Meet the Academic building trust in Artificial Intelligence

As you might expect, a Professor of Artificial Intelligence will have no hesitation in describing themselves as an ‘AI optimist’ and would likely remember being inspired by movies and TV shows from their childhood depicting exciting futures featuring self-driving vehicles.

“These things promised to make our lives better and more enjoyable, and I was inspired by the technology; that’s why I got into this field,” remarks Gopal Ramchurn, Professor of AI at the University of Southampton. “I see lots of opportunities for AI to make things more efficient, improve the resilience of assets and reduce costs.

“But what worries me is the way the technology has evolved over the last decade,” he continues. “It has given power to certain companies to exploit, rather than serve, society. We risk taking away jobs and creating new technology that adds to cyber security concerns.

“AI technologies take time to get right,” he adds, “but if we rush into them to make a lot of money very quickly rather than thinking through how we deploy them, we will end up serving the richest people first. We need to take a long-term view of AI, rather than just letting it be in the hands of Silicon Valley.”

One major challenge that needs addressing, he adds, is ensuring that people have trust in such systems.

Making effective connections

Gopal has been involved in AI for over 20 years and currently directs large programmes for UK Research & Innovation such as developing autonomous vehicle systems to be safe and trustworthy. This summer he was appointed Chief Executive of the £31 million ‘Responsible AI UK’ programme.

He was introduced to Connected Places Catapult four years ago by a former colleague from BAE Systems, Henry Tse – the Catapult’s Director of New Mobility Technologies. Gopal was asked to build a training module about autonomous vehicles to help managers learn more about how AI is used in the sector.

In recent years, Gopal has created two AI related start-up companies: one focused on reducing carbon emissions to mitigate the effects of climate change (Empati); the other for sports teams to run AI simulations to help make better decisions, such as with player selection (Sentient Sports).

He has also worked with defence and energy clients; developing digital twins for energy management and disaster response. Over the last decade, his focus has been on the human-centred design of AI, “thinking about how machines can best interface with people”.

“When we’re deploying any technology that takes away control or decision making from humans – as with autonomous vehicles – security issues emerge, as do people’s safety concerns,” he explains.

Public trust is an issue that cannot be taken lightly, he adds. “Autonomous vehicles have been around a long time, such as on the Docklands Light Railway, and people generally feel very safe in those trains. But they didn’t always at first. Over time, you can help build a greater feeling of safety.”

Gopal adds that people’s relationship with aeroplanes went through a similar process. “Planes are highly automated and use AI technology; we trust them to land us safely.” On the other hand, autonomous cars are at the starting point in their journey to acceptance. “It may take a while for parents to feel that such vehicles are safe enough to let their children climb aboard.”

Gopal is currently exploring whether autonomous cars should be allowed to drive everywhere, or be restricted to certain locations; and is exploring how parcel delivery drones using AI systems could be made to feel more acceptable to the public.

Early years

Gopal was born in Mauritius and attended secondary school during the formative years of the Internet. His father encouraged his son into computing, and took out a loan to buy him a machine “which cost him a fortune”. Besides using it for schoolwork and playing video games, the family used the computer to communicate with Gopal’s brother in Australia.

“We used an early chat system, and I was keen to try and get the latest software to communicate better and more cheaply,” Gopal explains.

Disaster response and transport

Gopal later embarked on a research project to help rescue teams function more effectively in disaster response environments; co-ordinating the activities of emergency services. A simulation was developed called RoboCup Rescue, using drones to map and create a digital twin of any city in the world, allowing events to be simulated such as fires or blocked roads.

“We could simulate emergency service vehicles driving across a city at different speeds, to help optimise the routing of services,” he explains.

Gopal points out that drones can be very useful in delivering medicines or seeking out casualties in disaster areas, but recognises that their general use in commercial applications can present a public acceptability challenge. So he has embarked on a research project looking at the issue of trustworthiness in deploying autonomous craft.

He has also developed transport-based applications in AI including ride sharing algorithms to reduce the cost of travel and minimise passengers’ carbon footprints. Thought has also been given to ways of pairing people together in vehicles who are likely to be comfortable sitting with strangers; a concept taken forward by the Catalan government in Barcelona.

Moving forward thoughtfully

Gopal reiterates his confidence that AI will deliver many improvements to society, but is mindful of the costs and trade-offs of sharing data and personal privacy. “The question is whether we’re ready as individuals,” he says.

“Right now innovators are moving fast, but there needs to be more forward thinking around the deployment of AI technologies, to make sure they don’t just benefit the few in the short term. Innovation is great, but innovation at the cost of society is not.”

To be alerted about future opportunities at Connected Places Catapult, join our Academic Network.