Mum, are we there yet? Why should AI care.

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.

Children worry about lack of recourse and repercussions. So do I. Things done in passing, can take on a permanence they never had before and were never intended. If expert providers of the tech world such as Apple Inc, Facebook Inc, Google Inc, Microsoft Corp, Twitter Inc and Yahoo Inc are calling for change, why is the government not listening? This is more than very concerning, it will have disastrous implications for trust in the State, data use by others, self-censorship, and fear that it will lead to outright censorship of adults online too.

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.

AI at the moment has a very broad meaning to the general public. Is it living with life-supporting humanoids?  Do we consider assistive search tools as AI? There is a fairly general understanding of “What is A.I., really?” Some wonder if we are “probably one of the last generations of Homo sapiens,” as we know 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)

Datasharing, lawmaking and ethics: power, practice and public policy

“Lawmaking is the Wire, not Schoolhouse Rock. It’s about blood and war and power, not evidence and argument and policy.”

"We can't trust the regulators," they say. "We need to be able to investigate the data for ourselves." Technology seems to provide the perfect solution. Just put it all online - people can go through the data while trusting no one.  There's just one problem. If you can't trust the regulators, what makes you think you can trust the data?" 

Extracts from The Boy Who Could Change the World: The Writings of Aaron Swartz. Chapter: ‘When is Technology Useful? ‘ June 2009.

The question keeps getting asked, is the concept of ethics obsolete in Big Data?

I’ve come to some conclusions why ‘Big Data’ use keeps pushing the boundaries of what many people find acceptable, and yet the people doing the research, the regulators and lawmakers often express surprise at negative reactions. Some even express disdain for public opinion, dismissing it as ignorant, not ‘understanding the benefits’, yet to be convinced. I’ve decided why I think what is considered ‘ethical’ in data science does not meet public expectation.

It’s not about people.

Researchers using large datasets, often have a foundation in data science, applied computing, maths, and don’t see data as people. It’s only data. Creating patterns, correlations, and analysis of individual level data are not seen as research involving human subjects.

This is embodied in the nth number of research ethics reviews I have read in the last year in which the question is asked, does the research involve people? The answer given is invariably ‘no’.

And these data analysts using, let’s say health data, are not working in a subject that is founded on any ethical principle, contrasting with the medical world the data come from.

The public feels differently about the information that is about them, and may be known, only to them or select professionals. The values that we as the public attach to our data  and expectations of its handling may reflect the expectation we have of handling of us as people who are connected to it. We see our data as all about us.

The values that are therefore put on data, and on how it can and should be used, can be at odds with one another, the public perception is not reciprocated by the researchers. This may be especially true if researchers are using data which has been de-identified, although it may not be anonymous.

New legislation on the horizon, the Better Use of Data in Government,  intends to fill the [loop]hole between what was legal to share in the past and what some want to exploit today, and emphasises a gap in the uses of data by public interest, academic researchers, and uses by government actors. The first incorporate by-and-large privacy and anonymisation techniques by design, versus the second designed for applied use of identifiable data.

Government departments and public bodies want to identify and track people who are somehow misaligned with the values of the system; either through fraud, debt, Troubled Families, or owing Student Loans. All highly sensitive subjects. But their ethical data science framework will not treat them as individuals, but only as data subjects. Or as groups who share certain characteristics.

The system again intrinsically fails to see these uses of data as being about individuals, but sees them as categories of people – “fraud” “debt” “Troubled families.” It is designed to profile people.

Services that weren’t built for people, but for government processes, result in datasets used in research, that aren’t well designed for research. So we now see attempts to shoehorn historical practices into data use  by modern data science practitioners, with policy that is shortsighted.

We can’t afford for these things to be so off axis, if civil service thinking is exploring “potential game-changers such as virtual reality for citizens in the autism spectrum, biometrics to reduce fraud, and data science and machine-learning to automate decisions.”

In an organisation such as DWP this must be really well designed since “the scale at which we operate is unprecedented: with 800 locations and 85,000  colleagues, we’re larger than most retail operations.”

The power to affect individual lives through poor technology is vast and some impacts seem to be being badly ignored. The ‘‘real time earnings’ database improved accuracy of benefit payments was widely agreed to have been harmful to some individuals through the Universal Credit scheme, with delayed payments meaning families at foodbanks, and contributing to worse.

“We believe execution is the major job of every business leader,” perhaps not the best wording in on DWP data uses.

What accountability will be built-by design?

I’ve been thinking recently about drawing a social ecological model of personal data empowerment or control. Thinking about visualisation of wants, gaps and consent models, to show rather than tell policy makers where these gaps exist in public perception and expectations, policy and practice. If anyone knows of one on data, please shout. I think it might be helpful.

