“Former counter-terrorism police chief attacks Rishi Sunak’s Prevent plans“, reads a headline in today’s Guardian. “Former counter-terrorism chief Sir Peter Fahy […] said: “The widening of Prevent could damage its credibility and reputation. It makes it more about people’s thoughts and opinions. Fahy said: “The danger is the perception it creates that teachers and health workers are involved in state surveillance.”
This article leaves out that today’s reality is already far ahead of proposals or perception. School children and staff are already surveilled in these ways. Not only are things monitored that people think type or read or search for online and offline in the digital environment, but copies may be collected, retained by companies and interventions made.
The products don’t only permit monitoring of trends on aggregated data in overviews of student activity but the behaviours of individual students. And these can be deeply intrusive and sensitive when you are talking about self harm, abuse, and terrorism.
(For more on the safety tech sector, often using AI in proactive monitoring, see my previous post (May 2021) The Rise of Safety Tech.)
Intrusion through inference and interventions
From 1 July 2015 all schools have been subject to the Prevent duty under section 26 of the Counter-Terrorism and Security Act 2015, in the exercise of their functions, to have “due regard to the need to prevent people from being drawn into terrorism”. While these products are about monitoring far more than the remit of Prevent, many companies actively market online filtering, blocking and monitoring safety products as a way of meeting that in the digital environment. Such as, “Lightspeed Filter™helps you meet all of the Prevent Duty’s online regulations…”
Despite there being no obligation to date, to fulfil this duty through technology, some companies’ way of selling such tools could be interpreted as threatening if schools don’t use it. Like this example:
“Failure to comply with the requirements may result in intervention from the Prevent Oversight Board, prompt an Ofsted inspection or incur loss of funding.”
Such products may create and send real-time alerts to company or school staff when children attempt to reach sites or type “flagged words” related to radicalisation or extremism on any online platform.
Under the auspices of the safeguarding-in-schools data sharing and web monitoring in the Prevent programme children may be labelled with terrorism or extremism labels, data which may be passed on to others or stored outside the UK without their knowledge. The drift in what is considered significant, has been from terrorism into now more vague and broad terms of extremism and radicalisation. Away from some assessment of intent and capability of action, into interception and interventions for potentially insignificant potential vulnerabilities and inferred assumptions of disposition towards such ideas. This is not potentially going to police thoughts as suggested by Fahy of Sunak’s views. It is already doing so. Policing thoughts in the developing child and holding them accountable for it like this in ways that are unforeseeable, is inappropriate and requires thorough investigation into its effects on children, including mental health.
But it’s important to understand that these libraries of thousands of words, ever changing and in multiple languages, and what the systems are looking for and flag, often claiming to do it using Artificial Intelligence, go far beyond Prevent. ‘Legal but harmful’ is their bread and butter. Self harm, harm to or from others.
While companies have no obligations to publish how the monitoring or flagging operates, what the words or phrases or blocked websites are, their error rates (positive and negative) or the effects on children or school staff and their behaviour as a result, these companies have a great deal of influence what gets inferred from what children do online, and who decides what to act on.
Why does it matter?
Schools have normalized the premise that systems they introduce should monitor activity outside of the school network, and hours. And that strangers or their private companies’ automated systems should be involved in inferring or deciding what children are ‘up to’ before the school staff who know the children in front of them.
In a defenddigitalme report, The State of Data 2020, we included a case study on one company that has since been bought out. And bought again. As of August 2018 eSafe was monitoring approximately one million school children plus staff across the UK. This case study they used in their public marketing raised all sorts of questions on professional confidentiality and school boundaries, personal privacy, ethics, and companies’ role and technical capability, as well as the lack of any safety tech accountability.
“A female student had been writing an emotionally charged letter to her Mum using Microsoft Word, in which she revealed she’d been raped. Despite the device used being offline, eSafe picked this up and alerted John and his care team who were able to quickly intervene.”
“Bearing in mind we are doing this throughout the year, the behaviours we detect are not confined to the school bell starting in the morning and ringing in the afternoon, clearly; it is 24/7 and it is every day of the year. Lots of our incidents are escalated through activity on evenings, weekends and school holidays.”
Similar products offer a photo capturing feature of users (pupils while using the device being monitored) described as “common across most solutions in the sector” by this company:
When a critical safeguarding keyword is copied, typed or searched for across the school network, schools can turn on NetSupport DNA’s webcams capture feature (this feature is turned-off by default) to capture an image of the user (not a recording) who has triggered the keyword.
Who decides what does and does not trigger interventions across different products? In the month of December 2017 alone, eSafe claims they added 2254 words to their threat libraries.
Famously, Impero’s system even included the word “biscuit” which they say is a term used to define a gun. Their system was used by more than “half a million students and staff in the UK” in 2018. And students had better not talk about “taking a wonderful bath.” Currently there is no understanding or oversight of the accuracy of this kind of software and black-box decision-making is often trusted without openness to human question or correction.
“The result of vague and broad definitions of harmful information, for example in determining how to set Internet filters, can prevent children from gaining access to information that can support them to make informed choices, including honest, objective and age-appropriate information about issues such as sex education and drug use. This may exacerbate rather than diminish children’s vulnerability to risk.” (2014)
U.S. safety tech creates harms
Today in the U.S. the CDT published a report on school monitoring systems there, many of which are also used over here. The report revealed that 13 percent of students knew someone who had been outed as a result of student-monitoring software. Another conclusion the CDT draws, is that monitoring is used for discipline more often than for student safety.
We don’t have that same research for the UK, but we’ve seen IT staff openly admit to using the webcam feature to take photos of young boys who are “mucking about” on the school library computer.
The Online Safety Bill scales up problems like this
The Online Safety Bill seeks to expand how such ‘behavioural identification technology’ can be expanded outside schools.
“Proactive technology include content moderation technology, user profiling technology or behaviour identification technology which utilises artificial intelligence or machine learning.” (p151 Online Safety Bill, August 3, 2022)
The “proactive technology requirement” is as yet rather open ended, left to OFCOM in Codes of Practice but the scope creep of such AI-based tools has become ever more intrusive in education. Legal but harmful is decided by companies and the IWF and any number of opaque third parties whose process and decision-making we know little about. It’s important not to conflate filtering, blocking lists of ‘unsuitable’ websites that can be accessed in schools, with monitoring and tracking individual behaviours.
‘Technological developments that have the capacity to interfere with our freedom of thought fall clearly within the scope of “morally unacceptable harm,”‘ according to Algere (2017), and yet this individual interference is at the very core of school safeguarding tech and policy by design.
In 2018, the ‘lawful but harmful’ list of activities in the Online Harms White paper was nearlyidenticalwiththose termsused by school SafetyTech companies. The Bill now appears to be trying to create a new legitimate basis for these practices, more about underpinning a developing market, than supporting children’s safety or rights.
Chilling speech is itself controlling content
While a lot of debate about the Bill has been the free speech impacts of content removal, there has been less about what is unwritten but how it will operate to prevent speech and participation in the digital environment for children. Thechillingeffectofsurveillanceon access and participationonlineiswell documented. Younger people and women are more likely to be negatively affected (Penney, 2017). The chilling effect on thought and opinion is worsened in these types of tools that trigger an alert even when what is typed is quickly deleted or remains unsent or shared. Thoughts are no longer private.
The ability to use end-to-end encryption on private messaging platforms is simply worked around by these kinds of tools, trading security for claims of children’s safety. Anything on screen may be read in the clear by some systems, even capturing passwords and bank details.
Graham Smith has written, “It may seem like overwrought hyperbole to suggest that the [Online Harms] Bill lays waste to several hundred years of fundamental procedural protections for speech. But consider that the presumption against prior restraint appeared in Blackstone’s Commentaries (1769). It endures today in human rights law. That presumption is overturned by legal duties that require proactive monitoring and removal before an independent tribunal has made any determination of illegality.”
More than this, there is no determination of illegality in legal but harmful activity. It’s opinion. The government is prone to argue that, “nothing in the Bill says X…” but you need to understand this context of how such proactive behavioural monitoring tools work is through threat and the resultant chilling effect to impose unwritten control. This Bill does not create a safer digital environment, it creates threat models for users and companies, to control how we think and behave.
What do children and parents think?
Young people’s own views that don’t fit the online harms narrative have been ignored by Westminster scrutiny Committees. A 2019 survey by the Australian e-safety commissioner found that over half (57%) of child respondents were uncomfortable with background monitoring processes, and 43 %were unsure about these tools’ effectiveness in ensuring online safety.
And what of the role of parents? Article 3(2) of the UNCRC says: “States Parties undertake to ensure the child such protection and care as is necessary for his or her wellbeing, taking into account the rights and duties of his or her parents, legal guardians, or other individuals legally responsible for him or her, and, to this end, shall take all appropriate legislative and administrative measures.” (my emphasis)
In 2018, 84% of 1,004 parents in England who we polled through Survation, agreed that children and guardians should be informed how this monitoring activity works and wanted to know what the keywords were. (We didn’t ask if it should happen at all or not.)
The wide ranging nature [of general monitoring] rather than targeted and proportionate interference has been judged to be in breach of law and a serious interference with rights, previously. Neitherpolicymakersnor companies should assume parents want safety tech companies to remove autonomy, or make inferences about our children’s lives. Parents if asked, reject the secrecy in which it happens today and demand transparency and accountability. Teachers can feel anxious talking about it at all. There’s no clear routes for error corrections, in fact it’s not done because some claim in building up profiles staff should not delete anything and ignore claims of errors, in case a pattern of behaviour is missed. But there’s no independent assessments available to evidence these tools work or are worth the costs. There are no routes for redress or responsibility taken for tech-made mistakes. None of which makes children safer online.
Before broadening out where such monitoring tools are used, their use and effects on school children need to be understood and openly debated. Policy makers may justify turning a blind eye to harms created by one set of technology providers while claiming that only the other tech providers are the problem, because it suits political agendas or industry aims, but children’s rights and their wellbeing should not be sacrificed in doing so. Opaque, unlawful and unsafe practice must stop. A quid pro quo for getting access to millions of children’s intimate behaviour, should be transparent access to their product workings, and accepting standards on universal safe accountable practices. Families need to know what’s recorded. To have routes for redress when a daughter researching ‘cliff walks’ gets flagged as a suicide risk or an environmentally interested teenage son searching for information on ‘black rhinos’ is asked about his potential gang membership. The tools sold as solutions to online harms, shouldn’t create more harm like these reported real-life case studies.
Teachers are ‘involved in state surveillance’ as Fahy put it, through Prevent. Sunak was wrong to point away from the threats of the far right in his comments. But the far broader unspoken surveillance of children’s personal lives, behaviours and thoughts through general monitoring in schools, and what will be imposed through the Online Safety Bill more broadly, should concern us far more than was said.
At the CRISP hosted, Rise of Safety Tech, event this week, the moderator asked an important question: What is Safety Tech? Very honestly Graham Francis of the DCMS answered among other things, “It’s an answer we are still finding a question to.”
From ISP level to individual users, limitations to mobile phone battery power and app size compatibility, a variety of aspects within a range of technology were discussed. There is a wide range of technology across this conflated set of products packaged under the same umbrella term. Each can be very different from the other, even within one set of similar applications, such as school Safety Tech.
It worries me greatly that in parallel to the run up to the Online Harms legislation that their promotion appears to have assumed the character of a done deal. Some of these tools are toxic to children’s rights because of the policy that underpins them. Legislation should not be gearing up to make the unlawful lawful, but fix what is broken.
The current drive is towards the normalisation of the adoption of such products in the UK, and to make them routine. It contrasts with the direction of travel of critical discussion outside the UK.
Some Safety Tech companies have human staff reading flagged content and making decisions on it, while others claim to use only AI. Both might be subject to any future EU AI Regulation for example.
The big questions include who watches the watchers, with what scrutiny and safeguards? Is it safe, lawful, ethical, and does it work?
Safety Tech isn’t only an answer we are still finding a question to. It is a world view, with a particular value set. Perhaps the only lens through which its advocates believe the world wide web should be seen, not only by children, but by anyone. And one that the DCMS is determined to promote with “the UK as a world-leader” in a worldwide export market.
