I’ve been thinking a great deal recently about the actual risks of artificial intelligence, not the amount of jobs that it’s going to take, which is the sort of thing that occupies most people’s concerns. Fears are of course everywhere, but there is something darker and more urgent afoot. When I heard Demis Hassabis, who is a Nobel Prize laureate and the head of Google DeepMind, I was a bit worried but I had a strange sense of clarity. AI putting people out of work doesn’t keep him up at night. He’s worried rather about how people can misuse this powerful tool of ours, how unprepared we appear to be.
If I’m honest, I am with him on this. The notion that people can now clone a voice or manufacture fake videos so realistic they are able to dupe families or entire communities is absolutely horrifying.
These are not distant threats, they are already happening. Artificial intelligence is being used to impersonate people you may not even know, via digitally altered videos that can spread misinformation on social media. And websites can now create fake stories that look real while being tailor made for particular companies or candidates. It’s like the truth can be harder to find.
What worries me is that AI is becoming more advanced more quickly than rules and protections are being devised to deal with it. Hassabis spoke about the necessity of global accords to prevent AI from getting into the wrong hands and I do think we need that sort of thing. Sometimes it’s the only way to get the memory or answer that we want to remember or have and if you don’t set those limits now, you may regret it when it’s beyond your control.
I’m not against AI. I think it can help us do wonderful things. But we need to be smart, careful and a bit humble. But just because we’re capable of building something doesn’t mean we’re prepared for what it brings.