But the data *is* all about people

Regardless whether they are in front of you or numbers on a screen, big or small datasets using data about real lives are data about people. And that triggers a need to treat the data with an ethical approach as you would people involved face-to-face.

Researchers need to stop treating data about people as meaningless data because that’s not how people think about their own data being used. Not only that, but if the whole point of your big data research is to have impact, your data outcomes, will change lives.

Tosh, I know some say. But, I have argued, the reason being is that the applications of the data science/ research/ policy findings / impact of immigration in education review / [insert purposes of the data user’s choosing] are designed to have impact on people. Often the people about whom the research is done without their knowledge or consent. And while most people say that is OK, where it’s public interest research, the possibilities are outstripping what the public has expressed as acceptable, and few seem to care.

Evidence from public engagement and ethics all say, hidden pigeon-holing, profiling, is unacceptable. Data Protection law has special requirements for it, on autonomous decisions. ‘Profiling’ is now clearly defined under article 4 of the GDPR as ” any form of automated processing of personal data consisting of using those data to evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects concerning that natural person’s performance at work, economic situation, health, personal preferences, interests, reliability, behaviour, location or movements.”

Using big datasets for research that ‘isn’t interested in individuals’ may still intend to create results profiling groups for applied policing, or discriminate, to make knowledge available by location. The data may have been deidentified, but in application becomes no longer anonymous.

Big Data research that results in profiling groups with the intent for applied health policy impacts for good, may by the very point of research, with the intent of improving a particular ethnic minority access to services, for example.

Then look at the voting process changes in North Carolina and see how that same data, the same research knowledge might be applied to exclude, to restrict rights, and to disempower.

Is it possible to have ethical oversight that can protect good data use and protect people’s rights if they conflict with the policy purposes?

The “clear legal basis”is not enough for public trust

Data use can be legal and can still be unethical, harmful and shortsighted in many ways, for both the impacts on research – in terms of withholding data and falsifying data and avoiding the system to avoid giving in data – and the lives it will touch.

What education has to learn from health is whether it will permit the uses by ‘others’ outside education to jeopardise the collection of school data intended in the best interests of children, not the system. In England it must start to analyse what is needed vs wanted. What is necessary and proportionate and justifies maintaining named data indefinitely, exposed to changing scope.

In health, the most recent Caldicott review suggests scope change by design – that is a red line for many: “For that reason the Review recommends that, in due course, the opt-out should not apply to all flows of information into the HSCIC. This requires careful consideration with the primary care community.”

The community spoke out already, and strongly in Spring and Summer 2014 that there must be an absolute right to confidentiality to protect patients’ trust in the system. Scope that ‘sounds’ like it might sneakily change in future, will be a death knell to public interest research, because repeated trust erosion will be fatal.

Laws change to allow scope change without informing people whose data are being used for different purposes

Regulators must be seen to be trusted, if the data they regulate is to be trustworthy. Laws and regulators that plan scope for the future watering down of public protection, water down public trust from today. Unethical policy and practice, will not be saved by pseudo-data-science ethics.

Will those decisions in private political rooms be worth the public cost to research, to policy, and to the lives it will ultimately affect?

What happens when the ethical black holes in policy, lawmaking and practice collide?

At the last UK HealthCamp towards the end of the day, when we discussed the hard things, the topic inevitably moved swiftly to consent, to building big databases, public perception, and why anyone would think there is potential for abuse, when clearly the intended use is good.

The answer came back from one of the participants, “OK now it’s the time to say. Because, Nazis.” Meaning, let’s learn from history.

Given the state of UK politics, Go Home van policies, restaurant raids, the possibility of Trump getting access to UK sensitive data of all sorts from across the Atlantic, given recent policy effects on the rights of the disabled and others, I wonder if we would hear the gentle laughter in the room in answer to the same question today.

With what is reported as Whitehall’s digital leadership sharp change today, the future of digital in government services and policy and lawmaking does indeed seem to be more “about blood and war and power,” than “evidence and argument and policy“.

The concept of ethics in datasharing using public data in the UK is far from becoming obsolete. It has yet to begin.

We have ethical black holes in big data research, in big data policy, and big data practices in England. The conflicts between public interest research and government uses of population wide datasets, how the public perceive the use of our data and how they are used, gaps and tensions in policy and practice are there.

We are simply waiting for the Big Bang. Whether it will be creative, or destructive we are yet to feel.

*****

image credit: LIGO – graphical visualisation of black holes on the discovery of gravitational waves

References:

Report: Caldicott review – National Data Guardian for Health and Care Review of Data Security, Consent and Opt-Outs 2016

Report: The OneWay Mirror: Public attitudes to commercial access to health data

Royal Statistical Society Survey carried out by Ipsos MORI: The Data Trust Deficit