AI scientist-author Toby Walsh: 'We are increasingly outsourcing some of our thinking to machines'

Sydney-based Toby Walsh, one of the world's leading scientists in Artificial Intelligence, and author of '2062: The World That AI Made', was a prominent speaker at last month's Jaipur Literature Festival.

In his book, 2062: The World That AI Made, Toby Walsh, one of the world's leading scientists in Artificial Intelligence (AI), predicts the transformation of homo sapiens into homo digitalis, a state in which human thought will be replaced by digital thought. Born just outside London and educated at Cambridge University, Walsh lives in Sydney, Australia, where he is a professor of AI at the University of New South Wales.

'2062: The World That AI Made' (2023, La Trobe University Press) by Toby Walsh.

In such books as It's Alive! Artificial Intelligence From the Logic Piano to Killer Robots (2017), Machines That Think: The Future of Artificial Intelligence (2018) and Machines Behaving Badly: The Morality of AI (2022), he has sought limits on use of machines and supported algorithms for positive use. A prominent speaker at the Jaipur Literature Festival last month, in an exclusive interview to Moneycontrol, Walsh talked about morality and ethics in the development of technology and the danger of handing over decision-making to machines. Edited excerpts:

What are the aspects of development in AI that make experts like you worried?

There are a couple of areas which concern me. Artificial Intelligence is unfortunately a dual-use technology, there are positive uses, and the very same algorithms have negative uses. One that concerns me the most is AI warfare. It is a fine example of positive-negative uses. The algorithms that are going to go into the autonomous cars, they are going to identify tracks and destroy commands on the ground. The same algorithms that are going to go into autonomous cars are going to identify pedestrian traffic and avoid them, saving millions of lives.

What should ordinary people be worried about AI?

There are many positive uses that are going to improve our lives. We should be more concerned about how AI is going to transform warfare, how we already see it transforming democratic divide, we see the AI algorithms running social media distorting our conversation, and algorithms to generate fake news and fake tweets that are overwhelming social media channels.

What are the ethics shaping the role of technology like AI?

I have written a whole book, Machines Behaving Badly: The Morality of AI, looking at these questions. I think one of the misconceptions that people have is that some sense of new ethics is required to deal with the new challenges that AI poses. Clearly there are new challenges, but most of these challenges are about speed, scale and cost. Are there things that we can do at greater speed, greater scale or lesser cost? The ethical questions we should be asking are the same questions we have asked about any other technology. Questions about whether it is done in a transparent way, is it done in a way that is fair and equitable. Those are the things we are worried about with all technologies. There is nothing new about it. There is one small exception, one thing that is new and was not true of any previous technologies. which is the idea of autonomy. Machines have some independence to make decisions on their own, to go off an act without much or no oversight. Which is why the most difficult ethical challenges AI pose are questions about autonomy.

Who are the people asking these questions? Is there a regional or global forum or campaign or activism around ethics and AI?

The good news is that there is a growing academic subfield of AI looking at how to view AI in a responsible manner. Many researchers are turning their minds to these questions. One of the challenges of this field though is that it really requires people from all different disciplines. It requires not just technologists like myself to understand how the technology works or how the technology breaks, but it also requires philosophers who understand fundamental questions about philosophy. It requires anthropologists to understand how humans respond in any situations, it requires psychologists who understand the human psyche. It requires economists to understand economic imperative questions. It requires political scientists to understand how it could affect changes in regulations to do with these challenges. It requires people from the humanities who frame their own perspective about ultimately what sort of society we want to build. And the novelists and poets who have actually asked these questions very well in the past.

Are humans becoming machines themselves?

We are increasingly outsourcing some of our thinking to machines. Who remembers telephone numbers anymore? No one. We like machines to do that. There are other things to be outsourced where we don't give up important parts of ourselves by doing that. There is an important role to be played by machines doing the dull and the repetitive. Equally there are places where we may undermine our own humanity if we hand those decisions over to machines.

Are anti-tech communities alive today? Do you meet people who stay away from the digital world?

I do predict in my book that there will be these punk-like reactionaries who enjoy old-fashioned pursuits and pleasures. Indeed in many respects we are already doing that to ourselves. There is a greater appreciation for handmade bread and artisan cheese. Those are old-fashioned pursuits where we make things with our own hands. Those are things that always gave us great reward. I always point out to people that one of the newest jobs on the planet is being an Uber driver. That is soon going to be automated. In San Francisco today they are already trialling autonomous taxis 24x7. There aren't going to be Uber drivers around for very long. One of the oldest jobs on the planet is being a carpenter, it is also one of the longest-lived jobs. We are always going to appreciate hand carved wood, we are always going to appreciate the carpenter telling us how she or he found a piece of wood and how they turned that into a wooden bowl.

Why is 2062 as the year homo sapiens become 'homo digitalis' as you predict in your book 2062: The World That AI Made?

It is 2062 because I surveyed 300 of my colleagues who are experts around the world in AI when machines would be as capable as humans and the average answer they said was 2062. 'Homo digitalis' is the digital selves that we will have. I am not sure we are necessarily going to be physically one. With the machines increasingly we will be spending our lives in digital spaces, interacting with other digital beings, avatars of our friends and our colleagues and our lives are going to stop being so physical and be much more digital. And we are going to become one in many respects with our digital selves.

Will robots become conscious?

Who knows? It is one of the biggest scientific questions. They are not conscious today as far as we can tell. It is possible, there is not any law of physics we know of today that will be violated if they become conscious in the future. I think it is one of the great scientific mysteries of the next century that AI will help us answer. Is consciousness something that is distinct to biology or something that we can recreate in silicon? It is going to be a fascinating century to find out the answer to that question.

What kind of questions on AI and ethics that your students ask you?

That is an interesting question. I think mostly they ask about how we can ensure these technologies will be used for good. That is a challenging question. The best answer, not the perfect answer, I have for this question is what we could work on the good use of technology, to help solve many of the 17 United Nations sustainable development goals, SDGs, with the help of AI, at least then we can ensure those uses of AI offer positive benefits to all of humanity.

How do you see governments and technology companies performing their roles in the future?

Politicians around the world are waking up to the idea that you can and you should regulate the tech space. Ten years ago there was a lot of pushback to this idea, it was said you couldn't because the tech space was something different, it didn't have physical presence, these tech companies crossed international borders and were as powerful like nation-states in their own right and you shouldn't because it would stifle innovation. And we had a lot of good innovation in the tech space in the last 2,000 years. Now there is a growing realisation that you can and it is entirely possible to regulate the tech space.