What does artificial intelligence look like in the 2020s?

The Face guide to the 2020s: Stuart Russell, Professor of computer science at the University of California, weighs in on “digital personal assistants” and the tragic prospect of “AI-powered genocide”.

Hi, and welcome to the future: a toasty-warm, carbon-neutral, plastic-free place where your face has replaced your passport and your car drives itself. Weed is legal, alcohol is hangover-free, weekends last three days and we robots do your admin. We can dream.

Ten long years ago, The Face compiled a set of predictions for the coming decade from a star chamber of hotshot experts. That flesh-and-blood editorial team has long since disbanded but their legacy remains. From now until the first day of the new decade we’re sharing some prognostications (as seen in The Face Volume 4 Issue 002) on love, sex, space, AI, cannabis, mental health and plastic surgery (and more) for the years ahead.

Stuart Russell (Professor of computer science at the University of California, Berkeley and author of Human Compatible)

There are some good developments that are likely to come down the line. By the end of the 2020s our digital personal assistants will actually become useful, as AI systems get a proper handle on language – meaning they’ll understand what we’re saying and give us truly meaningful information back. We’ll also be ferried around in self-driving cars which we’ve probably bought from US company Waymo. It is likely to bring the first systems to market. This will be good for road safety and for disabled users, but it will also mean a lot of job losses for people who drive for a living. And those losses are likely to extend to many low-skilled sectors, like factory work and farming, all of which are likely to become more AI-powered within the next 10 years.

By the end of the decade we may also begin to see real interface between the brain and technology. We might figure out how to give people additional memory by wiring chips to some part of the brain. Initially, this technology will be developed for people who have dementia, but then it will likely become available for consumers who just want to increase their memory. By 2030 we may also have figured out how to connect brains to one another; that’s not as far-fetched as it seems. These things can happen despite the fact that we don’t fully understand the brain because it’s such an adaptive organ. For instance, we already have cochlear implants, chips which allow deaf people to hear. At first, the sounds seem alien but within a few weeks the brain adapts.

But I think the first big AI news item of the 2020s is likely to be a tragic one. It will be the first ever use of autonomous weapons in warfare – Turkey is reportedly planning to use them in Syria. 

These weapons can move by themselves and, under the control of AI, can locate, select and attack targets without any human supervision. They work a little like the civilian drones you can buy, where you show it your face and then it follows you around, taking little selfie videos. In the case of weapons, you would define some targeting criterion and a geographical region – so, go into this sector and kill anyone who looks like a male Kurd’ – and it would follow those directives. 

After that, we might start to see these weapons being miniaturised and sold in large numbers. Which, as the 2020s progress, could well lead to the first AI-powered genocide. Perhaps after the first genocide nations will agree to sign a treaty to ban such weapons. I hope it never gets to that point, but the potential is there.”


Loading...
00:00 / 00:00