It’s been a drama-filled week for OpenAI, the creator of ChatGPT. Its wunderkind CEO Sam Altman has been unceremoniously booted out by its board and more than 600 staff members are now threatening to quit unless he’s allowed back in. (Don’t be too sad for Altman – the 38-year-old has already been snapped up by Microsoft for an undisclosed sum.)
As a writer, I am of course duty-bound to swear on my copy of McNae’s Essential Law for Journalists that I did not use OpenAI’s chatbot to write this column – or did I? Even if I did, why would I fess up to it? Thanks to disastrously unpopular attempts by the likes of BuzzFeed to create AI-assisted content, its name is mud in the media industry. Saying you use ChatGPT is like admitting you thought the T-1000 was just misunderstood or that Skynet had a point.
I always regarded the chatbot as something of a pariah, a dirty little productivity tool. Using it was effectively declaring that your job was redundant – the equivalent of screaming “Automate me, Agent Smith!” in The Matrix. But then I realised that people I know are using it in ways I’d never considered. “Oh, I just use it to write emails,” one friend said airily over lunch this summer. “People expect these long, polite responses when I just want to say ‘Yes, do it’ or ‘No, don’t do it’, so I ask ChatGPT to pad my message out.”
In a taxi on the way to a wedding, a guest told me that he feeds data into ChatGPT when he’s feeling unimaginative: “Just to see if it can spot any trends. Sometimes it can get it totally wrong, but it does come up with some interesting stuff.” And just this month, I went to a friend’s place for dinner and found him typing the ingredients of his fridge into ChatGPT so it could come up with a recipe. Turns out you can make spaghetti bolognese with the leftover pulp from carrot juice – who knew?
Not exactly the stuff of I, Robot. But while we get wrapped up in the intrigue of tech company drama or breathless hypothesising about the as-yet unseen future, it’s easy to forget that people are already using AI tools in utterly mundane, banal ways to cut corners and save a bit of time and effort. ChatGPT alone claims to attract around 100m users a week. Not everyone on the internet is a sinister pervert making deepfakes of Bugs Bunny doing unmentionable things to Scrooge McDuck.
Scanning the X (formerly Twitter) timelines of some of its most ardent cheerleaders is enough to make you believe that anyone can be a Picasso or a Jane Austen nowadays, with the right tools. Artists, writers and film-makers are right to fear what AI might bring, but fortunately the only Van Goghs that AI has produced so far are hideous “AI-enhanced” versions of Starry Night, which resemble Van Gogh if he took too much acid and lost all sense of taste and artistic perspective.
What will probably happen is that AI will create a slush pile of what is considered “good enough”. That ChatGPT-written email? Maybe it would have been better off penned by a human, but it will suffice if you’re on deadline. If your client has stiffed you on a day rate, perhaps an AI-assisted trends report will do just fine. Think of it as reclaiming your time, if you must. The real question is: if your boss asks if you’re done with work, will you admit that you used ChatGPT to finish two hours ahead of time?
As for that AI-generated spaghetti bolognese recipe – well, it was perfectly edible. But with apologies to Chef-GPT, it could have done with a little more oomph: a sprinkling of sea salt, more garlic, perhaps even a cheeky little anchovy slipped in towards the end for a bit of umami unctuousness. Nigella doesn’t need to worry: it’s still the human touch that adds flavour – for the time being, at least.
• Zing Tsjeng is an author and freelance journalist
