The AI apocalypse is coming! Personally, I welcome our new robot overlords

What’s the coolest apocalypse scenario? I’m quite partial to zombies myself, but that’s probably because I was in university during the great zombie boom of the late 2000s, where it was impossible to engage with a piece of media that didn’t have some kind of “zombie mode”. Spending five years asking strangers in the smoking areas of pubs, “hey, how do you think you’d survive a zombie apocalypse?” is basically millennial navy seals training.

This slow collapse of civilisation that we’re living through right now is probably the least interesting version of Armageddon, so I’m pleased to report that it will soon be overtaken by my second-favourite doomsday scenario: the murder robots are coming.

At least, that seems to be the case according to Geoffrey Hinton, a man so famous for his work in artificial intelligence that he’s known as the “godfather of AI”. Hinton recently left his job at Google, where he helped develop the underlying technology that that paved the way for current AI systems such as ChatGPT. He has quit his job, ostensibly, to warn humanity about the dangers of AI.

Hinton believes that in the short term, AI will be used to create a deluge of fake images, videos and text, which will leave the general public unable to distinguish between fact and fiction. Longer term, however, could be worse, as Hinton suggests the technology could leader to a future where robots surpass human intelligence, and ultimately decide to kill us off.

Finally, some good news.

I can’t speak to the validity of Hinton’s claims to the latter, but we should absolutely concerned about the former. We were already struggling with “fake news” (ie shameless propaganda) before there were systems in place that could essentially automate it. Imagine how bad it’s going to be when it’s not just people tweeting fake information or dodgy news sites writing heavily biased or fabricated news articles, but fake pictures and videos that are indistinguishable from the real thing.

Remember that picture that was going around a few weeks ago of the Pope in a big puffy rapper jacket? I thought that was real for two full days. Now, I’m very gullible, but at this rate this technology will be sophisticated enough to trick non-gullible people in no time at all.

But what about his claims about a Terminator-style future where robots hunt us for sport? Honestly, as crazy as they might sound, nothing would really surprise me anymore. The rule for the past 10 years seems to have been: think of the worst, least likely thing that could possibly happen. I guarantee you, something even worse will end up happening.

The funniest part of Hinton’s statement is definitely the bit where he goes: “The idea that this stuff could actually get smarter than people – a few people believed that. But most people thought it was way off. And I thought it was way off. I thought it was 30 to 50 years or even longer away.”

Wait, you thought the AI doomsday scenario was only 30 years away, and you ploughed ahead with it anyway? 30 years isn’t “way off”! That’s no time! Jurassic Park came out 30 years ago!

Hinton isn’t even the only AI expert that believes the technology they helped create will eventually destroy us. In a recent podcast appearance Paul Christiano, who ran the language model alignment team at OpenAI, said he takes “quite seriously” the idea that artificial intelligence will enslave and ultimately destroy humanity.

So what’s the solution? Do we outlaw AI? Do we pass a bunch of laws to try and regulate it? That doesn’t seem likely, since our policymakers still haven’t figured out a way to stop neo-Nazis from recruiting on Twitter, or organising hate campaigns against individuals they disagree with. We’ll be hiding from Skynet while the government is still trying to figure out whether or not unplugging the AssassinBot 3000 is a violation of its right to free speech.

What I do know if this: whether or not we get a viscerally satisfying apocalypse out of this, with cyber tanks and sentient laser turrets, the much greater threat in the short term is to the labour market. This is something that Hinton touches on, but it’s probably the thing that is likely to affect you, the reader, most.

It’s certainly going to affect me. I don’t think I would have bothered putting all my energy into a writing career if I’d known that people put so little value into the quality of the texts they consume that they’ll happily outsource that work to an AI language model.

You might think you’re safe, but if you work in a field that can be handed over to an algorithm, rest assured that it eventually will be. AI doesn’t take smoke breaks, it doesn’t sexually harass its co-workers, and the only compensation it requires is your sweet, sweet information.

For now, the issue whether or not the robots are coming to harvest your weak, human, fleshy organs is irrelevant. They’re definitely coming for your jobs.