If a week is a long time in politics, a year is an eternity in tech. Just over 12 months ago, the industry was humming along in its usual way. The big platforms were deep into what Cory Doctorow calls “enshittification” – the process in which platforms go from being initially good to their users, to abusing them to make things better for their business customers and finally to abusing those customers in order to claw back all the value for themselves. Elon Musk was ramping up his efforts to alienate advertisers on Twitter/X and accelerate the death spiral of his expensive toy. TikTok was monopolising every waking hour of teenagers. FTX had just gone bankrupt and at least $1bn of investors’ money had gone awol. Here in the UK, the bedraggled online safety bill was wending its way through parliament. And nobody outside the tech world had ever heard of Geoffrey Hinton or Sam Altman.
And then one day – 30 November 2022, to be precise – everything changed. OpenAI, an upstart tech company headed by Altman that had been building so-called large language models (LLMs) for some years, released ChatGPT. The strange thing, though, was that, even weeks earlier, ChatGPT wasn’t a product. OpenAI’s focus was elsewhere – on GPT-4, the biggest and most powerful model the company had built. This was a machine that could apparently answer almost any question using information gleaned from having “read” everything ever published, but which would sometimes also make stuff up and was therefore deemed not ready for public consumption. Altman, possibly spooked by the fear that a rival company, Anthropic, would launch something big, then made a fateful decision: to release an older, less powerful version of the GPT technology – GPT-3 with a bolted-on chatbot front end – and see what happened.
Well, we know what happened. The public went for ChatGPT in an unprecedented way. In less than a week, it went from zero to 1 million users, making it the fastest-growing app ever. We also know that to several of the tech giants it came as a “Pearl Harbor moment”, metaphorically akin to the surprise Japanese attack that brought the US into the second world war. ChatGPT triggered panic and U-turns at the highest levels of these corporations, but we lacked a ringside view of how the crisis played out in those elevated circles. Until now.
Last Tuesday, the New York Times came out with the first overall narrative of what went on. The story its reporters tell is entertaining, instructive and sobering: entertaining because it isn’t often one sees these guys (and they are all guys, BTW) in a flap; instructive because it provides a vivid picture of how they operate when they’re threatened; and sobering because it reveals how they are likely to behave in the new landscape shaped by “generative AI”.
The thing is that these outfits had been working for years on the machine-learning technology underpinning LLMs, but had held back from launching them because of concerns about their implicit limitations and dangers. And then Altman had the temerity to release ChatGPT despite its flaws, hoping to get a head start on everyone else. The normally sober Microsoft boss, Satya Nadella (who had provided the cloud computing muscle behind ChatGPT), folded the technology into his Bing search engine as a way of “making Google dance”. Google – terrified that its search monopoly was threatened – obligingly complied, in the process rushing out its own LLM, complete with its own howlers.
Meanwhile, over at Meta (née Facebook), Supreme Leader Zuckerberg was compelled to divert his imperial attention from the Metaverse and agree to launch Meta’s Llama LLM as “open source” – after which the code was leaked on to 4Chan, which meant that Meta lost control of its creation. And then Elon Musk, taking a break from destroying Twitter/X, announced that he, too, would be releasing a chatbot, though his would have a sense of humour. (I am not making this up.)
And so it went on, this riveting saga of allegedly smart guys caught in an arms race and doing stupid and risky things. Underpinning it all, though, is the real takeaway: the tech giants knowingly rushed imperfect products to market, worried that being cautious would allow rivals to dominate AI. Safety and ethical concerns took a backseat to the profit motive. Plus ça change. And these are the guys who aspire to be masters of our universe.
What I’ve been reading
Kissinger assessed
A People’s Obituary of Henry Kissinger is a scarifying account by the historian Gregg Grandin in the Nation of the US secretary of state, who died aged 100 at the end of last month.
Marienbad mullings
Granta has published a nice witty and ironic essay by Lauren Oyler, titled Last Week at Marienbad.
The fact is…
Ethan Zuckerman has written an interesting blogpost – Is the Web Eating Itself? – about Heather Ford’s presentation on the creation and dissemination of knowledge.
