Our AI leaders promised us cures for diseases and even computer gods. They’re giving us slop

For years, OpenAI and others have been searching for investment with the terrifying threats and glorious promises. AI is coming, the companies warned, and it was going to be significant, though they were often unclear about whether that significance would bring malevolent computer gods or cures for cancer, and often suggested both.

All of that will happen when the companies finally achieved the long-promised goal of AGI, or artificial general intelligence. Nobody quite knows exactly what that means, but it roughly refers to a point at which AI will be able to compete with the kind of broad intelligence of humans.

Really, all of this was a plea for funding. The AI industry has been gripped by the idea of scaling – of increasing the data, energy and computers that are being used to power artificial intelligence – and the claim that more would eventually bring better, and eventually be so much better that it changed everything.

But that idea of scaling has been in trouble in recent weeks. OpenAI boss Sam Altman had been hyping the release of GPT-5 for months – suggesting that the company had achieved AGI internally, and that the power of the new version was such that it could bring huge changes – but when it arrived, it was weak and disappointing, and suggested that simply feeing systems more data and more computers to process it on would not bring exponentially better results.

But the company has continued to stress that it will eventually get there. With more money – Sam Altman is reported to be hoping that he will receive trillions of dollars to fund this plan – eventually the computers will rival humans.

For now, though, he seems to believe that the pressing problem is that there are not enough short-form videos. In recent days, both Meta and OpenAI have launched new ways of making them: each now has an app in which you can quickly make and share AI-generated videos of whatever bizarre, meaningless flavour you’d like.

Unsatisfied with the fears about TikTok and YouTube Shorts shortening our attention spans, overwhelming our dopamine systems and obliterating a common understanding of truth, both companies have seemingly delayed their mission to cure diseases and get rid of the need for work to instead provide us with more six second videos.

Criticised for the release on Twitter, Altman has suggested that the short videos are part of a broader plan to get to AGI, but that it was also aimed at making people "smile". "i get the vibe here, but... we do mostly need the capital for build AI that can do science, and for sure we are focused on AGI with almost all of our research effort," he wrote on Twitter, with his trademark relied lack of capital letters. "it is also nice to show people cool new tech/products along the way, make them smile, and hopefully make some money given all that compute need. When we launched chatgpt there was a lot of 'who needs this and where is AGI'. reality is nuanced when it comes to optimal trajectories for a company."

There hasn't yet been all that much to smile about. Soon after Meta launched its new app, ‘Vibes’, a post taken from Facebook went viral that has to be seen to be believed – its narrative is both bizarre and largely impossible to capture in words – but which many did see and seemingly believe, sharing it joyously on the app. And then when OpenAI launched its new Sora app, which lets people ask for videos and then watch them before sharing them in a TikTok-like feed, the company’s chief executive Sam Altman did so with a long post that was largely focused on trying to reassure the world it wouldn’t be used only for bad purposes.

The thing that's good about videos is not that they exist. Anyone can produce more video files; my phone regularly does it in my pocket. There is, actually, far more content already in existence than you could ever care to watch. One of the central anxieties of being alive, for me at least, is the constant knowledge that you could never watch everything good.

This was one of the great breakthroughs of the TikTok scroll and the algorithm that powers it: it makes you feel that there is infinite content available to watch, and also that the next one will be the one you've been looking for all this time. Its power is in the way that it gives you no real choice about what you see but the promise that you will see all of the best stuff.

The only way we can cope with the sheer amount of video that is in offer is the belief that we are coming into contact with at least some of the best of it. And those really good videos are not simply content – automatically generated, with however much help from a computer – but that they are a statement about the world, and about the people in it. In that sense the videos we love might be plucked from near-infinity but they never feel arbitrary: they are a record of something that happened, or something that someone imagined might happen, a document of ideas in a person or group of people's brain.

In its (AI-generated) launch video, OpenAI's Sam Altman called Sora "the most powerful imagination engine ever built". (It's arguable whether Sora or the videos it makes actually count as "imagined", but that's mostly philosophical.) But imagination without ideas are just arbitrary pictures; there's a reason that dreams are rarely interesting, and only usually become so if you know the people dreaming and the meanings their visions might carry.

Today's automatically generated short-form videos are a vision of creative industry without creators. (Again, there's another philosophical question about whether this can even be true given that these AI systems must be trained on real and existing video made by humans.) On the release of Sora, that removal of people seemed to some commentators to be the exciting thing: at launch, the video was being shared with semi-gleeful messages such as "RIP Hollywood". Not only do you not filmmakers but you also don't need anyone to choose the videos, when they are being piped to people through an algorithmically sorted swipeable feed. (iJustine, a tech content creator, suggested in a video about the launch of the Sora app that it is "comforting" because you know everything on there is AI; it not only gets rid of filmmakers and programmers, but also the need to think about whether things are real.)

Of course, there's the possibility that what is fun here is that the videos are personalised. The fact that people can add their own faces in – through "cameos" – means that they can be so personalised as to be entirely individual, which might mean that the videos are especially popular in group chats and similar. (They could become a new and more uncanny version of JibJab, the app that let people create videos with their own face stitched in, or Snapchat filters.) They are a kind of instant, elaborate inside joke.

But really important video feels like that which transcends those specific moments, and offers a new way of saying and seeing things. That might be a grandiose way of talking about video in a world where one of the most popular moving pictures are Skibidi Toilet, a series of videos about a singing toilet bowl, and which feels the very definition of meaningless brain rot. But the thrill of these videos – as far as I can tell – is in part the shared joy at watching them as well as the bemused reactions of people who don’t get it. Even those older people who gleefully watch the bizarre slop that appears to be taking over Facebook are doing so on a platform built for sharing and communal enjoyment. Among the toilet bowls and the slop, the human keeps creeping through.

A version of this article was published in the IndyTech newsletter. You can subscribe to it and the rest of The Independent’s updates here.