But does the Department understand what they are promoting? The DCMS Minister responsible, Oliver Dowden said in Parliament on December 15th 2020: “Clearly, if it was up to individuals within those companies to identify content on private channels, that would not be acceptable—that would be a clear breach of privacy.”
He’s right. It is. And yet he and his Department are promoting it.
So how is this going to play out if at all, in the Online Harms legislation expected soon, that he owns together with the Home Office? Sadly the needed level of understanding by the Minister or in the third sector and much of the policy debate in the media, is not only missing, but is actively suppressed by the moral panic whipped up in emotive personal stories around a Duty of Care and social media platforms. Discussion is siloed about identifying CSAM, or grooming, or bullying or self harm, and actively ignores the joined-up, wider context within which Safety Tech operates.
That context is the world of the Home Office. Of anti-terrorism efforts. Of mass surveillance and efforts to undermine encryption that are as nearly old as the Internet. The efforts to combat CSAM or child grooming online, operate in the same space. WePROTECT for example, sits squarely amid it all, established in 2014 by the UK Government and the then UK Prime Minister, David Cameron. Scrutiny of UK breaches of human rights law are well documented in ECHR rulings. Other state members of the alliance including the UAE stand accused of buying spyware to breach activists’ encrypted communications. It is disingenuous for any school Safety Tech actors to talk only of child protection without mention of this context. School Safety Tech while all different, operate by tagging digital activity with categories of risk, and these tags can include terrorism and extremism.
Once upon a time, school filtering and blocking services meant only denying access to online content that had no place in the classroom. Now it can mean monitoring all the digital activity of individuals, online and offline, using school or personal devices, working around encryption, whenever connected to the school network. And it’s not all about in-school activity. No matter where a child’s account is connected to the school network, or who is actually using it, their activity might be monitored 24/7, 365 days a year. A user’s activity that matches with the thousands of words or phrases on watchlists and in keyword libraries gets logged, and profiles individuals with ‘vulnerable’ behaviour tags, sometimes creating alerts. Their scope has crept from flagging up content, to flagging up children. Some schools create permanent records including false positives because they retain everything in a risk-averse environment, even things typed that a child subsequently deleted, and may be distributed and accessible by an indefinite number of school IT staff and stored in further third parties’ systems like CPOMS or Capita SIMS.
A wide range of the rights of the child are breached by mass monitoring in the UK, such as outlined in the UN Committee on the Rights of the Child General Comment No.25 which states that, “Any digital surveillance of children, together with any associated automated processing of personal data, should respect the child’s right to privacy and should not be conducted routinely, indiscriminately or without the child’s knowledge or, in the case of very young children, that of their parent or caregiver; nor should it take place without the right to object to such surveillance, in commercial settings and educational and care settings, and consideration should always be given to the least privacy-intrusive means available to fulfil the desired purpose.” (para 75)
Even the NSPCC, despite their recent public policy that opposes secure messaging using end-to-send encryption, recognises on its own Childline webpage the risk for children from content monitoring of children’s digital spaces, and that such monitoring may make them less safe.
In my work in 2018, one school Safety Tech company accepted our objections from defenddigitalme, that this monitoring went too far in its breach of children’s confidentially and safe spaces, and it agreed to stop monitoring counselling services. But there are roughly fifteen active companies here in the UK and the data protection regulator, the ICO despite being publicly so keen to be seen to protect children’s rights, has declined to act to protect children from the breach of their privacy and data protection rights across this field.
There are questions that should be straightforward to ask and answer, and while some CEOs are more willing to engage constructively with criticism and ideas for change than others, there is reluctance to address the key question: what is the lawful basis for monitoring children in school, at home, in- or out-side school hours?
Another important question often without an answer, is how do these companies train their algorithms whether in age verification or child safety tech? How accurate are the language inferences for an AI designed to catch children out who are being deceitful and where are assumptions, machine or man-made, wrong or discriminatory? It is overdue that our Regulator, the ICO, should do what the FTC did with Paravision, and require companies that develop tools through unlawful data processing to delete the output from it, the trained algorithm, plus products created from it.
Many of the harms from profiling children were recognised by the ICO in the Met Police gangs matrix: discrimination, conflation of victim and perpetrator, notions of ‘pre-crime’ without independent oversight, data distributed out of context, and excessive retention.
Harm is after all why profiling of children should be prohibited. And where, in exceptional circumstances, States may lift this restriction, it is conditional that appropriate safeguards are provided for by law.
While I believe any of the Safety Tech generated category profiles could be harmful to a child through mis-interventions, being treated differently by staff as a result, or harm a trusted relationship, perhaps the potentially most devastating to a child’s prospects are from mistakes that could be made under the Prevent duty.
The UK Home Office has pushed its Prevent agenda through schools since 2015, and it has been built into school Safety Tech by-design. School Safety Tech while all different, operate by tagging digital activity with categories of risk, and these tags can include terrorism and extremism. I know of schools that have flags attached to children’s records that are terrorism related, but who have had no Prevent referral. But there is no transparency of these numbers at all. There is no oversight to ensure children do not stay wrongly tagged with those labels. Families may never know.
Perhaps the DCMS needs to ask itself, are the values of the UK Home Office really what the UK should export to children globally from “the UK as a world-leader” without independent legal analysis, without safeguards, and without taking accountability for their effects?
The Home Office talk is of safeguarding children, but it is opposed to them having safe spaces online. School Safety Tech tools actively work around children’s digital security, can act as a man-in-the-middle, and can create new risks. There is no evidence I have seen that on balance convinces me that school Safety Tech does in fact make children safer. But plenty of evidence that the Home Office appears to want to create the conditions that make children less secure so that such tools could thrive, by weakening the security of digital activity through its assault on end-to-end encryption. My question is whether Online Harms is to be the excuse to give it a lawful basis.
Today there are zero statutory transparency obligations, testing or safety standards required of school Safety Tech before it can be procured in UK state education at scale.
So what would a safe and lawful framework for operation look like? It would be open to scrutiny and require regulatory action, and law.
There are no published numbers of how many records are created about how many school children each year. There are no safeguards in place to protect children’s rights or protection from harm in terms of false positives, error retention, transfer of records to the U.S. or third party companies, or how many covert photos they have enabled to be taken of children via webcam by school staff. There is no equivalent of medical device ‘foreseeable misuse risk assessment’ such as ISO 14971 would require, despite systems being used for mental health monitoring with suicide risk flags. Children need to know what is on their record and to be able to seek redress when it is wrong. The law would set boundaries and safeguards and both existing and future law would need to be enforced. And we need independent research on the effects of school surveillance, and its chilling effects on the mental health and behaviour of developing young people.
But no one is yet willing to say in the school Safety Tech sector, these are thousands of words that if your child types may trigger a flag, or indeed, here’s an annual report of all the triggered flags and your own or your child’s saved profile. A school’s interactions with children’s social care already offers a framework for dealing with information that could put a child at risk from family members, so reporting should be do-able.
At the end of the event this week, the CRISP event moderator said of their own work, outside schools, that, “we are infiltrating bad actor networks across the globe and we are looking at everything they are saying. […] We have a viewpoint that there are certain lines where privacy doesn’t exist anymore.”
Their company website says their work involves, “uncovering and predicting the actions of bad actor, activist, agenda-driven and interest groups“. That’s a pretty broad conflation right there. Their case studies include countering social media activism against a luxury apparel brand. And their legal basis of ‘legitimate interests‘ for their data processing might seem flimsy at best, for such a wide ranging surveillance activity where, ‘privacy doesn’t exist anymore’.
I must often remind myself that the people behind Safety Tech may epitomise the very best of what some believe is making the world safer online as they see it. But it is *as they see it*. And if policy makers or CEOs have convinced themselves that because ‘we are doing it for good, a social impact, or to safeguard children’, that breaking the law is OK, then it should be a red flag that these self-appointed ‘good guys’ appear to think themselves above the law.
My takeaway time and time again, is that companies alongside governments, policy makers, and a range of lobbying interests globally, want to redraw the lines around human rights, so that they can overstep them. There are “certain lines” that don’t suit their own business models or agenda. The DCMS may talk about seeing its first safety tech unicorn, but not about the private equity funding, or where they pay their taxes. Children may be the only thing they talk about protecting but they never talk of protecting children’s rights.
In the school Safety Tech sector, there is activity that I believe is unsafe, or unethical, or unlawful. There is no appetite or motivation so far to fix it. If in upcoming Online Harms legislation the government seeks to make lawful what is unlawful today, I wonder who will be held accountable for the unsafe and the unethical, that come with the package deal—and will the Minister run that reputational risk?
Here’s some thoughts about the Prevent programme, after the half day I spent at the event this week, Youth Empowerment and Addressing Violent Youth Radicalisation in Europe.
Firstly, I appreciated the dynamic and interesting youth panel. Young people, themselves involved in youth work, or early researchers on a range of topics. Panelists shared their thoughts on:
Removal of gang databases and systemic racial targeting
Questions over online content takedown with the general assumption that “someone’s got to do it.”
The purposes of Religious Education and lack of religious understanding as cause of prejudice, discrimination, and fear.
From these connections comes trust.
Next, Simon Chambers, from the British Council, UK National Youth Agency, and Erasmus UK, talked about the programme of Erasmus Plus, under the striking sub theme, from these connections comes trust.
42% of the world’s population are under 25
Young people understand that there are wider, underlying complex factors in this area and are disproportionately affected by conflict, economic change and environmental disaster.
Many young people struggle to access education and decent work.
Young people everywhere can feel unheard and excluded from decision-making — their experience leads to disaffection and grievance, and sometimes to conflict.
We then heard a senior Home Office presenter speak about Radicalisation: the threat, drivers and Prevent programme.
On Contest 2018 Prevent / Pursue / Protect and Prepare
What was perhaps most surprising was his statement that the programme believes there is no checklist, [but in reality there are checklists] no single profile, or conveyer belt towards radicalisation.
“This shouldn’t be seen as some sort of predictive model,” he said. “It is not accurate to say that somehow we can predict who is going to become a terrorist, because they’ve got poor education levels, or because necessarily have a deprived background.”
But he then went on to again highlight the list of identified vulnerabilities in Thomas Mair‘s life, which suggests that these characteristics are indeed seen as indicators.
When I look at the ‘safeguarding-in-school’ software that is using vulnerabilities as signals for exactly that kind of prediction of intent, the gap between theory and practice here, is deeply problematic.
One slide included Internet content take downs, and suggested 300K pieces of illegal terrorist material have been removed since February 2010. That number he later suggested are contact with CTIRU, rather than content removal defined as a particular form. (For example it isn’t clear if this is a picture, a page, or whole site). This is still somewhat unclear and there remain important open questions, given its focus in the online harms policy and discussion.
The big gap that was not discussed and that I believe matters, is how much autonomy teachers have, for example, to make a referral. He suggested “some teachers may feel confident” to do what is needed on their own but others, “may need help” and therefore make a referral. Statistics on those decision processes are missing, and it is very likely I believe that over referral is in part as a result of fearing that non-referral, once a computer has tagged issues as Prevent related, would be seen as negligent, or not meeting the statutory Prevent duty as it applies to schools.
On the Prevent Review, he suggested that the current timeline still stands, of August 2020, even though there is currently no Reviewer. It is for Ministers to make a decision, who will replace Lord Carlile.
Safeguarding children and young people from radicalisation
Mark Chalmers of Westminster City Council., then spoke about ‘safeguarding children and young people from radicalisation.’
He started off with a profile of the local authority demographic, poverty and wealth, migrant turnover, proportion of non-English speaking households. This of itself may seem indicative of deliberate or unconscious bias.
He suggested that Prevent is not a security response, and expects that the policing role in Prevent will be reduced over time, as more is taken over by Local Authority staff and the public services. [Note: this seems inevitable after the changes in the 2019 Counter Terrorism Act, to enable local authorities, as well as the police, to refer persons at risk of being drawn into terrorism to local channel panels. Should this have happened at all, was not consulted on as far as I know]. This claim that Prevent is not a security response, appears different in practice, when Local Authorities refuse FOI questions on the basis of security exemptions in the FOI Act, Section 24(1).
