On November 1 and 2, the British Government will hold a summit on the security of artificial intelligence (AI). In this context, Demis Hassabis, head of Deep Mind, the AI company owned by Alphabet (Google), has made some loud statements about how concerned we should be about the risks of AI.
Specifically, Hassabis has assured the British newspaper The Guardian that "we must take the risks of AI as seriously as other important global challenges, such as climate change," he said. “It took too long for the international community to coordinate an effective global response to the climate crisis, and now we are living with the consequences. “We can’t afford the same delay with AI.”
Hassabis Hassabis said the world must act immediately to address the dangers of artificial intelligence, such as aiding the creation of biological weapons and the threat to existence posed by superintelligent systems. Deep Mind's chief executive said AI has the potential to "be one of the most important and beneficial technologies ever invented", but an oversight regime is nonetheless needed.
And Hassabis has a proposal for the model on which this AI monitoring system could be based. "I think we need to start with something like the Intergovernmental Panel on Climate Change (IPCC), which is based on scientific agreement and research, and then move forward from there." The next steps would be – according to Hassabis – “an equivalent of a Cern for AI safety, which investigates this aspect, but at an international level. And then, maybe one day, there will be some kind of equivalent of the IAEA – the International Atomic Energy Agency – that actually audits this kind of thing.”
Hassabis' idea is not the first time it has been raised. Last week, Eric Schmidt, former CEO of Google, and Mustafa Suleyman, co-founder of DeepMind, already called for the creation of an IPCC-style panel on AI. Hassabis was one of the signatories in May of an open letter warning that the threat of AI should be considered a societal-scale risk on par with pandemics and nuclear war. “We should start thinking and researching now. I mean yesterday, actually,” he said. “That's why I signed, and many people signed, that letter. "It's because we wanted to highlight that it is a reasonable topic to discuss."
The CEO of Deep Mind was especially concerned about the possible development of artificial general intelligence (AGI), systems with human or higher-than-human levels of intelligence that could evade human control, although at the same time he recognized that they could represent a great advance in fields such as medicine and science. Hassabis also believes that AGI is still a long way off from becoming a reality, but "we can see that the road is taking us there, so we should discuss it now."
