Sir Nigel Shadbolt is chairman of the Open Data Institute which he co-founded with Sir Tim Berners-Lee. Nigel is one of the UK’s foremost computer scientists. He is a leading researcher in artificial intelligence and was one of the originators of the interdisciplinary field of web science. He is Principal of Jesus College Oxford, a Professor of Computer Science at the University of Oxford and a visiting Professor of Artificial Intelligence at the University of Southampton.
In 2009 the Prime Minister appointed him and Sir Tim Berners-Lee as Information Advisors to transform access to Public Sector Information. This work led to the highly acclaimed data.gov.uk site that now provides a portal to tens of thousands of datasets. In 2010, he joined the UK government’s Public Sector Transparency Board — overseeing Open Data releases across the public sector. He is a Fellow of the Royal Society, the Royal Academy of Engineering, a Fellow and former President of the British Computer Society. He was knighted in 2013 for ‘services to science and engineering’
In this exclusive interview, I spoke to Sir Nigel about open data and artificial intelligence are transforming humanity.
Q: How did you find your passion for computing & Artificial Intelligence?
[Sir Nigel Shadbolt]: I grew up in the age of space exploration, reading Asimov, Le Guin and Clarke, getting lost in the worlds they had created which were littered with aliens, robots and AI. At quite an early age I started thinking about whether there was extra-terrestrial intelligence, what intelligence is, and whether we could make it. I went through school and university, took my first degrees in philosophy and psychology and then went on to do a PhD at the department of artificial intelligence at the University of Edinburgh. It was a field which- at the time- was deeply out of favour, but which was saved by the huge efforts of a few researchers and academics. It was in the 80’s and 90’s where AI started to gain popularity again; and since then- as an area of study- it has gone from strength to strength. For me, I think it was a combination of Star Trek, far too much Asimov and the niggling question of understanding ourselves better through computing technologies.
Q: What is driving the growth of AI and ML technology today?
[Sir Nigel Shadbolt]: The field of Artificial Intelligence was given its name in ’56 by John McCarthy though Turing and others had been thinking of AI for years before that. To understand where we are today, you need to go back to think of the early internet. It had always been doubling in numbers year to year, but that doubling starts from a small number and requires the right conditions for growth. When it comes to AI, there a number of driving forces. Firstly, the underlying growth in the power of computer hardware (we often use density of transistors on a circuit as an approximation of this) shows that computing power per unit of currency doubles every 15 months and has done for 4 decades (Moore’s Law).
This spills over into communication infrastructure, memory and all those areas. We mustn’t forget that we owe a huge amount of our success to electrical engineers who have built the hardware on which our programs run. We are also in an era where data is becoming more available, with the tools and methods we have getting more complex, available and capable. Many of the methods we use in AI today are not dissimilar from those we used 4 decades ago, but they have been supplanted with new tools and extraordinary computing power. The latest generation of machine learning is an intersection of brute force (huge computing power), huge amounts of data and deep insights into how machine learning can happen efficiently, at scale.
Q: As a society, do we understand our relationship with data?
[Sir Nigel Shadbolt]: Practically everything we do has a relationship with data, and we’re generating data constantly which is often held on large corporate or government platforms. People rightly worry about getting the balance right in terms of our data, but we are being overwhelmed by the volume of data – it’s a tsunami – there’s huge amounts of it about, it’s becoming easier to capture, and vanishingly inexpensive to capture through the amount of low cost sensors in our environment. How do we put regulations, checks and balances around this? How do we put adequate controls on the generation and use of data whilst still retaining the huge benefits from this extraordinary ability we now have? The pace of change we have experienced recently has meant that individually, and as a society, we’re still catching-up as we better understand the role that data plays in the decisions that affect our day to day lives.
We have to have a conversation about how data can empower or oppress us- we have the opportunity to re-imagine this, it’s not a deal that’s been done, and it’s not too late to imagine a different way of organising, regulating, collecting, contributing and benefiting. Right now, it’s a bit like the early stages of the universe. Lots of hot plasma gas everywhere, but structure takes time to emerge…
Q: Are technologies such as Ai shaping what it means to be human?
[Sir Nigel Shadbolt]: Roger Hampson and I wrote a book called The Digital Ape, which is all about how we could live in peace with smart machines, but it was literally about the fact that this device here <holds up mobile phone> is the hand-axe of the modern era. Tools and technologies augment us, and amplify our capability, they always have done. Early human species have been found to have shaped and made tools. That tool making shaped our neurology and musculature. We didn’t’ just make technology, it made us. In the modern context, this phenomenon terrifies some people and excites others- but it’s going to happen. We have to understand how humans and their tools and technologies blend at scale – it’s going to be an absolutely fascinating journey.
Q: How do researchers deal with the ethical challenges of AI?
[Sir Nigel Shadbolt]: My fear has never been the machines waking up and deciding to do away with us, but rather that we- in our own bone headed way- deploy systems inappropriately, or without thinking through the unintended consequences that may occur.
There have been successful conversations around bio-ethics for example. Take a look at what happened in the UK with the human fertilization and embryology authority; before the science was resolved they were imagining where artificial insemination and embryology were going, the ethical issues that would be raised, and how we could deal with them.
In a similar way, we’re seeing a debate around the ethics of AI. Key questions are being asked such as when (and when not) to use certain technologies…. How algorithms can be held accountable…. How explicable should algorithms be… what happens when algorithms appear biased against a particular demographic… and also the fundamental question of what happens if AI is available to some parts of the population, and not to others. The nature of AI means that these ethical questions will also bleed into areas such as medical ethics (such as how AI is enabling precision genomics, personalised medicine and therapeutics).