Both speakers declined to accept my suggestion that Prevent and Channel is not consensual. Participation in the programme, they were adamant is voluntary and confidential. The reality is that children do not feel they can make a freely given informed choice, in the face of an authority and the severity of the referral. They also do not understand where their records go to, how confidential are they really, and how long they are kept or why.
The recently concluded legal case and lengths one individual had to go to, to remove their personal record from the Prevent national database, shows just how problematic the mistaken perception of a consensual programme by authorities is.
I knew nothing of the Prevent programme at all in 2015. I only began to hear about it once I started mapping the data flows into, across and out of the state education sector, and teachers started coming to me with stories from their schools.
I found it fascinating to hear those speak at the conference that are so embedded in the programme. They seem unable to see it objectively or able to accept others’ critical point of view as truth. It stems perhaps from the luxury of having the privilege of believing you yourself, will be unaffected by its consequences.
“Yes,” said O’Brien, “we can turn it off. We have that privilege” (1984)
There was no ground given at all for accepting that there are deep flaws in practice. That in fact ‘Prevent is having the opposite of its intended effect: by dividing, stigmatising and alienating segments of the population, Prevent could end up promoting extremism, rather than countering it’ as concluded in the 2016 report Preventing Education: Human Rights and Countering terrorism in UK Schools by Rights Watch UK .
Mark Chalmers conclusion was to suggest perhaps Prevent is not always going to be the current form, of bolt on ‘big programme’ and instead would be just like any other form of child protection, like FGM. That would mean every public sector worker, becomes an extended arm of the Home Office policy, expected to act in counter terrorism efforts.
But the training, the nuance, the level of application of autonomy that the speakers believe exists in staff and in children is imagined. The trust between authorities and people who need shelter, safety, medical care or schooling must be upheld for the public good.
No one asked, if and how children should be seen through the lens of terrorism, extremism and radicalisation at all. No one asked if and how every child, should be able to be surveilled online by school imposed software and covert photos taken through the webcam in the name of children’s safeguarding. Or labelled in school, associated with ‘terrorist.’ What happens when that prevents trust, and who measures its harm?
[click to view larger file]
Far too little is known about who and how makes decisions about the lives of others, the criteria for defining inappropriate activity or referrals, or the opacity of decisions on online content.
What effects will the Prevent programme have on our current and future society, where everyone is expected to surveil and inform upon each other? Failure to do so, to uphold the Prevent duty, becomes civic failure. How is curiosity and intent separated? How do we safeguard children from risk (that is not harm) and protect their childhood experiences, their free and full development of self?
No one wants children to be caught up in activities or radicalisation into terror groups. But is this the correct way to solve it?
“The research provides new evidence that by attempting to profile and predict violent youth radicalisation, we may in fact be breeding the very reasons that lead those at risk to violent acts.” (Professor Theo Gavrielides).
Current case studies of lived experience, and history also say it is mistaken. Prevent when it comes to children, and schools, needs massive reform, at very least, but those most in favour of how it works today, aren’t the ones who can be involved in its reshaping.
“Who denounced you?” said Winston.
“It was my little daughter,” said Parsons with a sort of doleful pride. “She listened at the keyhole. Heard what I was saying, and nipped off to the patrols the very next day. Pretty smart for a nipper of seven, eh? I don’t bear her any grudge for it. In fact I’m proud of her. It shows I brought her up in the right spirit, anyway.” (1984).
The event was the launch of the European study on violent youth radicalisation from YEIP: The project investigated the attitudes and knowledge of young Europeans, youth workers and other practitioners, while testing tools for addressing the phenomenon through positive psychology and the application of the Good Lives Model.
Its findings include that young people at risk of violent radicalisation are “managed” by the existing justice system as “risks”. This creates further alienation and division, while recidivism rates continue to spiral.
Notes [and my thoughts] from the Women Leading in AI launch event of the Ten Principles of Responsible AI report and recommendations, February 6, 2019.
Speakers included Ivana Bartoletti (GemServ), Jo Stevens MP, Professor Joanna J Bryson, Lord Tim Clement-Jones, Roger Taylor (Centre for Data Ethics and Innovation, Chair), Sue Daley (techUK), Reema Patel, Nuffield Foundation and Ada Lovelace Institute.
Challenging the unaccountable and the ‘inevitable’ is the title of the conclusion of the Women Leading in AI reportTen Principles of Responsible AI, launched this week, and this makes me hopeful.
“There is nothing inevitable about how we choose to use this disruptive technology. […] And there is no excuse for failing to set clear rules so that it remains accountable, fosters our civic values and allows humanity to be stronger and better.”
Everyone’s talking about ethics, she said, but it has limitations. I agree with that. This was by contrast very much a call to action.
It was nearly impossible not to cheer, as she set out without any of the usual bullshit, the reasons why we need to stop “churning out algorithms which discriminate against women and minorities.”
Professor Joanna J Bryson took up multiple issues, such as why
innovation, ‘flashes in the pan’ are not sustainable and not what we’re looking for things in that work for us [society].
The power dynamics of data, noting Facebook, Google et al are global assets, and are also global problems, and flagged the UK consultation on taxation open now.
And that it is critical that we do not have another nation with access to all of our data.
She challenged the audience to think about the fact that inequality is higher now than it has been since World War I. That the rich are getting richer and that imbalance of not only wealth, but of the control individuals have in their own lives, is failing us all.
This big picture thinking while zooming in on detailed social, cultural, political and tech issues, fascinated me most that evening. It frustrated the man next to me apparently, who said to me at the end, ‘but they haven’t addressed anything on the technology.’
[I wondered if that summed up neatly, some of why fixing AI cannot be a male dominated debate. Because many of these issues for AI, are not of the technology, but of people and power.]
Jo Stevens, MP for Cardiff Central, hosted the event and was candid about politicians’ level of knowledge and the need to catch up on some of what matters in the tech sector.
We grapple with the speed of tech, she said. We’re slow at doing things and tech moves quickly. It means that we have to learn quickly.
While discussing how regulation is not something AI tech companies should fear, she suggested that a constructive framework whilst protecting society against some of the problems we see is necessary and just, because self-regulation has failed.
She talked about their enquiry which began about “fake news” and disinformation, but has grown to include:
wider behavioural economics,
how it affects democracy.
understanding the power of data.
disappointment with social media companies, who understand the power they have, and fail to be accountable.
She wants to see something that changes the way big business works, in the way that employment regulation challenged exploitation of the workforce and unsafe practices in the past.
The bias (conscious or unconscious) and power imbalance has some similarity with the effects on marginalised communities — women, BAME, disabilities — and she was looking forward to see the proposed solutions, and welcomed the principles.
Right now there are so many different bodies, groups in parliament and others looking at this [AI / Internet / The Digital World] he said, so it was good that the topic is timely, front and centre with a focus on women, diversity and bias.
He highlighted, the importance of maintaining public trust. How do you understand bias? How do you know how algorithms are trained and understand the issues? He fessed up to being a big fan of DotEveryone and their drive for better ‘digital understanding’.
[Though sometimes this point is over complicated by suggesting individuals must understand how the AI works, the consensus of the evening was common sensed — and aligned with the Working Party 29 guidance — that data controllers must ensure they explain clearly and simply to individuals, how the profiling or automated decision-making process works, and what its effect is for them.]
The way forward he said includes:
Designing ethics into algorithms up front.
Data audits need to be diverse in order to embody fairness and diversity in the AI.
Questions of the job market and re-skilling.
The enforcement of ethical frameworks.
He also asked how far bodies will act, in different debates. Deciding who decides on that is still a debate to be had.
For example, aware of the social credit agenda and scoring in China, we should avoid the same issues. He also agreed with Joanna, that international cooperation is vital, and said it is important that we are not disadvantaged in this global technology. He expected that we [the Government Office for AI] will soon promote a common set of AI ethics, at the G20.
Facial recognition and AI are examples of areas that require regulation for safe use of the tech and to weed out those using it for the wrong purposes, he suggested.
However, on regulation he held back. We need to be careful about too many regulators he said. We’ve got the ICO, FCA, CMA, OFCOM, you name it, we’ve already got it, and they risk tripping over one another. [What I thought as CDEI was created para 31.]
We [the Lords Committee] didn’t suggest yet another regulator for AI, he said and instead the CDEI should grapple with those issues and encourage ethical design in micro-targeting for example.
Roger Taylor (Chair of the CDEI), — after saying it felt as if the WLinAI report was like someone had left their homework on his desk, — supported the concept of the WLinAI principles are important, and agreed it was time for practical things, and what needs done.
Can our existing regulators do their job, and cover AI? he asked, suggesting new regulators will not be necessary. Bias he rightly recognised, already exists in our laws and bodies with public obligations, and in how AI is already operating;
What evidence is needed, what process is required, what is needed to assure that we know how it is actually operating? Who gets to decide to know if this is fair or not? While these are complex decisions, they are ultimately not for technicians, but a decision for society, he said.
[So far so good.]
Then he made some statements which were rather more ambiguous. The standards expected of the police will not be the same as those for marketeers micro targeting adverts at you, for example.
[I wondered how and why.]
Start up industries pay more to Google and Facebook than in taxes he said.
[I wondered how and why.]
When we think about a knowledge economy, the output of our most valuable companies is increasingly ‘what is our collective truth? Do you have this diagnosis or not? Are you a good credit risk or not? Even who you think you are — your identity will be controlled by machines.’
What can we do as one country [to influence these questions on AI], in what is a global industry? He believes, a huge amount. We are active in the financial sector, the health service, education, and social care — and while we are at the mercy of large corporations, even large corporations obey the law, he said.
The power to use systems to nudge our decisions, he suggested, is one that needs careful thought. The desire to use the tech to help make decisions is inbuilt into what is actually wrong with the technology that enables us to do so. [With this I strongly agree, and there is too little protection from nudge in data protection law.]
The real question here is, “What is OK to be owned in that kind of economy?” he asked.
This was arguably the neatest and most important question of the evening, and I vigorously agreed with him asking it, but then I worry about his conclusion in passing, that he was, “very keen to hear from anyone attempting to use AI effectively, and encountering difficulties because of regulatory structures.“
[And unpopular or contradictory a view as it may be, I find it deeply ethically problematic for the Chair of the CDEI to be held by someone who had a joint-venture that commercially exploited confidential data from the NHS without public knowledge, and its sale to the Department of Health was described by the Public Accounts Committee, as a “hole and corner deal”. That was the route towards care.data, that his co-founder later led for NHS England. The company was then bought by Telstra, where Mr Kelsey went next on leaving NHS Engalnd. The whole commodification of confidentiality of public data, without regard for public trust, is still a barrier to sustainable UK data policy.]
Sue Daley (Tech UK) agreed this year needs to be the year we see action, and the report is a call to action on issues that warrant further discussion.
Business wants to do the right thing, and we need to promote it.
We need two things — confidence and vigilance.
We’re not starting from scratch, and talked about GDPR as the floor not the ceiling. A starting point.
[I’m not quite sure what she was after here, but perhaps it was the suggestion that data regulation is fundamental in AI regulation, with which I would agree.]
What is the gap that needs filled she asked? Gap analysis is what we need next and avoid duplication of effort —need to avoid complexity and duplicity of work with other bodies. If we can answer some of the big, profound questions need to be addressed to position the UK as the place where companies want to come to.
Sue was the only speaker that went on to talk about the education system that needs to frame what skills are needed for a future world for a generation, ‘to thrive in the world we are building for them.’
She finished with the hope that young people watching BBC icons the night before would see, Alan Turing [winner of the title] and say yes, I want to be part of that.
Listening to Reema Patel, representative of the Ada Lovelace Institute, was the reason I didn’t leave early and missed my evening class. Everything she said resonated, and was some of the best I have heard in the recent UK debate on AI.
Civic engagement, the role of the public is as yet unclear with not one homogeneous, but many publics.
The sense of disempowerment is important, with disconnect between policy and decisions made about people’s lives.
Transparency and literacy are key.
Accountability is vague but vital.
What does the social contract look like on people using data?
Data may not only be about an individual and under their own responsibility, but about others and what does that mean for data rights, data stewardship and articulation of how they connect with one another, which is lacking in the debate.
