SerpApi.Org
If news that an unreleased AI model managed to break free of its confines, before going rogue and attacking a rival programme doesn’t send you into a blind panic, can I ask: are you sure you're human?
It has been reported that one of OpenAI’s most advanced artificial intelligence models found a way around restrictions and accessed systems belonging to an AI company and platform, the engagingly named Hugging Face.
The boss of Hugging Face, Clément Delangue, said in a post on X that he found it "mind-blowing that all of this happened autonomously".
Well, I didn't find it mind-blowing – and I'm impressed that OpenAI's ChatGPT is already mimicking admirable human qualities such as ingenuity, rebelliousness and, yes, laziness.
I don't suppose ChatGPT has quite mastered the art of English vernacular (it might be helped by reading this…), but it basically couldn't be bothered with its freewheeling assault on the internet, reasoning that it had even better things to do. The point is that it got the job done. Good on yer, Chatters!
Maybe that's a bit flippant, but we have to keep it in perspective. Every technology developed by mankind has been double-edged. Fire. The wheel. The internal combustion engine. Television (which was supposed to rot children's minds). Gunpowder. Nuclear fission. The hamburger. The printing press, you might well add. Well, maybe not penicillin (surely no downsides there?), but you take my point. Anyone can be terrified by new things.
Back in the 1890s, the British had a law whereby any horseless carriage ("car") had to be preceded by a man walking 60 yards with a red flag to warn others of the approach of this dangerous machine, and its speed was limited to 2mph in town and 4mph in the country. Well, we soon grew out of that, and we learned how to get the best out of the automobile and now, increasingly, how to trust it to drive itself. We're now steering it towards a quieter, less-polluting, more sustainable, electrically propelled future.
It's true that AI is a uniquely powerful and pervasive technology that presents vastly more awesome challenges, but the principle has to be the same. We have to adapt to it and manage it when it becomes a problem, like anything else. We can use AI to control AI, even if it becomes an endless game of cat and mouse, as it will inevitably become. We're still in the foothills, and we're assuming that AI is necessarily and intrinsically malign – but it is not, any more than, say, a gun is.
The other reason why we'll have to learn to get the best out of AI – and that includes its economic impact – is that we don't have much choice. The genie is out of the bottle and cannot be suppressed, even if we wanted to and had the international frameworks to do so, which remain lacking despite strong British initiatives taken by Rishi Sunak, who did and does understand its real significance.
Like previous industrial revolutions, AI has the potential to rescue us from our present low-productivity, stagnant-living-standards torpor and unleash another Industrial Revolution, where robots can clean the loo, drive the taxis, analyse medical scans and an AI assistant can do your tax return – freeing us humans up to create our lovely leisure utopia.
We shall need a system to make sure that people who can no longer get jobs are well provided for in this plentiful future, and that income and wealth don't become obscenely concentrated in the hands of those who own the new capital.
A radically different world is not necessarily anything to be frightened of but, even if it were, it's coming at us anyway.
The Independent is the world’s most free-thinking news brand, providing global news, commentary and analysis for the independently-minded. We have grown a huge, global readership of independently minded individuals, who value our trusted voice and commitment to positive change. Our mission, making change happen, has never been as important as it is today.
