Skip to main content

Neil Lawrence - Atomic Human interview

Neil Lawrence is the DeepMind Professor of Machine Learning at the University of Cambridge where he leads the university-wide initiative on AI, and a Senior AI Fellow at the Alan Turing Institute. Previously he was Director of Machine Learning at Amazon, deploying solutions for Alexa, Prime Air and the Amazon supply chain. Co-host of the Talking Machines podcast, he's written a series for The Guardian and appeared regularly on other media. Known for his policy and societal work with the UK's AI Council, the Centre for Data Ethics and Innovation, and the OECD's Global Partnership on AI, his research focuses on improving data governance, accelerating scientific discovery, and how humans can take back control of large AI systems. His latest title is The Atomic Human.

What would you like your book to achieve?

I wanted it to speak to individuals from different backgrounds in a way that didn’t preach or tell, but told stories in a way the reader could relate to. I hope the book allows people to bring their own perspectives and combine them with a more nuanced understanding of intelligence. In a way the book is an edited version of me, one that’s more coherent and tempers my more polemic instincts. I’d like to be in a world where everyone feels confident (and informed) when having a say in how we use this technology. I hope the reader leaves the book feeling more empowered to have that say.

You emphasise how humans have to cope with very limited bandwidth compared with AIs - is there anything we can do make the best of our limitations?

I think how we deal with this varies according to our personalities. For me, it makes me realise I have to better understand someone’s culture before I understand their words. I feel arguments arise when other people’s words are combined with our premises, but to understand people you have to understand their premises. I think I’ve got better at that as I’ve got older.

I take your point that we shouldn't allow AIs to make life-changing decisions without human supervision, but can we realistically impose this?

Likely not fully, although existing legistlation already attempts to do this. In GDPR consequential decisions should be taken by humans or be rendered explainable. And I think that’s trying to get to this point. Even if we can’t enforce this precisely, I’m worried that the current debate around AI doesn’t seem to have this principle at its heart. Referring back to your first question, having this principle at the heart would be a change I hope the book can help enact. 

Given concerns about AI decision making, should autonomous vehicles be allowed on the roads (particularly narrow, windy European roads)?

I think I worry more about the busyness of the roads and the  shared usage between pedestrians, cyclists, etc that you get e.g. in city centres. Motorways would seem ideal for deployment already, but they are much more tightly regulated than other roads (no cyclists or pedestrians). This reflects what we’ve seen in previous waves of automation, humans have to adapt to computers to accommodate them. What I call the 'great AI fallacy' is the idea that this trend would be different for AI.

A colleague recently visited the US and said that one of the major challenges the Wayve vehicle he rode in had was in dropping off. It struggled to find a legal place to park near the destination. I suspect that most taxi drivers would just have parked illegally. That’s something we seem to tolerate from humans but likely won’t tolerate from machines. I think that relates to the core theme of the book, which is really about the shared jeopardy of human society. Since machines don’t share in the jeopardy I think we will never indulge them in the same way we do humans (or other animals). I think that has significant implications not just for autonomous vehicles but in a range of other societal domains where human decision making comes into contact with machines.

What’s exciting you at the moment (in AI or otherwise)?

The potential that AI could allow the computer to become a tool, not just of software engineers, but of regular people. In a similar manner to the printing press enabled writing to become accessible to regular people.

Comments

Popular posts from this blog

Six Maths Essentials - Terence Tao *****

This is a book in the 'compact collection of science/maths essays' category beloved of publishers because they can charge hardback prices for fewer than 80 pages of text - but don't let that put you off, because it's excellent. Terence Tao introduces six areas of mathematics: numbers, algebra, geometry, probability, analysis and dynamics. Each of these is descriptive rather than practical (there are only a couple of school-level equations in sight), done in a way that gets to the heart of 'what's the point of this subject' with precision and a good level of storytelling. It's a shame in a way that analysis and dynamics form the final third of the book (apart from a short epilogue), because they don't work as well as the other sections. For the first four, practically everyone will have a vague idea of what they are and what they do - but Tao is then able to open our eyes to their wonders and power. Because Tao has to stay away from any detail of what...

Bradley Efron - Five Way Interview

Bradley Efron is Max Stein Professor emeritus of Statistics and of Biomedical Data Science at Stanford University. He is the inventor of the bootstrap method for assessing statistical accuracy and has published extensively on the theory and practice of data analysis. He has served as president of the American Statistical Association and the Institute of Mathematical Statistics.  A MacArthur fellow, he is a member of the National Academy of Sciences and received the National Medal of Science in 2005. His remarkable contributions to his field were recognized again with the 2018 International Prize in Statistics, which noted his influence on 'statistical thinking in every way, including new pathways in how we communicate and teach statistics to all.' His latest book is To Think Like a Statistician . Why mathematics?   Abstract mathematics and I had an amicable divorce following my Caltech graduation. (I really wasn’t very good at it.) Statistics is a branch of applied mathematics...

Time Now - Colin Gillespie ***

A quote on the back cover of this book describes it as monumental - it certainly feels big enough to use as part of a structure with 664 pages plus a further 112 pages of end notes. Colin Gillespie's aim is to 'bring together centuries of ancient thought and modern insights to unearth what is real and what is not'. That certainly sounds like it would need a lot of pages, if it is to be achieved at all. A starting point is to establish what 'real' means. Gillespie gives us the guidance that 'to accept something as truly real, it must be an accurately described objective aspect of the universe.' That may seem obvious, but as he points out 'little in philosophy or physics even aims to pass this test.' Gillespie builds a new conception of physics based entirely on space. As he puts it 'space isn't something; it is literally everything.' In this worldview, the universe began with a single 'space quantum' - a Planck-sized six-dimensiona...