Legitimacy; If people don’t believe it is working for them, it won’t work at all.
Ensuring tech design is responsive to societal values.
2018 was a terrible year she thought. Let’s make 2019 better. [Yes!]
Comments from the floor and questions included Professor Noel Sharkey, who spoke about the reasons why it is urgent to act especially where technology is unfair and unsafe and already in use. He pointed to Compass (Durham police), and predictive policing using AI and facial recognition, with 5% accuracy, and that the Met was not taking these flaws seriously. Liberty produced a strong report on it out this week.
Caroline, from Women in AI echoed my own comments on the need to get urgent review in place of these technologies used with children in education and social care. [in particular where used for prediction of child abuse and interventions in family life].
Joanna J Bryson added to the conversation on accountability, to say people are not following existing software and audit protocols, someone just needs to go and see if people did the right thing.
The basic question of accountability, is to ask if any flaw is the fault of a corporation, of due diligence, or of the users of the tool? Telling people that this is the same problem as any other software, makes it much easier to find solutions to accountability.
Tim Clement-Jones asked, on how many fronts can we fight on at the same time? If government has appeared to exempt itself from some of these issues, and created a weak framework for itself on handing data, in the Data Protection Act — critically he also asked, is the ICO adequately enforcing on government and public accountability, at local and national levels?
Sue Daley also reminded us that politicians need not know everything, but need to know what the right questions are to be asking? What are the effects that this has on my constituents, in employment, my family? And while she also suggested that not using the technology could be unethical, a participant countered that it’s not the worst the thing to have to slow technology down and ensure it is safe before we all go along with it.
My takeaways of the evening, included that there is a very large body of women, of whom attendees were only a small part, who are thinking, building and engineering solutions to some of these societal issues embedded in policy, practice and technology. They need heard.
It was genuinely electric and empowering, to be in a room dominated by women, women reflecting diversity of a variety of publics, ages, and backgrounds, and who listened to one another. It was certainly something out of the ordinary.
There was a subtle but tangible tension on whether or not regulation beyond what we have today is needed.
While regulating the human behaviour that becomes encoded in AI, we need to ensure ethics of human behaviour, reasonable expectations and fairness are not conflated with the technology [ie a question of, is AI good or bad] but how it is designed, trained, employed, audited, and assess whether it should be used at all.
Why there’s not more women or people from minorities working in the sector, was a really interesting if short, part of the discussion. Why should young women and minorities want to go into an environment that they can see is hostile, in which they may not be heard, and we still hold *them* responsible for making work work?
And while there were many voices lamenting the skills and education gaps, there were probably fewer who might see the solution more simply, as I do. Schools are foreshortening Key Stage 3 by a year, replacing a breadth of subjects, with an earlier compulsory 3 year GCSE curriculum which includes RE, and PSHE, but means that at 12, many children are having to choose to do GCSE courses in computer science / coding, or a consumer-style iMedia, or no IT at all, for the rest of their school life. This either-or content, is incredibly short-sighted and surely some blend of non-examined digital skills should be offered through to 16 to all, at least in parallel importance with RE or PSHE.
I also still wonder, about all that incredibly bright and engaged people are not talking about and solving, and missing in policy making, while caught up in AI. We need to keep thinking broadly, and keep human rights at the centre of our thinking on machines. Anaïs Nin wrote over 70 years ago about the risks of growth in technology to expand our potential for connectivity through machines, but diminish our genuine connectedness as people.
“I don’t think the [American] obsession with politics and economics has improved anything. I am tired of this constant drafting of everyone, to think only of present day events”.
And as I wrote about nearly 3 years ago, we still seem to have no vision for sustainable public policy on data, or establishing a social contract for its use as Reema said, which underpins the UK AI debate. Meanwhile, the current changing national public policies in England on identity and technology, are becoming catastrophic.
Challenging the unaccountable and the ‘inevitable’ in today’s technology and AI debate, is an urgent call to action.
I look forward to hearing how Women Leading in AI plan to make it happen.
“A new, a vast, and a powerful language is developed for the future use of analysis, in which to wield its truths so that these may become of more speedy and accurate practical application for the purposes of mankind than the means hitherto in our possession have rendered possible.” [on Ada Lovelace, The First tech Visionary, New Yorker, 2013]
What would Ada Lovelace have argued for in today’s AI debates? I think she may have used her voice not only to call for the good use of data analysis, but for her second strength.The power of her imagination.
James Ball recently wrote in The European [1]:
“It is becoming increasingly clear that the modern political war isn’t one against poverty, or against crime, or drugs, or even the tech giants – our modern political era is dominated by a war against reality.”
My overriding take away from three days spent at the Conservative Party Conference this week, was similar. It reaffirmed the title of a school debate I lost at age 15, ‘We only believe what we want to believe.’
James writes that it is, “easy to deny something that’s a few years in the future“, and that Conservatives, “especially pro-Brexit Conservatives – are sticking to that tried-and-tested formula: denying the facts, telling a story of the world as you’d like it to be, and waiting for the votes and applause to roll in.”
These positions are not confined to one party’s politics, or speeches of future hopes, but define perception of current reality.
I spent a lot of time listening to MPs. To Ministers, to Councillors, and to party members. At fringe events, in coffee queues, on the exhibition floor. I had conversations pressed against corridor walls as small press-illuminated swarms of people passed by with Queen Johnson or Rees-Mogg at their centre.
In one panel I heard a primary school teacher deny that child poverty really exists, or affects learning in the classroom.
In another, in passing, a digital Minister suggested that Pupil Referral Units (PRU) are where most of society’s ills start, but as a Birmingham head wrote this week, “They’ll blame the housing crisis on PRUs soon!” and “for the record, there aren’t gang recruiters outside our gates.”
This is no tirade on failings of public policymakers however. While it is easy to suspect malicious intent when you are at, or feel, the sharp end of policies which do harm, success is subjective.
It is clear that an overwhelming sense of self-belief exists in those responsible, in the intent of any given policy to do good.
Where policies include technology, this is underpinned by a self re-affirming belief in its power. Power waiting to be harnessed by government and the public sector. Even more appealing where it is sold as a cost-saving tool in cash strapped councils. Many that have cut away human staff are now trying to use machine power to make decisions. Some of the unintended consequences of taking humans out of the process, are catastrophic for human rights.
The disconnect between perception of risk, the reality of risk, and real harm, whether perceived or felt from these applied policies in real-life, is not so much, ‘easy to deny something that’s a few years in the future‘ as Ball writes, but a denial of the reality now.
Concerningly, there is lack of imagination of what real harms look like.There is no discussion where sometimes these predictive policies have no positive, or even a negative effect, and make things worse.
I’m deeply concerned that there is an unwillingness to recognise any failures in current data processing in the public sector, particularly at scale, and where it regards the well-known poor quality of administrative data. Or to be accountable for its failures.
Harms, existing harms to individuals, are perceived as outliers. Any broad sweep of harms across policy like Universal Credit, seem perceived as political criticism, which makes the measurable failures less meaningful, less real, and less necessary to change.
There is a worrying growing trend of finger-pointing exclusively at others’ tech failures instead. In particular, social media companies.
Imagination and mistaken ideas are reinforced where the idea is plausible, and shared. An oft heard and self-affirming belief was repeated in many fora between policymakers, media, NGOs regards children’s online safety. “There is no regulation online”. In fact, much that applies offline applies online. The Crown Prosecution Service Social Media Guidelines is a good place to start. [2] But no one discusses where children’s lives may be put at risk or less safe, through the use of state information about them.
Policymakers want data to give us certainty. But many uses of big data, and new tools appear to do little more than quantify moral fears, and yet still guide real-life interventions in real-lives.
Child abuse prediction, and school exclusion interventions should not be test-beds for technology the public cannot scrutinise or understand.
In one trial attempting to predict exclusion, this recent UK research project in 2013-16 linked children’s school records of 800 children in 40 London schools, with Metropolitan Police arrest records of all the participants. It found interventions created no benefit, and may have caused harm. [3]
“Anecdotal evidence from the EiE-L core workers indicated that in some instances schools informed students that they were enrolled on the intervention because they were the “worst kids”.”
“Keeping students in education, by providing them with an inclusive school environment, which would facilitate school bonds in the context of supportive student–teacher relationships, should be seen as a key goal for educators and policy makers in this area,” researchers suggested.
But policy makers seem intent to use systems that tick boxes, and create triggers to single people out, with quantifiable impact.
Some of these systems are known to be poor, or harmful.
When it comes to predicting and preventing child abuse, there is concern with the harms in US programmes ahead of us, such as both Pittsburgh, and Chicago that has scrapped its programme.
The Illinois Department of Children and Family Services ended a high-profile program that used computer data mining to identify children at risk for serious injury or death after the agency’s top official called the technology unreliable, and children still died.
“We are not doing the predictive analytics because it didn’t seem to be predicting much,” DCFS Director Beverly “B.J.” Walker told the Tribune.
Many professionals in the UK share these concerns. How long will they be ignored and children be guinea pigs without transparent error rates, or recognition of the potential harmful effects?
Why on earth not? At least for these high risk projects.
How long should children be the test subjects of machine learning tools at scale, without transparent error rates, audit, or scrutiny of their systems and understanding of unintended consequences?
Is harm to any child a price you’re willing to pay to keep using these systems to perhaps identify others, while we don’t know?
Is there an acceptable positive versus negative outcome rate?
The evidence so far of AI in child abuse prediction is not clearly showing that more children are helped than harmed.
Surely it’s time to stop thinking, and demand action on this.
It doesn’t take much imagination, to see the harms. Safe technology, and safe use of data, does not prevent the imagination or innovation, employed for good.
Where you are willing to sacrifice certainty of human safety for the machine decision, I want someone to be accountable for why.
References
[1] James Ball, The European, Those waging war against reality are doomed to failure, October 4, 2018.
[2] Thanks to Graham Smith for the link. “Social Media – Guidelines on prosecuting cases involving communications sent via social media. The Crown Prosecution Service (CPS) , August 2018.”
[3] Obsuth, I., Sutherland, A., Cope, A. et al. J Youth Adolescence (2017) 46: 538. https://doi.org/10.1007/s10964-016-0468-4 London Education and Inclusion Project (LEIP): Results from a Cluster-Randomized Controlled Trial of an Intervention to Reduce School Exclusion and Antisocial Behavior (March 2016)
What would it mean for you to trust an Internet connected product or service and why would you not?
What has damaged consumer trust in products and services and why do sellers care?
What do we want to see different from today, and what is necessary to bring about that change?
These three pairs of questions implicitly underpinned the intense day of #iotmark discussion at the London Zoo last Friday.
The questions went unasked, and could have been voiced before we started, although were probably assumed to be self-evident:
Why do you want one at all [define the problem]?
What needs to change and why [define the future model]?
How do you deliver that and for whom [set out the solution]?
If a group does not agree on the need and drivers for change, there will be no consensus on what that should look like, what the gap is to achieve it, and even less on making it happen.
So who do you want the trustmark to be for, why will anyone want it, and what will need to change to deliver the aims? No one wants a trustmark per se. Perhaps you want what values or promises it embodies to demonstrate what you stand for, promote good practice, and generate consumer trust. To generate trust, you must be seen to be trustworthy. Will the principles deliver on those goals?
The Open IoT Certification Mark Principles, as a rough draft was the outcome of the day, and are available online.
Here’s my reflections, including what was missing on privacy, and the potential for it to be considered in future.
I’ve structured this first, assuming readers attended the event, at ca 1,000 words. Lists and bullet points. The background comes after that, for anyone interested to read a longer piece.
Many thanks upfront, to fellow participants, to the organisers Alexandra D-S and Usman Haque and the colleague who hosted at the London Zoo. And Usman’s Mum. I hope there will be more constructive work to follow, and that there is space for civil society to play a supporting role and critical friend.
The mark didn’t aim to fix the IoT in a day, but deliver something better for product and service users, by those IoT companies and providers who want to sign up. Here is what I took away.
