In my podcast this weekI took a closer look at OpenAI’s new video generation model. Sora 2which can turn simple text descriptions into impressively realistic videos. If you type the message “a man rides a horse that is riding another horse,” for example, you’ll get, well, this:
AI video generation is technically interesting and ethically concerning in every way you would expect. But there is another element of this story that is worth highlighting: OpenAI accompanied the launch of its new Sora 2 model with a new “social app for iOS” called simply Sora.
This app, clearly inspired by TikTok, makes it easy for users to quickly generate short videos based on text descriptions and consume others’ creations through an algorithmically curated feed. The videos circulating on this new platform are as outrageously stupid or morally suspect as you may have guessed; e.g,
EITHER,
In other words, the Sora app takes the already purified engagement that fuels TikTok and removes any vestiges of human action, resulting in an artificial high-octane drop.
It is not clear if this application will last. A major problem is the final expense of producing these videos. For now, OpenAI requires a paid ChatGPT Plus account to generate its own content. At the $20 tier, you can post up to 50 low-resolution videos per month. For a whopping $200 a month, you can generate more videos at higher resolutions. None of this compares favorably to competitors like TikTok, which is exponentially cheaper to run and can therefore not only remain truly free for all users, but also pay your creators.
However, whether Sora lasts or not is beside the point. What stands out to me the most is that OpenAI released this app in the first place.
It wasn’t long ago that Sam Altman was still Comparing the launch of GPT-5 with the testing of the first atomic bomband many commentators took Dario Amodei at his word when he proclaimed50% of administrative jobs could soon be automated using LLM-based tools.
A company that still believes its technology will imminently dominate large swaths of the economy and would be powerful enough to reconfigure our experience of the world as we know it would not be looking to make a quick buck by selling ads alongside completely fake videos of historical wrestling figures. They wouldn’t even consider the idea, like Altman did last weekthat they could soon start offering an age-controlled version of ChatGPT so that adults can enjoy AI-generated “erotica.”
To me, these are the acts of a company that poured tens of billions of dollars in investments into creating what they hoped would be the most momentous invention in modern history, only to ultimately realize that what they created, while very cool and powerful, is not powerful enough on its own to create a new world all at once.
In his famous 2021 essay,“Moore’s law for everything” Altman made the following grandiose prediction:
“My work at OpenAI reminds me every day of the magnitude of socioeconomic change that will come sooner than most people think. Software that can think and learn will do more and more of the work that people do now. Even more power will shift from labor to capital. If public policies are not adapted accordingly, most people will end up worse off than they are today.”
Four years later, he’s betting his company on its ability to sell ads against AI crap and computer-generated porn. Don’t be distracted by the hype. This change matters.
