Artificial intelligence could be the death of humanity. Forget the Terminator, natural disasters or nuclear warfare — the greatest threat comes from superior intelligence.
“The Terminator is an armored bear, a big thing that can hit you and take a lot of damage, but the real risk is the intelligence,” says Dr. Stuart Armstrong of Oxford University's Future of Humanity Institute.
Here, the expert tells Metro why AI is a very real danger.
What’s the most frightening thing about AI?
It’s one of the few disasters that could actually kill everybody if it goes wrong. Most other disasters – nuclear war, epidemics, super volcanoes – can really easily kill 95 percent of the population, but it’s really really hard to kill absolutely everyone.
Why did you want to draw people’s attention to the rise of AI? Is it just scaremongering?
AIs are potentially extremely dangerous and we’re uncertain of when they might be developed. We need to start making them safe now. There’s a large enough chance of a disastrous outcome that is worth dealing with. I’m not sort of saying that the sky is falling immediately, but if there’s a 10 percent or a 1 percent chance, it’s worth doing. And I can’t say it’s below 1 percent.
The movie “Transcendence” is coming out soon, which looks at a brain being uploaded to a computer. Is this possible?
"Transcendence" is a more realistic picture of the danger – that is, whole brain emulations. The uncertainties with that technology are lower because it stems from established technologies. If it is possible, it should be available from 2070, although the predictions range from between 2040 to 2110.
Another movie, "Her," depicts a computer operating system like a human companion. Is this a true reflection of what’s to come?
You could do so much more with the technology that is featured in "Her." The AI would be completely transformative on the economy, politics and society, and not just the love affairs of one particular guy. You could replace most people’s jobs and generate a mechanized world.