I learned three things
A sense of privacy is not homogenous, even within people who like and care about privacy in theoretical and applied ways. (I very much look forward to reading suggestions promised by fellow participants, even if enforced personal openness and ‘watching the watchers’ may mean ‘privacy is theft‘.)
Awareness of current data protection regulations needs improved in the field. For example, Subject Access Requests already apply to all data controllers, public and private. Few have read the GDPR, or the e-Privacy directive, despite importance for security measures in personal devices, relevant for IoT.
I truly love working on this stuff, with people who care.
And it reaffirmed things I already knew
Change is hard, no matter in what field.
People working together towards a common goal is brilliant.
Group collaboration can create some brilliantly sharp ideas. Group compromise can blunt them.
Some men are particularly bad at talking over each other, never mind over the women in the conversation. Women notice more. (Note to self: When discussion is passionate, it’s hard to hold back in my own enthusiasm and not do the same myself. To fix.)
The IoT context, and risks within it are not homogenous, but brings new risks and adverseries. The risks for manufacturers and consumers and the rest of the public are different, and cannot be easily solved with a one-size-fits-all solution. But we can try.
Concerns I came away with
If the citizen / customer / individual is to benefit from the IoT trustmark, they must be put first, ahead of companies’ wants.
If the IoT group controls both the design, assessment to adherence and the definition of success, how objective will it be?
The group was not sufficiently diverse and as a result, reflects too little on the risks and impact of the lack of diversity in design and effect, and the implications of dataveillance .
Critical minority thoughts although welcomed, were stripped out from crowdsourced first draft principles in compromise.
More future thinking should be built-in to be robust over time.
What was missing
There was too little discussion of privacy in perhaps the most important context of IoT – inter connectivity and new adversaries. It’s not only about *your* thing, but things that it speaks to, interacts with, of friends, passersby, the cityscape , and other individual and state actors interested in offense and defense. While we started to discuss it, we did not have the opportunity to discuss sufficiently at depth to be able to get any thinking into applying solutions in the principles.
One of the greatest risks that users face is the ubiquitous collection and storage of data about users that reveal detailed, inter-connected patterns of behaviour and our identity and not seeing how that is used by companies behind the scenes.
What we also missed discussing is not what we see as necessary today, but what we can foresee as necessary for the short term future, brainstorming and crowdsourcing horizon scanning for market needs and changing stakeholder wants.
Future thinking
Here’s the areas of future thinking that smart thinking on the IoT mark could consider.
We are moving towards ever greater requirements to declare identity to use a product or service, to register and log in to use anything at all. How will that change trust in IoT devices?
Single identity sign-on is becoming ever more imposed, and any attempts for multiple presentation of who I am by choice, and dependent on context, therefore restricted. [not all users want to use the same social media credentials for online shopping, with their child’s school app, and their weekend entertainment]
Is this imposition what the public wants or what companies sell us as what customers want in the name of convenience? What I believe the public would really want is the choice to do neither.
There is increasingly no private space or time, at places of work.
Limitations on private space are encroaching in secret in all public city spaces. How will ‘handoffs’ affect privacy in the IoT?
There is too little understanding of the social effects of this connectedness and knowledge created, embedded in design.
What effects may there be on the perception of the IoT as a whole, if predictive data analysis and complex machine learning and AI hidden in black boxes becomes more commonplace and not every company wants to be or can be open-by-design?
Ubiquitous collection and storage of data about users that reveal detailed, inter-connected patterns of behaviour and our identity needs greater commitments to disclosure. Where the hand-offs are to other devices, and whatever else is in the surrounding ecosystem, who has responsibility for communicating interaction through privacy notices, or defining legitimate interests, where the data joined up may be much more revealing than stand-alone data in each silo?
Define with greater clarity the privacy threat models for different groups of stakeholders and address the principles for each.
What would better look like?
The draft privacy principles are a start, but they’re not yet aspirational as I would have hoped. Of course the principles will only be adopted if possible, practical and by those who choose to. But where is the differentiator from what everyone is required to do, and better than the bare minimum? How will you sell this to consumers as new? How would you like your child to be treated?
The wording in these 5 bullet points, is the first crowdsourced starting point.
The supplier of this product or service MUST be General Data Protection Regulation (GDPR) compliant.
This product SHALL NOT disclose data to third parties without my knowledge.
I SHOULD get full access to all the data collected about me.
I MAY operate this device without connecting to the internet.
My data SHALL NOT be used for profiling, marketing or advertising without transparent disclosure.
Yes other points that came under security address some of the crossover between privacy and surveillance risks, but there is as yet little substantial that is aspirational to make the IoT mark a real differentiator in terms of privacy. An opportunity remains.
It was that and how young people perceive privacy that I hoped to bring to the table. Because if manufacturers are serious about future success, they cannot ignore today’s children and how they feel. How you treat them today, will shape future purchasers and their purchasing, and there is evidence you are getting it wrong.
The timing is good in that it now also offers the opportunity to promote consistent understanding, and embed the language of GDPR and ePrivacy regulations into consistent and compatible language in policy and practice in the #IoTmark principles.
User rights I would like to see considered
These are some of the points I would think privacy by design would mean. This would better articulate GDPR Article 25 to consumers.
Data sovereignty is a good concept and I believe should be considered for inclusion in explanatory blurb before any agreed privacy principles.
Goods should by ‘dumb* by default’ until the smart functionality is switched on. [*As our group chair/scribe called it] I would describe this as, “off is the default setting out-of-the-box”.
Privact by design. Deniability by default. i.e. not only after opt out, but a company should not access the personal or identifying purchase data of anyone who opts out of data collection about their product/service use during the set up process.
The right to opt out of data collection at a later date while continuing to use services.
A right to object to the sale or transfer of behavioural data, including to third-party ad networks and absolute opt-in on company transfer of ownership.
A requirement that advertising should be targeted to content, [user bought fridge A] not through jigsaw data held on users by the company [how user uses fridge A, B, C and related behaviour].
An absolute rejection of using children’s personal data gathered to target advertising and marketing at children
Background: Starting points before privacy
After a brief recap on 5 years ago, we heard two talks.
The first was a presentation from Bosch. They used the insights from the IoT open definition from 5 years ago in their IoT thinking and embedded it in their brand book. The presenter suggested that in five years time, every fridge Bosch sells will be ‘smart’. And the second was a fascinating presentation, of both EU thinking and the intellectual nudge to think beyond the practical and think what kind of society we want to see using the IoT in future. Hints of hardcore ethics and philosophy that made my brain fizz from Gerald Santucci, soon to retire from the European Commission.
The principles of open sourcing, manufacturing, and sustainable life cycle were debated in the afternoon with intense arguments and clearly knowledgeable participants, including those who were quiet. But while the group had assigned security, and started work on it weeks before, there was no one pre-assigned to privacy. For me, that said something. If they are serious about those who earn the trustmark being better for customers than their competition, then there needs to be greater emphasis on thinking like their customers, and by their customers, and what use the mark will be to customers, not companies. Plan early public engagement and testing into the design of this IoT mark, and make that testing open and diverse.
To that end, I believe it needed to be articulated more strongly, that sustainable public trust is the primary goal of the principles.
Trust that my device will not become unusable or worthless through updates or lack of them.
Trust that my device is manufactured safely and ethically and with thought given to end of life and the environment.
Trust that my source components are of high standards.
Trust in what data and how that data is gathered and used by the manufacturers.
Fundamental to ‘smart’ devices is their connection to the Internet, and so the last for me, is therefore key to successful public perception and it actually making a difference, beyond the PR value to companies. The value-add must be measured from consumers point of view.
All the openness about design functions and practice improvements, without attempting to change privacy infringing practices, may be wasted effort. Why? Because the perceived benefit of the value of the mark, will be proportionate to what risks it is seen to mitigate.
Why?
Because I assume that you know where your source components come from today. I was shocked to find out not all do and that ‘one degree removed’ is going to be an improvement? Holy cow, I thought. What about regulatory requirements for product safety recalls? These differ of course for different product areas, but I was still surprised. Having worked in global Fast Moving Consumer Goods (FMCG) and food industry, semiconductor and optoelectronics, and medical devices it was self-evident for me, that sourcing is rigorous. So that new requirement to know one degree removed, was a suggested minimum. But it might shock consumers to know there is not usually more by default.
Customers also believe they have reasonable expectations of not being screwed by a product update, left with something that does not work because of its computing based components. The public can take vocal, reputation-damaging action when they are let down.
While these are visible, the full extent of the overreach of company market and product surveillance into our whole lives, not just our living rooms, is yet to become understood by the general population. What will happen when it is?
The Internet of Things is exacerbating the power imbalance between consumers and companies, between government and citizens. As Wendy Grossman wrote recently, in one sense this may make privacy advocates’ jobs easier. It was always hard to explain why “privacy” mattered. Power, people understand.
That public discussion is long overdue. If open principles on IoT devices mean that the signed-up companies differentiate themselves by becoming market leaders in transparency, it will be a great thing. Companies need to offer full disclosure of data use in any privacy notices in clear, plain language under GDPR anyway, but to go beyond that, and offer customers fair presentation of both risks and customer benefits, will not only be a point-of-sales benefit, but potentially improve digital literacy in customers too.
The morning discussion touched quite often on pay-for-privacy models. While product makers may see this as offering a good thing, I strove to bring discussion back to first principles.
Privacy is a human right. There can be no ethical model of discrimination based on any non-consensual invasion of privacy. Privacy is not something I should pay to have. You should not design products that reduce my rights. GDPR requires privacy-by-design and data protection by default. Now is that chance for IoT manufacturers to lead that shift towards higher standards.
We also need a new ethics thinking on acceptable fair use. It won’t change overnight, and perfect may be the enemy of better. But it’s not a battle that companies should think consumers have lost. Human rights and information security should not be on the battlefield at all in the war to win customer loyalty. Now is the time to do better, to be better, demand better for us and in particular, for our children.
Privacy will be a genuine market differentiator
If manufacturers do not want to change their approach to exploiting customer data, they are unlikely to be seen to have changed.
Today feelings that people in US and Europe reflect in surveys are loss of empowerment, feeling helpless, and feeling used. That will shift to shock, resentment, and any change curve will predict, anger.
“The poll of just over two thousand British adults carried out by Ipsos MORI found that the media, internet services such as social media and search engines and telecommunication companies were the least trusted to use personal data appropriately.” [2014, Data trust deficit with lessons for policymakers, Royal Statistical Society]
In the British student population, one 2015 survey of university applicants in England, found of 37,000 who responded, the vast majority of UCAS applicants agree that sharing personal data can benefit them and support public benefit research into university admissions, but they want to stay firmly in control. 90% of respondents said they wanted to be asked for their consent before their personal data is provided outside of the admissions service.
In 2010, a multi method model of research with young people aged 14-18, by the Royal Society of Engineering, found that, “despite their openness to social networking, the Facebook generation have real concerns about the privacy of their medical records.” [2010, Privacy and Prejudice, RAE, Wellcome]
When people use privacy settings on Facebook set to maximum, they believe they get privacy, and understand little of what that means behind the scenes.
Are there tools designed by others, like Projects by If licenses, and ways this can be done, that you’re not even considering yet?
What if you don’t do it?
“But do you feel like you have privacy today?” I was asked the question in the afternoon. How do people feel today, and does it matter? Companies exploiting consumer data and getting caught doing things the public don’t expect with their data, has repeatedly damaged consumer trust. Data breaches and lack of information security have damaged consumer trust. Both cause reputational harm. Damage to reputation can harm customer loyalty. Damage to customer loyalty costs sales, profit and upsets the Board.
Where overreach into our living rooms has raised awareness of invasive data collection, we are yet to be able to see and understand the invasion of privacy into our thinking and nudge behaviour, into our perception of the world on social media, the effects on decision making that data analytics is enabling as data shows companies ‘how we think’, granting companies access to human minds in the abstract, even before Facebook is there in the flesh.
Governments want to see how we think too, and is thought crime really that far away using database labels of ‘domestic extremists’ for activists and anti-fracking campaigners, or the growing weight of policy makers attention given to predpol, predictive analytics, the [formerly] Cabinet Office Nudge Unit, Google DeepMind et al?
Had the internet remained decentralized the debate may be different.
