Michael Osborne: “Perhaps driving a car will be illegal by 2050.”
Where is automated driving headed? In the third part of our interview video series, machine learning expert Michael Osborne explains why we need to prevent driverless cars from becoming black boxes.
Let’s look at the big picture! At a recent developer conference of technology company Continental in Copenhagen, 2025AD had the chance to talk to several renowned experts from various fields. We asked them what impact automated driving will have on our society, how public trust in driverless cars can be created and whether humans will still drive cars in 2050. In the first part, bestselling author Martin Ford explained why he thinks humans will continue to drive in the future. In the second part, Peter Cochrane, technology consultant and former CTO of British Telecom, explained why sensors represent the biggest challenge for driverless cars . In the third part, we interview artificial intelligence Oxford expert Michael Osborne. Stay tuned for an interview with Club of Rome Co-President Prof. Ernst Ulrich von Weizsäcker.
2025AD: What will be the implications of AI for our society in the year 2030?
Michael Osborne: So, the implications of AI on our society in the year of 2030 are really quite profound and contain both enormous opportunities alongside potentially equally huge risks. So, the opportunities involve improving productivity across a wide range of different sectors: improving health outcomes, tackling environmental challenges… The risks, however, include things like the risk to labor that automation enabled by machine learning might pose, alongside an increasing concern about the impact of AI on privacy and what the data that we share with the tech firms, for instance, means for our roles as citizens. On the positive side, AI means automating away some of the tasks that we actually find least pleasant. Put another way, the things that will remain after AI has truly established itself in our workforce, are things like creativity and social intelligence: the kind of things we find most pleasant, that we do in our spare time. In a way, AI will actually make work more pleasant.
2025AD: What impact will automated driving have on our society?
Michael Osborne: Autonomous driving, just as with AI, will have transformational impact across the whole range of our society – changing everything from the way cities are designed to the way that data is gathered on transport and downstream consequences for mobility services.
2025AD: What does it take to create public trust in driverless cars?
Michael Osborne: Creating trust in AI systems is something that is very much in the thoughts of machine learning practitioners today. One of the challenges here is ensuring that the decisions that are made by machine learning systems are interpretable to some degree. The state-of-the-art machine learning is not entirely unfairly characterized as black boxes that are fundamentally not able to give good explanations for why they make the decisions that they do. So, I think that’s the technical challenge that we as machine learning developers must assume - which is to give some of that degree of transparency to the algorithms we develop. That will go a long way to giving downstream users trust in the decisions made by these algorithms.
2025AD: What is the biggest challenge on the way towards driverless mobility?
Michael Osborne: I think the biggest challenges to autonomous driving are actually more regulatory than technical in some ways. Even today we have the technology to deliver autonomous driving within certain well-defined areas, where the rules exist to do so. Autonomous vehicles, for instance, are quite operational today in places like airports or in factories or warehouses. The question is: where do we want to allow these vehicles to be? That means putting appropriate safeguards in place to ensure that people don’t get in the way, that there aren’t unnecessary risks.
2025AD: In 2050, will humans still drive cars themselves?
Michael Osborne: So, in 2050, assuming that autonomous vehicles truly have reached their potential, perhaps it will be illegal for humans to drive. Obviously, humans are far from perfect drivers: we’re the major course of accidents. There might actually be a case for regulation that insists that if you do want to drive you do it only within special parks for enjoyment rather than actually for transport.
2025AD: How will you spend your time during the ride once cars are self-driving?
Michael Osborne: I’m afraid to say my life would not be so different from my current life, because my major mode of transportation is cycling or walking and if I do get into a vehicle, it’s normally driven by someone else. What I do then is largely check twitter! So this is exciting technology but it might not change my life so much.
About our expert:
Michael A Osborne (DPhil Oxon) works to develop machine intelligence in sympathy with societal needs. His work in Machine Learning has been successfully applied in diverse contexts, from aiding the detection of planets in distant solar systems to enabling self-driving cars to determine when their maps may have changed due to roadworks. Dr Osborne also has deep interests in the broader societal consequences of machine learning and robotics. His work on the significance of machine learning and robotics to future labour markets has resulted in both sustained coverage in most major media venues (e.g. his being interviewed on BBC Newsnight, a cover feature in the Economist) and policy impact (including presenting oral evidence to the House of Commons Science and Technology Committee). Dr Osborne is the Dyson Associate Professor in Machine Learning, a co-director of the Oxford Martin programme on Technology and Employment, an Official Fellow of Exeter College, and a co-director of the EPSRC Centre for Doctoral Training in Autonomous Intelligent Machines and Systems, all at the University of Oxford.