Don’t expect AI to just fix everything, professor warns

One of the UK’s leading artificial intelligence researchers has criticised what he said was an “extremely dangerous” approach to the technology after comments from a government minister about its role in fixing public services.
Neil Lawrence, professor of machine learning at the University of Cambridge, was responding to remarks at The Times Tech Summit by Feryal Clark, Labour’s new minister for AI and digital government.
Lawrence told the event: “When you hear Feryal Clark saying public services have completely failed, that’s simply not true. Most things work, actually. There’s a lot of stuff we’d like to fix [but] once you get this notion that, ‘that’s all failed, let’s replace it with AI, and some AI general intelligence can fix it’, that’s extremely dangerous.”
Clark had said that the government was seeking to use AI to tackle problems with “broken” public services, including the length of NHS waiting lists. Lawrence noted that Sir Tony Blair, the former prime minister, had made similar suggestions.
The professor said the idea that AI could be imposed on public services to “fix everything” represented an “absurd and reductionist understanding of the complexities of human intelligence and human society”.
He said: “It ignores the complexity of our distributive intelligence — the fact that many of the ways in which we hold each other to account are based on the fact that we have shared human vulnerabilities,”.
Lawrence, who spent a period as a field engineer on oil rigs in the North Sea, is former director of machine-learning at Amazon, where he worked on deploying solutions for Prime Air, Alexa and the Amazon supply chain.
The professor also criticised “Wall Street and Silicon Valley’s perception that an artificial general intelligence, the term for AI matching human cognitive capabilities across a wide range of cognitive tasks, could one day replace human intelligence.
“It’s threatening everyone’s futures not with a technical existential risk but with a socio-technical existential risk of dumb decisions by people who should be thinking about things in a more nuanced way,” he said.
Lawrence said large language model-based AI, such as ChatGPT, had the potential to be “utterly transformational” because it allowed ordinary people to converse with machines much more comprehensively, whereas previously computer experts and programmers had been like the “scribes of Mesopotamia”.
“It comes with many, many challenges that undermine existing societal structures,” he said.
He added: “The horror of the conversation in this country in particular over the last two years is [that] the people who are most confident about how the future is going to work out have dominated the conversation. The only thing we can confidently say is confident projections of the future won’t pan out.
“We need people who ‘know they don’t know’ around the table, the ones who understand it requires a conversation across society and that we need to get together and listen to all the voices before we make the critical decisions about what our children’s … future is going to look like.”
Lawrence also criticised Rishi Sunak, the former prime minister, for failing to challenge a prediction at a UK-hosted technology summit last year from Elon Musk, the tech billionaire, that AI would put an end to all work.
“This is a total absurdity,” Lawrence said. “That … it was just accepted shows what a difficult situation we are in, in terms of public understanding of this technology. I found it horrific, but it was the centre of our summit.”
Marc Warner, founder and chief executive of Faculty, one of the UK’s most promising AI companies, also said that forecasts of AI dominating GDP should be treated with caution. “Just don’t believe predictions like that,” he said. “They’re utterly clueless. They’re just selling you something.”
Lawrence said that the rise of AI “gives us a chance to stand and look back at our intelligence and start marvelling at its extraordinary capabilities”.
He added: “In the same way that the famous ‘blue marble’ photo from 1972 gave everyone a sense of how fragile Earth is in the vastness of space, comparing us to the vast capabilities of a machine which can access 300 million times more information than us per second — the difference between walking space and light speed — should give us more respect for who we are and what we are.”