I am starting to think of the IoT not as the Internet of Things, but as the Internet of Tracking. If some have their way, it will be the Internet of Thinking.
Considering our centralised Internet of Things model, our personal data from human interactions has become the network infrastructure, and data flows, are controlled by others. Our brains are the new data servers.
In the Internet of Tracking, people become the end nodes, not things.
And it is this where the future users will be so important. Do you understand and plan for factors that will drive push back, and crash of consumer confidence in your products, and take it seriously?
Companies have a choice to act as Empires would – multinationals, joining up even on low levels, disempowering individuals and sucking knowledge and power at the centre. Or they can act as Nation states ensuring citizens keep their sovereignty and control over a selected sense of self.
Look at Brexit. Look at the GE2017. Tell me, what do you see is the direction of travel? Companies can fight it, but will not defeat how people feel. No matter how much they hope ‘nudge’ and predictive analytics might give them this power, the people can take back control.
What might this desire to take-back-control mean for future consumer models? The afternoon discussion whilst intense, reached fairly simplistic concluding statements on privacy. We could have done with at least another hour.
Some in the group were frustrated “we seem to be going backwards” in current approaches to privacy and with GDPR.
But if the current legislation is reactive because companies have misbehaved, how will that be rectified for future? The challenge in the IoT both in terms of security and privacy, AND in terms of public perception and reputation management, is that you are dependent on the behaviours of the network, and those around you. Good and bad. And bad practices by one, can endanger others, in all senses.
If you believe that is going back to reclaim a growing sense of citizens’ rights, rather than accepting companies have the outsourced power to control the rights of others, that may be true.
There was a first principle asked whether any element on privacy was needed at all, if the text was simply to state, that the supplier of this product or service must be General Data Protection Regulation (GDPR) compliant. The GDPR was years in the making after all. Does it matter more in the IoT and in what ways? The room tended, understandably, to talk about it from the company perspective. “We can’t” “won’t” “that would stop us from XYZ.” Privacy would however be better addressed from the personal point of view.
What do people want?
From the company point of view, the language is different and holds clues. Openness, control, and user choice and pay for privacy are not the same thing as the basic human right to be left alone. Afternoon discussion reminded me of the 2014 WAPO article, discussing Mark Zuckerberg’s theory of privacy and a Palo Alto meeting at Facebook:
“Not one person ever uttered the word “privacy” in their responses to us. Instead, they talked about “user control” or “user options” or promoted the “openness of the platform.” It was as if a memo had been circulated that morning instructing them never to use the word “privacy.””
In the afternoon working group on privacy, there was robust discussion whether we had consensus on what privacy even means. Words like autonomy, control, and choice came up a lot. But it was only a beginning. There is opportunity for better. An academic voice raised the concept of sovereignty with which I agreed, but how and where to fit it into wording, which is at once both minimal and applied, and under a scribe who appeared frustrated and wanted a completely different approach from what he heard across the group, meant it was left out.
This group do care about privacy. But I wasn’t convinced that the room cared in the way that the public as a whole does, but rather only as consumers and customers do. But IoT products will affect potentially everyone, even those who do not buy your stuff. Everyone in that room, agreed on one thing. The status quo is not good enough. What we did not agree on, was why, and what was the minimum change needed to make a enough of a difference that matters.
I share the deep concerns of many child rights academics who see the harm that efforts to avoid restrictions Article 8 the GDPR will impose. It is likely to be damaging for children’s right to access information, be discriminatory according to parents’ prejudices or socio-economic status, and ‘cheating’ – requiring secrecy rather than privacy, in attempts to hide or work round the stringent system.
In ‘The Class’ the research showed, ” teachers and young people have a lot invested in keeping their spheres of interest and identity separate, under their autonomous control, and away from the scrutiny of each other.” [2016, Livingstone and Sefton-Green, p235]
Employers require staff use devices with single sign including web and activity tracking and monitoring software. Employee personal data and employment data are blended. Who owns that data, what rights will employees have to refuse what they see as excessive, and is it manageable given the power imbalance between employer and employee?
What is this doing in the classroom and boardroom for stress, anxiety, performance and system and social avoidance strategies?
A desire for convenience creates shortcuts, and these are often met using systems that require a sign-on through the platforms giants: Google, Facebook, Twitter, et al. But we are kept in the dark how by using these platforms, that gives access to them, and the companies, to see how our online and offline activity is all joined up.
Any illusion of privacy we maintain, we discussed, is not choice or control if based on ignorance, and backlash against companies lack of efforts to ensure disclosure and understanding is growing.
“The lack of accountability isn’t just troubling from a philosophical perspective. It’s dangerous in a political climate where people are pushing back at the very idea of globalization. There’s no industry more globalized than tech, and no industry more vulnerable to a potential backlash.”
If your connected *thing* requires registration, why does it? How about a commitment to not forcing one of these registration methods or indeed any at all? Social Media Research by Pew Research in 2016 found that 56% of smartphone owners ages 18 to 29 use auto-delete apps, more than four times the share among those 30-49 (13%) and six times the share among those 50 or older (9%).
Does that tell us anything about the demographics of data retention preferences?
In 2012, they suggested social media has changed the public discussion about managing “privacy” online. When asked, people say that privacy is important to them; when observed, people’s actions seem to suggest otherwise.
Does that tell us anything about how well companies communicate to consumers how their data is used and what rights they have?
There is also data with strong indications about how women act to protect their privacy more but when it comes to basic privacy settings, users of all ages are equally likely to choose a private, semi-private or public setting for their profile. There are no significant variations across age groups in the US sample.
Now think about why that matters for the IoT? I wonder who makes the bulk of purchasing decsions about household white goods for example and has Bosch factored that into their smart-fridges-only decision?
Do you *need* to know who the user is? Can the smart user choose to stay anonymous at all?
The day’s morning challenge was to attend more than one interesting discussion happening at the same time. As invariably happens, the session notes and quotes are always out of context and can’t possibly capture everything, no matter how amazing the volunteer (with thanks!). But here are some of the discussion points from the session on the body and health devices, the home, and privacy. It also included a discussion on racial discrimination, algorithmic bias, and the reasons why care.data failed patients and failed as a programme. We had lengthy discussion on ethics and privacy: smart meters, objections to models of price discrimination, and why pay-for-privacy harms the poor by design.
Smart meter data can track the use of unique appliances inside a person’s home and intimate patterns of behaviour. Information about our consumption of power, what and when every day, reveals personal details about everyday lives, our interactions with others, and personal habits.
Why should company convenience come above the consumer’s? Why should government powers, trump personal rights?
Smart meter is among the knowledge that government is exploiting, without consent, to discover a whole range of issues, including ensuring that “Troubled Families are identified”. Knowing how dodgy some of the school behaviour data might be, that helps define who is “troubled” there is a real question here, is this sound data science? How are errors identified? What about privacy? It’s not your policy, but if it is your product, what are your responsibilities?
If companies do not respect children’s rights, you’d better shape up to be GDPR compliant
For children and young people, more vulnerable to nudge, and while developing their sense of self can involve forming, and questioning their identity, these influences need oversight or be avoided.
In terms of GDPR, providers are going to pay particular attention to Article 8 ‘information society services’ and parental consent, Article 17 on profiling, and rights to restriction of processing (19) right to erasure in recital 65 and rights to portability. (20) However, they may need to simply reassess their exploitation of children and young people’s personal data and behavioural data. Article 57 requires special attention to be paid by regulators to activities specifically targeted at children, as ‘vulnerable natural persons’ of recital 75.
Human Rights, regulations and conventions overlap in similar principles that demand respect for a child, and right to be let alone:
(a) The development of the child ‘s personality, talents and mental and physical abilities to their fullest potential;
(b) The development of respect for human rights and fundamental freedoms, and for the principles enshrined in the Charter of the United Nations.
A weakness of the GDPR is that it allows derogation on age and will create inequality and inconsistency for children as a result. By comparison Article one of the Convention on the Rights of the Child (CRC) defines who is to be considered a “child” for the purposes of the CRC, and states that: “For the purposes of the present Convention, a child means every human being below the age of eighteen years unless, under the law applicable to the child, majority is attained earlier.”<
Article two of the CRC says that States Parties shall respect and ensure the rights set forth in the present Convention to each child within their jurisdiction without discrimination of any kind.
CRC Article 16 says that no child shall be subjected to arbitrary or unlawful interference with his or her honour and reputation.
Article 8 CRC requires respect for the right of the child to preserve his or her identity […] without unlawful interference.
Article 12 CRC demands States Parties shall assure to the child who is capable of forming his or her own views the right to express those views freely in all matters affecting the child, the views of the child being given due weight in accordance with the age and maturity of the child.
That stands in potential conflict with GDPR article 8. There is much on GDPR on derogations by country, and or children, still to be set.
What next for our data in the wild
Hosting the event at the zoo offered added animals, and during a lunch tour we got out on a tour, kindly hosted by a fellow participant. We learned how smart technology was embedded in some of the animal enclosures, and work on temperature sensors with penguins for example. I love tigers, so it was a bonus that we got to see such beautiful and powerful animals up close, if a little sad for their circumstances and as a general basic principle, seeing big animals caged as opposed to in-the-wild.
Freedom is a common desire in all animals. Physical, mental, and freedom from control by others.
I think any manufacturer that underestimates this element of human instinct is ignoring the ‘hidden dragon’ that some think is a myth. Privacy is not dead. It is not extinct, or even unlike the beautiful tigers, endangered. Privacy in the IoT at its most basic, is the right to control our purchasing power. The ultimate people power waiting to be sprung. Truly a crouching tiger. People object to being used and if companies continue to do so without full disclosure, they do so at their peril. Companies seem all-powerful in the battle for privacy, but they are not. Even insurers and data brokers must be fair and lawful, and it is for regulators to ensure that practices meet the law.
When consumers realise our data, our purchasing power has the potential to control, not be controlled, that balance will shift.
“Paper tigers” are superficially powerful but are prone to overextension that leads to sudden collapse. If that happens to the superficially powerful companies that choose unethical and bad practice, as a result of better data privacy and data ethics, then bring it on.
I hope that the IoT mark can champion best practices and make a difference to benefit everyone.
While the companies involved in its design may be interested in consumers, I believe it could be better for everyone, done well. The great thing about the efforts into an #IoTmark is that it is a collective effort to improve the whole ecosystem.
I hope more companies will realise their privacy rights and ethical responsibility in the world to all people, including those interested in just being, those who want to be let alone, and not just those buying.
“If a cat is called a tiger it can easily be dismissed as a paper tiger; the question remains however why one was so scared of the cat in the first place.”
Further reading: Networks of Control – A Report on Corporate Surveillance, Digital Tracking, Big Data & Privacy by Wolfie Christl and Sarah Spiekermann
Mike Loukides drew similarities between the current status of AI and children’s learning in an article I read this week.
The children I know are always curious to know where they are going, how long will it take, and how they will know when they get there. They ask others for guidance often.
Loukides wrote that if you look carefully at how humans learn, you see surprisingly little unsupervised learning.
If unsupervised learning is a prerequisite for general intelligence, but not the substance, what should we be looking for, he asked. It made me wonder is it also true that general intelligence is a prerequisite for unsupervised learning? And if so, what level of learning must AI achieve before it is capable of recursive self-improvement? What is AI being encouraged to look for as it learns, what is it learning as it looks?
What is AI looking for and how will it know when it gets there?
Loukides says he can imagine a toddler learning some rudiments of counting and addition on his or her own, but can’t imagine a child developing any sort of higher mathematics without a teacher.
I suggest a different starting point. I think children develop on their own, given a foundation. And if the foundation is accompanied by a purpose — to understand why they should learn to count, and why they should want to — and if they have the inspiration, incentive and assets they’ll soon go off on their own, and outstrip your level of knowledge. That may or may not be with a teacher depending on what is available, cost, and how far they get compared with what they want to achieve.
It’s hard to learn something from scratch by yourself if you have no boundaries to set knowledge within and search for more, or to know when to stop when you have found it.
You’ve only to start an online course, get stuck, and try to find the solution through a search engine to know how hard it can be to find the answer if you don’t know what you’re looking for. You can’t type in search terms if you don’t know the right words to describe the problem.
