AI chatbots could hit a ceiling after 2026 as training data runs dry

The stock of language data that artificial intelligences like ChatGPT train on could run out by 2026, because AIs consume it faster than we produce it

By Chris Stokel-Walker

The supply of high-quality language data used to train machine-learning artificial intelligence models may run out in three years, leading AI advancement to stagnate.

Machine learning powers AI programs like text-prompted image generator Midjourney and OpenAI’s chat-based text generator ChatGPT. Such models train on vast reams of human-created data from the internet to learn, for instance, when asked to draw a banana that it should be …