We have to have these conversations, and we have to come up with viable frameworks, otherwise people will continue with dystopian claims about how the future may end-up impoverished with these technologies. I remain a technical optimist… the problem is not artificial intelligence, it’s natural stupidity. Through treaties, regimes and pressure (not always applied brilliantly or perfectly) we have at least stopped ourselves from the worst excesses in chemical, biological and nuclear science.
Q: How can we build societal resilience to AI?
[Sir Nigel Shadbolt]: Societal resilience is not about having a defensive posture, nor about laws and regulations, it’s also about the norms and what is and is not accepted as reasonable behaviour.
People often talk about the ‘Turing test’ and whether or not you can tell if it’s a machine at the end of the phone in a call centre. We’re already well beyond that and we know there are systems that sound and behave like people, and rather than trying to get us (as people) to be able to distinguish between the human and the digital, surely it’s more sensible that we should expect that systems say what they are. Notwithstanding that, there is nothing behind that algorithm at the call centre… nothing equivalent to human intuition, human responsibility, human sensibility….
As you go all the way from individual call centres, you get to the place where you find state actors who are trying to take down civilian cyber infrastructure. With armed context you can see this clearly with tanks and soldiers, but what people can’t see is that there is a 24/7 set offensive and defensive operations happening between state actors, every single day. The digital is invisible, pervasive, ubiquitous, it’s difficult to pin down and make clear rules. That’s where our multilateral organisations need to be… to help us come to a better settlement between our nation states… And that’s at least one place we should start.
Q: What is the difference between data and information?
[Sir Nigel Shadbolt]: When people talk about data, they’re invariably talking about information. If I say to you 41, it’s data. If I tell you 41 is the temperature in centigrade, that’s information. If I tell you 41 is the temperature, in centigrade, of a human patient, that really is information – in fact, it’s in danger of becoming knowledge because of the contextualisation it gives you.
The notion is that you go from data to information when the data has meaning or actionable information. As you look into the whole area of what we mean by data as infrastructure we get into the area of open data which argues that data comes in all-sorts of species and types… some data needs to be kept pretty localized to the individual, state or company as it has high sensitivity or confidentiality… but it’s also obvious that the broader your platform of open data is, the more innovation you get. For example, it would be wrong for anyone to assert intellectual property rights over the periodic table… there are just some facts of the matter that are so useful to society! The question is how broad you get with this… Take the example of a modern state; the vast majority of the data that makes it run isn’t the personal data of citizens, but the data about when trains run, where they are running to, where the schools are, what equipment is in the hospitals… that’s what we mean by data infrastructure and it’s this data which- if it is made open- can be repurposed and reused quickly. People might say, ‘ah, but if you tell people where the telephone masts are, they’ll use that information to blow them up…’ but in reality that’s the trade-off. Those bad actors could easily crowdsource that information, but the risk of that unlikely outcome is outweighed by the value of the information to society, especially if we can balance the risk with confidentiality and accessibility provisions. Information is more powerful if it can be used and shared.
For many of us, we feel this through the laws around privacy. The reasonable expectation of privacy varies through time, and we must be careful to not drive it down to something meaningless. We’re living through a moment now where we have a public health emergency and the collective sharing of sensitive data is in our mutual interest. It doesn’t mean we have to share this level of information forever, but you can clearly make an argument that there are trade-offs between privacy and mutual benefit.
Q: What is the role of trust when we discuss open data?
[Sir Nigel Shadbolt]: There is an idea of institutional trust, and over the years we’ve gained some nuanced distinctions between those institutions that we trust or don’t trust. We might not trust certain public institutions, but we might trust the BBC, our local library, a university or the medical profession. That trust can be moderated or withdrawn if behaviours turn a certain way- if a platform suddenly reveals that they’ve been misusing your data, it will destroy the trust you have with them very quickly, and it’s very hard to get that trust back. We need a new kind of literacy to live in the 21st century – a literacy in computational thinking that extends from data and analytics into how we see a world that’s partly organised by al
And for all of this to happen, fundamentally data as infrastructure yes, but actually a new kind of literacy that we now need to live in the 21st century. It’s a literacy around computational thinking that extends from data, analytics, through to the idea that our world is in some sense organised partly through algorithms.
Q: What are your greatest hopes and fears for AI in society?
[Sir Nigel Shadbolt]: My greatest fear is that we drive ourselves into a sense of hopeless despair with a dystopian trope which says ‘there’s nothing we can do about this… the deal is done and these things have got too complicated for mere mortals to understand…’ this is a dangerous narrative. We educated children from the 19th century onwards in how to do long division, we have a history of making people learn difficult things to advance society.
I also worry that we launch into a narrative around sovereign Ai, that we have to compete with Chinese AI for ‘x’ reason…
We’re on the edge of some extraordinary discoveries around our biological inheritance and health, and AI will help us intervene in a precise and focused way. AI will revolutionise medicine and areas of scientific discovery- not through making decisions for us- but through augmenting our intellect.
The goals of our systems are our goals and it’s we who have a duty to put systems to productive use (medicine, scientific discovery, climate change, sustainability). At the heart of this, we’re humans. We’re located in the space of social interactions even now, thanks to AI shaping the traffic between us.