I described this recently to a fellow codebar-goer, more experienced than me, and she pointed out something much better to me. Don’t search for the solution or describe what you’re trying to do, ask the search engine to find others with the same error message.
In effect she said, your search is wrong. Google knows the answer, but can’t tell you what you want to know, if you don’t ask it in the way it expects.
So what will AI expect from people and will it care if we dont know how to interrelate? How does AI best serve humankind and defined by whose point-of-view? Will AI serve only those who think most closely in AI style steps and language? How will it serve those who don’t know how to talk about, or with it? AI won’t care if we don’t.
If as Loukides says, we humans are good at learning something and then applying that knowledge in a completely different area, it’s worth us thinking about how we are transferring our knowledge today to AI and how it learns from that. Not only what does AI learn in content and context, but what does it learn about learning?
His comparison of a toddler learning from parents — who in effect are ‘tagging’ objects through repetition of words while looking at images in a picture book — made me wonder how we will teach AI the benefit of learning? What incentive will it have to progress?
“the biggest project facing AI isn’t making the learning process faster and more efficient. It’s moving from machines that solve one problem very well (such as playing Go or generating imitation Rembrandts) to machines that are flexible and can solve many unrelated problems well, even problems they’ve never seen before.”
Is the skill to enable “transfer learning” what will matter most?
For AI to become truly useful, we need better as a global society to understand *where* it might best interface with our daily lives, and most importantly *why*. And consider *who* is teaching and AI and who is being left out in the crowdsourcing of AI’s teaching.
Who is teaching AI what it needs to know?
The natural user interfaces for people to interact with today’s more common virtual assistants (Amazon’s Alexa, Apple’s Siri and Viv, Microsoft and Cortana) are not just providing information to the user, but through its use, those systems are learning. I wonder what percentage of today’s population is using these assistants, how representative are they, and what our AI assistants are being taught through their use? Tay was a swift lesson learned for Microsoft.
In helping shape what AI learns, what range of language it will use to develop its reference words and knowledge, society co-shapes what AI’s purpose will be — and for AI providers to know what’s the point of selling it. So will this technology serve everyone?
Are providers counter-balancing what AI is currently learning from crowdsourcing, if the crowd is not representative of society?
So far we can only teach machines to make decisions based on what we already know, and what we can tell it to decide quickly against pre-known references using lots of data. Will your next image captcha, teach AI to separate the sloth from the pain-au-chocolat?
One of the task items for machine processing is better searches. Measurable goal driven tasks have boundaries, but who sets them? When does a computer know, if it’s found enough to make a decision. If the balance of material about the Holocaust on the web for example, were written by Holocaust deniers will AI know who is right? How will AI know what is trusted and by whose measure?
What will matter most is surely not going to be how to optimise knowledge transfer from human to AI — that is the baseline knowledge of supervised learning — and it won’t even be for AI to know when to use its skill set in one place and when to apply it elsewhere in a different context; so-called learning transfer, as Mike Loukides says. But rather, will AI reach the point where it cares?
Will AI ever care what it should know and where to stop or when it knows enough on any given subject?
How will it know or care if what it learns is true?
If in the best interests of advancing technology or through inaction we do not limit its boundaries, what oversight is there of its implications?
Online limits will limit what we can reach in Thinking and Learning
If you look carefully at how humans learn online, I think rather than seeing surprisingly little unsupervised learning, you see a lot of unsupervised questioning. It is often in the questioning that is done in private we discover, and through discovery we learn. Often valuable discoveries are made; whether in science, in maths, or important truths are found where there is a need to challenge the status quo. Imagine if Galileo had given up.
The freedom to think freely and to challenge authority, is vital to protect, and one reason why I and others are concerned about the compulsory web monitoring starting on September 5th in all schools in England, and its potential chilling effect. Some are concerned who might have access to these monitoring results today or in future, if stored could they be opened to employers or academic institutions?
If you tell children do not use these search terms and do not be curious about *this* subject without repercussions, it is censorship. I find the idea bad enough for children, but for us as adults its scary.
As Frankie Boyle wrote last November, we need to consider what our internet history is:
“The legislation seems to view it as a list of actions, but it’s not. It’s a document that shows what we’re thinking about.”
Children think and act in ways that they may not as an adult. People also think and act differently in private and in public. It’s concerning that our private online activity will become visible to the State in the IP Bill — whether photographs that captured momentary actions in social media platforms without the possibility to erase them, or trails of transitive thinking via our web history — and third-parties may make covert judgements and conclusions about us, correctly or not, behind the scenes without transparency, oversight or recourse.
By narrowing our parameters what will we not discover? Not debate? Or not invent? Happy are the clockmakers, and kids who create. Any restriction on freedom to access information, to challenge and question will restrict children’s learning or even their wanting to. It will limit how we can improve our shared knowledge and improve our society as a result. The same is true of adults.
So in teaching AI how to learn, I wonder how the limitations that humans put on its scope — otherwise how would it learn what the developers want — combined with showing it ‘our thinking’ through search terms, and how limitations on that if users self-censor due to surveillance, will shape what AI will help us with in future and will it be the things that could help the most people, the poorest people, or will it be people like those who programme the AI and use search terms and languages it already understands?
Who is accountable for the scope of what we allow AI to do or not? Who is accountable for what AI learns about us, from our behaviour data if it is used without our knowledge?
How far does AI have to go?
The leap for AI will be if and when AI can determine what it doesn’t know, and it sees a need to fill that gap. To do that, AI will need to discover a purpose for its own learning, indeed for its own being, and be able to do so without limitation from the that humans shaped its framework for doing so. How will AI know what it needs to know and why? How will it know, what it knows is right and sources to trust? Against what boundaries will AI decide what it should engage with in its learning, who from and why? Will it care? Why will it care? Will it find meaning in its reason for being? Why am I here?
We assume AI will know better. We need to care, if AI is going to.
How far are we away from a machine that is capable of recursive self-improvement, asks John Naughton in yesterday’s Guardian, referencing work by Yuval Harari suggesting artificial intelligence and genetic enhancements will usher in a world of inequality and powerful elites. As I was finishing this, I read his article, and found myself nodding, as I read the implications of new technology focus too much on technology and too little on society’s role in shaping it.
If the purpose of AI is to improve human lives, who defines improvement and who will that improvement serve? Is there a consensus on the direction AI should and should not take, and how far it should go? What will the global language be to speak AI?
As AI learning progresses, every time AI turns to ask its creators, “Are we there yet?”, how will we know what to say?
image: Stephen Barling flickr.com/photos/cripsyduck (CC BY-NC 2.0)
This blog post is also available as an audio file on soundcloud.
Anais Nin, wrote in her 1946 diary of the dangers she saw in the growth of technology to expand our potential for connectivity through machines, but diminish our genuine connectedness as people. She could hardly have been more contemporary for today:
“This is the illusion that might cheat us of being in touch deeply with the one breathing next to us. The dangerous time when mechanical voices, radios, telephone, take the place of human intimacies, and the concept of being in touch with millions brings a greater and greater poverty in intimacy and human vision.” [Extract from volume IV 1944-1947]
Echoes from over 70 years ago, can be heard in the more recent comments of entrepreneur Elon Musk. Both are concerned with simulation, a lack of connection between the perceived, and reality, and the jeopardy this presents for humanity. But both also have a dream. A dream based on the positive potential society has.
How will we use our potential?
Data is the connection we all have between us as humans and what machines and their masters know about us. The values that masters underpin their machine design with, will determine the effect the machines and knowledge they deliver, have on society.
In seeking ever greater personalisation, a wider dragnet of data is putting together ever more detailed pieces of information about an individual person. At the same time data science is becoming ever more impersonal in how we treat people as individuals. We risk losing sight of how we respect and treat the very people whom the work should benefit.
Nin grasped the risk that a wider reach, can mean more superficial depth. Facebook might be a model today for the large circle of friends you might gather, but how few you trust with confidences, with personal knowledge about your own personal life, and the privilege it is when someone chooses to entrust that knowledge to you. Machine data mining increasingly tries to get an understanding of depth, and may also add new layers of meaning through profiling, comparing our characteristics with others in risk stratification.
Data science, research using data, is often talked about as if it is something separate from using information from individual people. Yet it is all about exploiting those confidences.
Today as the reach has grown in what is possible for a few people in institutions to gather about most people in the public, whether in scientific research, or in surveillance of different kinds, we hear experts repeatedly talk of the risk of losing the valuable part, the knowledge, the insights that benefit us as society if we can act upon them.
We might know more, but do we know any better? To use a well known quote from her contemporary, T S Eliot, ‘Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?’
What can humans achieve? We don’t yet know our own limits. What don’t we yet know? We have future priorities we aren’t yet aware of.
To be able to explore the best of what Nin saw as ‘human vision’ and Musk sees in technology, the benefits we have from our connectivity; our collaboration, shared learning; need to be driven with an element of humility, accepting values that shape boundaries of what we should do, while constantly evolving with what we could do.
The essence of this applied risk is that technology could harm you, more than it helps you. How do we avoid this and develop instead the best of what human vision makes possible? Can we also exceed our own expectations of today, to advance in moral progress?
Atlas, the Boston Dynamics created robot, won hearts and minds this week as it stoically survived man being mean. Our collective human response was an emotional defence of the machine, and criticism of its unfair treatment by its tester.
The concepts of fairness and of decision making algorithms for ‘abuse avoidance’ are interesting from perspectives of data mining, AI and the wider access to and use of tech in general, and in health specifically.
If the decision to avoid abuse can be taken out of an individual’s human hands and are based on unfathomable amounts of big data, where are its limits applied to human behaviour and activity?
When it is decided that an individual’s decision making capability is impaired or has been forfeited their consent may be revoked in their best interest.
Who has oversight of the boundaries of what is acceptable for one person, or for an organisation, to decide what is in someone else’s best interest, or indeed, the public interest?
Where these boundaries overlap – personal abuse avoidance, individual best interest and the public interest – and how society manage them, with what oversight, is yet to be widely debated.
We must get involved and it must be the start of a debate and dialogue not simply a tick-box to a done-deal, if data derived from us are to be used as a platform for future to “achieve great results for the NHS and everyone who depends on it.”
Administering applied “abuse avoidance” and Restraining Abilities
Administrative uses and secondary research using the public’s personal data are applied not only in health, but across the board of public bodies, including big plans for tech in the justice system.
The use of this technology as a monitoring tool, should not of itself be a punishment. It is said compliance is not intended to affectthedignity of individuals who are being monitored, but through the collection of personal and health data will ensure the deprivation of alcohol – avoiding its abuse for a person’s own good and in the public interest. Is it fair?
Abstinence orders might be applied to those convicted of crimes such as assault, being drunk and disorderly and drunk driving.
We’re yet to see much discussion of how these varying degrees of integration of tech with the human body, and human enhancement will happen through robot elements in our human lives.
How will the boundaries of what is possible and desirable be determined and by whom with what oversight?
What else might be considered as harmful as alcohol to individuals and to society? Drugs? Nictotine? Excess sugar?
As we wonder about the ethics of how humanoids will act and the aesthetics of how human they look, I wonder how humane are we being, in all our ‘public’ tech design and deployment?
Umberto Eco who died on Friday wrote in ‘The birth of ethics’ that there are universal ideas on constraints, effectively that people should not harm other people, through deprivation, restrictions or psychological torture. And that we should not impose anything on others that “diminishes or stifles our capacity to think.”
How will we as a society collectively agree what that should look like, how far some can impose on others, without consent?
Enhancing the Boundaries of Being Human
Technology might be used to impose bodily boundaries on some people, but tech can also be used for the enhancement of others. Antonio Santos retweeted this week, the brilliant Angel Giuffria’s arm.
While the technology in this case is literally hands-on in its application, increasingly it is not the technology itself but the data that it creates or captures which enables action through data-based decision making.
Robots that are tiny may be given big responsibilities to monitor and report massive amounts of data. What if we could swallow them?
Data if analysed and understood, become knowledge.
Knowledge can be used to inform decisions and take action.
So where are the boundaries of what data may be extracted, information collated, and applied as individual interventions?
