Elon Musk says the world is running out of data for AI training
Elon Musk says he's worried that "we’ve exhausted basically the cumulative sum of human knowledge" for training AIs. The solution: just have the AI make something up.
Tesla/X CEO Elon Musk seems to believe that training AI models with solely human-made data is becoming impossible. Musk claims that there’s a growing lack of real-world data with which to train AI models, including his Grok AI chatbot.
“We’ve now exhausted basically the cumulative sum of human knowledge … in AI training,” Musk said during an X live-stream interview conducted by Stagwell chairman Mark Penn. “That happened basically last year.”
Musk’s comments reflect those of former OpenAI researcher Ilya Sutskever, who predicted last December that the AI industry had reached “peak data.” Musk’s solution to this issue — synthetic data — also mirrors the larger industry. Google, OpenAI, Anthropic, and Meta already leverage synthetic data to train their models.
“The only way to supplement [real-world data] is with synthetic data, where the AI creates [training data],” Musk said. “With synthetic data … [AI] will sort of grade itself and go through this process of self-learning.”
While the use of synthetic data can offer significant cost savings to companies, some studies suggest that over-reliance on synthetic data can lead to model collapse where the AI’s responses become less creative and more biased over time as they’re repeatedly trained on recursively generated data.
The lack of human-derived data hasn’t stopped X from spinning off its Grok AI feature into its own iOS app on Thursday. The chatbot and image generator, notable for their complete lack of intellectual property or content guardrails, used to only be available to folks shelling out $8 a month for an X premium account. However, the new app is free for anyone to download.