In an interview with the “Times”, Matt Clifford insisted on the importance of immediately regulating the research done around AI, comparing the moment when we are at the beginning of the Covid epidemic, when in January 2020, we considered the coronavirus very far from us.
Faced with the dazzling and exponential progress of artificial intelligence software, we no longer count these warnings, each week brings a new one, each time from an AI godfather, or from a disillusioned pioneer in the sector. On Tuesday, it was Matthew Clifford, the artificial intelligence adviser of 10 Downing Street, therefore British Prime Minister Rishi Sunak, who made the front page of the Times which headlined with this terrible sentence: ” AI, we have two years to save the world”.
>> South Korea: several cities use artificial intelligence to talk on the phone to isolated elderly people
The sentence is alarming, and immediately, it was taken up everywhere, the daily The Independent even summed up by captioning that according to the Prime Minister’s adviser, ” artificial intelligence could kill many humans within two years”. But in reality, what Matt Clifford says is much more interesting than this catastrophic distortion of his words. Because the journalist Timesin this filmed interview, asks him the question that intrigues us all: of course, we talk about the “dangers” of AI, but concretely, what are these dangers?
“What do we do with what we know?”
To which Clifford responds that “the range of risks is very wide, but it can be quite frightening, for example an AI could be used to disseminate a biological weapon or launch a large computer attack“. But he adds that in the face of these probabilities, the reality is that no one, no expert knows what will happen or when, and that instead of trying to make predictions, we must frame , regulate.
1/3 Quick thread on AI risk: there’s a story in The Times today headlined “Two years to save the world, says AI adviser” based on comments I made in an interview. The headline doesn’t reflect my view. Here’s what I said: pic.twitter.com/5JwMfdn8YQ
— Matt Clifford (@matthewclifford) June 6, 2023
The journalist asks him all the same “How much time do we have left to act?”and Clifford replies: “If we don’t look at all that now, well, in two years we’re going to end up with really powerful software.” End of the quote. No threat of extinction of humanity, nor of a collapsed world, therefore. No, in fact, listening to Matt Clifford, we understand that what should alarm us are humans, those who do the AI, but also those who watch, who scare themselves, then end up saying “What’s the point of regulating if China doesn’t do it, if Russia doesn’t do it?“
A bit like for global warming, we know, but we wait. Clifford compares the challenge we face with AI to the Covid epidemic. He says that we have the same attitude today as in January 2020, when there were some contaminations, but we did not feel concerned. For the climate, the Covid, artificial intelligence, the question is “What do we do with what we know?