Defining the Boundaries of “in the Public Interest”
Where are boundaries of what data may be created, stored, and linked to create a detailed picture about us as individuals, if the purpose is determined to be in the public interest?
Who decides which purposes are in the public interest? What qualifies as research purposes? Who qualifies as meeting the criteria of ‘researcher’?
How far can research and interventions go without consent?
Should security services and law enforcement agencies always be entitled to get access to individuals’ data ‘in the public interest’?
That’s something Apple is currently testing in the US.
Should research bodies always be entitled to get access to individuals’ data ‘in the public interest’?
That’s something care.data tried and failed to assume the public supported and has yet to re-test. Impossible before respecting the opt out that was promised over two years ago in March 2014.
The question how much data research bodies may be ‘entitled to’ will be tested again in the datasharing consultation in the UK.
Where is the boundary between access and use of data not in enforcement of acts already committed but in their prediction and prevention?
If you believe there should be an assumption of law enforcement access to data when data are used for prediction and prevention, what about health?
Should there be any difference between researchers’ access to data when data are used for past analysis and for use in prediction?
If ethics define the boundary between what is acceptable and where actions by one person may impose something on another that “diminishes or stifles our capacity to think” – that takes away our decision making capacity – that nudges behaviour, or acts on behaviour that has not yet happened, who decides what is ethical?
How does a public that is poorly informed about current data practices, become well enough informed to participate in the debate of how data management should be designed today for their future?
How Deeply Mined should our Personal Data be?
The application of technology, non-specific but not yet AI, was also announced this week in the Google DeepMind work in the NHS.
Its first key launch app co-founder provided a report that established the operating framework for the Behavioural Insights Team established by Prime Minister David Cameron.
A number of highly respected public figures have been engaged to act in the public interest as unpaid Independent Reviewers of Google DeepMind Health. It will be interesting to see what their role is and how transparent its workings and public engagement will be.
The recent consultation on the NHS gave overwhelming feedback that the public does not support the direction of current NHS change. Even having removed all responses associated with ‘lefty’ campaigns, concerns listed on page 11, are consistent including a request the Government “should end further involvement of the private sector in healthcare”. It appears from the response that this engagement exercise will feed little into practice.
The strength of feeling should however be a clear message to new projects that people are passionate that equal access to healthcare for all matters and that the public wants to be informed and have their voices heard.
How will public involvement be ensured as complexity increases in these healthcare add-ons and changing technology?
Will Google DeepMind pave the way to a new approach to health research? A combination of ‘nudge’ behavioural insights, advanced neural networks, Big Data and technology is powerful. How will that power be used?
I was recently told that if new research is not pushing the boundaries of what is possible and permissible then it may not be worth doing, as it’s probably been done before.
Should anything that is new that becomes possible be realised?
I wonder how the balance will be weighted in requests for patient data and their application, in such a high profile project.
Will NHS Research Ethics Committees turn down research proposals in-house in hospitals that benefit the institution or advance their reputation, or the HSCIC, ever feel able to say no to data use by Google DeepMind?
Ethics committees safeguard the rights, safety, dignity and well-being of research participants, independently of research sponsors whereas these representatives are not all independent of commercial supporters. And it has not claimed it’s trying to be an ethics panel. But oversight is certainly needed.
The boundaries of ownership between what is seen to benefit commercial and state in modern health investment is perhaps more than blurred to an untrained eye. Genomics England – the government’s flagship programme giving commercial access to the genome of 100K people – stockholding companies, data analytics companies, genome analytic companies, genome collection, and human tissue research, commercial and academic research, often share directors, working partnerships and funders. That’s perhaps unsurprising given such a specialist small world.
It’s exciting to think of the possibilities if, “through a focus on patient outcomes, effective oversight, and the highest ethical principles, we can achieve great results for the NHS and everyone who depends on it.”
Where will an ageing society go, if medics can successfully treat more cancer for example? What diseases will be prioritised and others left behind in what is economically most viable to prevent? How much investment will be made in diseases of the poor or in countries where governments cannot afford to fund programmes?
What will we die from instead? What happens when some causes of ‘preventative death’ are deemed more socially acceptable than others? Where might prevention become socially enforced through nudging behaviour into new socially acceptable or ethical norms?
Don’t be Evil
Given the leading edge of the company and its curiosity-by-design to see how far “can we” will reach, “don’t be evil” may be very important. But “be good” might be better. Where is that boundary?
The boundaries of what ‘being human’ means and how Big Data will decide and influence that, are unclear and changing. How will the law and regulation keep up and society be engaged in support?
Data principles such as fairness, keeping data accurate, complete and up-to-date and ensuring data are not excessive retained for no longer than necessary for the purpose are being widely ignored or exempted under the banner of ‘research’.
Can data use retain a principled approach despite this and if we accept commercial users, profit making based on public data, will those principles from academic research remain in practice?
Exempt from the obligation to give a copy of personal data to an individual on request if data are for ‘research’ purposes, data about us and our children, are extracted and stored ‘without us’. Forever. That means in a future that we cannot see, but Google DeepMind among others, is designing.
Lay understanding, and that of many climical professionals is likely to be left far behind if advanced technologies and use of big data decision-making algorithms are hidden in black boxes.
Public transparency of the use of our data and future planned purposes are needed to create trust that these purposes are wise.
Data are increasingly linked and more valuable when identifiable.
Any organisation that wants to future-proof its reputational risk will make sure data collection and use today is with consent, since future outcomes derived are likely to be in interventions for individuals or society. Catching up consent will be hard unless designed in now.
A Dialogue on the Boundaries of Being Human and Big Data
Where the commercial, personal, and public interests are blurred, the highest ethical principles are going to be needed to ensure ‘abuse avoidance’ in the use of new technology, in increased data linkage and resultant data use in research of many different kinds.
How we as a society achieve the benefits of tech and datasharing and where its boundaries lie in “the public interest” needs public debate to co-design the direction we collectively want to partake in.
Once that is over, change needs supported by a method of oversight that is responsive to new technology, data use, and its challenges.
What a channel for ongoing public dialogue, challenge and potentially recourse might look like, should be part of that debate.
It is perhaps meaningful that pro-reform Wellington School last week opted out of some of the greatest uses of Big Data sharing in the UK. League tables. Citing their failures. Deciding they were “in fact, a key driver for poor educational practice.”
Most often we cannot tell from the data provided what we are told those Big Data should be telling us. And we can’t tell if the data are accurate, genuine and reliable.
Yet big companies are making big money selling the dream that Big Data is the key to decision making. Cumulatively through lack of skills to spot inaccuracy, and inability to do necessary interpretation, we’re being misled by what we find in Big Data.
Being misled is devastating for public trust, as the botched beginnings of care.data found in 2014. Trust has come to be understood as vital for future based on datasharing. Public involvement in how we are used in Big Data in the future, needs to include how our data are used in order to trust they are used well. And interpreting those data well is vital. Those lessons of the past and present must be learned, and not forgotten.
It’s time to invest some time in thinking about safeguarding trust in the future, in the unknown, and the unseen.
We need to be told which private companies like Cinven and FFT have copies of datasets like HES, the entire 62m national hospital records, or the NPD, our entire schools database population of 20 million, or even just its current cohort of 8+ million.
If the public is to trust the government and public bodies to use our data well, we need to know exactly how those data are used today and all these future plans that others have for our personal data.
When we talk about public bodies sharing data they hold for administrative purposes, do we know which private companies this may mean in reality?
The UK government has big plans for big data sharing, sharing across all public bodies, some tailored for individual interventions.
While there are interesting opportunities for public benefit from at-scale systems, the public benefit is at risk not only from lack of trust in how systems gather data and use them, but that interoperability gets lost in market competition.
Openness and transparency can be absent in public-private partnerships until things go wrong. Given the scale of smart-cities, we must have more than hope that data management and security will not be one of those things.
But how will we know if new plans design well, or not?
Who exactly holds and manages those data and where is the oversight of how they are being used?
Using Big Data to be predictive and personal
How do we definde “best use of data” in “public services” right across the board in a world in which boundaries between private and public in the provision of services have become increasingly blurred?
UK researchers and police are already analysing big data for predictive factors at postcode level for those at risk or harm, for example in combining health and education data.
What has grown across the Atlantic is now spreading here. When I lived there I could already see some of what is deeply flawed.
When your system has been as racist in its policing and equity of punishment as institutionally systemic as it is in the US, years of cumulative data bias translates into ‘heat lists’ and means “communities of color will be systematically penalized by any risk assessment tool that uses criminal history as a legitimate criterion.”
How can we ensure British policing does not pursue flawed predictive policies and methodologies, without seeing them?
What transparency have our use of predictive prisons and justice data?
What oversight will the planned new increase in use of satellite tags, and biometrics access in prisons have?
What policies can we have in place to hold data-driven decision-making processes accountable?<
What tools do we need to seek redress for decisions made using flawed algorithms that are apparently indisputable?
Is government truly committed to being open and talking about how far the nudge unit work is incorporated into any government predictive data use? If not, why not?
There is a need for a broad debate on the direction of big data and predictive technology and whether the public understands and wants it.If we don’t understand, it’s time someone explained it.
If I can’t opt out of O2 picking up my travel data ad infinitum on the Tube, I will opt out of their business model and try to find a less invasive provider. If I can’t opt out of EE picking up my personal data as I move around Hyde park, it won’t be them.
Most people just want to be left alone and their space is personal.
A public consultation on smart-technology, and its growth into public space and effect on privacy could be insightful.
Feed me Seymour?
With the encroachment of integrated smart technology over our cities – our roads, our parking, our shopping, our parks, our classrooms, our TV and our entertainment, even our children’s toys – surveillance and sharing information from systems we cannot see start defining what others may view, or decide about us, behind the scenes in everything we do.
As it expands city wide, it will be watched closely if data are to be open for public benefit, but not invade privacy if “The data stored in this infrastructure won’t be confidential.”
If the destination of digital in all parts of our lives is smart-cities then we have to collectively decide, what do we want, what do we design, and how do we keep it democratic?
What price is our freedom to decide how far its growth should reach into public space and private lives?
The cost of smart cities to individuals and the public is not what it costs in investment made by private conglomerates.
Already the cost of smart technology is privacy inside our homes, our finances, and autonomy of decision making.
Facebook and social media may run algorithms we never see that influence our mood or decision making. Influencing that decision making is significant enough when it’s done through advertising encouraging us to decide which sausages to buy for your kids tea.
Who influences most voters wins an election. If we can’t see the technology behind the influence, have we also lost sight of how democracy is decided? The power behind the mechanics of the cogs of Whitehall may weaken inexplicably as computer driven decision from the tech companies’ hidden tools takes hold.
What opportunity and risk to “every part of government” does ever expanding digital bring?
The means the public-interest values that could be built by design and their protection and oversight are currently outside our control.
There is no disincentive for companies that have taken private information that is none of their business, and quite literally, made it their business to not want to collect ever more data about us. It is outside our control.
We must plan by-design for the values we hope for, for ethics, to be embedded in systems, in policies, embedded in public planning and oversight of service provision by all providers. And that the a fair framework of values is used when giving permission to private providers who operate in public spaces.
We must plan for transparency and interoperability.
We must plan by-design for the safe use of data that does not choke creativity and innovation but both protects and champions privacy as a fundamental building block of trust for these new relationships between providers of private and public services, private and public things, in private and public space.
If “digital is changing how we deliver every part of government,” and we want to “harness the best of digital and technology, and the best use of data to improve public services right across the board” then we must see integration in the planning of policy and its application.
Across the board “the best use of data” must truly value privacy, and enable us to keep our autonomy as individuals.
Without this, the cost of smart cities growing unchecked, will be an ever growing transfer of power to the funders behind corporations and campaign politics.
The ultimate price of this loss of privacy, will be democracy itself.
****
This is the conclusion to a four part set of thoughts: On smart technology and data from the Sprint16 session (part one). I thought about this more in depth on “Smart systems and Public Services” here (part two), and the design and development of smart technology making “The Best Use of Data” here looking at today in a UK company case study (part three) and this part four, “The Best Use of Data” used in predictions and the Future.
