tech
AI giants learn what everyone else on the modern internet already knows
Anthropic's distillation complaints expose an awkward question: does AI's fair use argument cut both ways?
TL;DR
- AI companies have long argued that information on the internet can be used for AI model development under fair use.
- Now, companies like Anthropic, OpenAI, and Google are complaining about 'distillation,' where competitors use their AI model outputs to improve their own models.
- This situation creates an ironic symmetry, as AI companies have been scraping web content without permission for years, similar to how rivals are now using their AI outputs.
- Concerns are raised about the cost-effectiveness of developing AI models if competitors can replicate intelligence cheaply through distillation.
- The article suggests that once information is online, it's difficult to control its use, a lesson the AI giants are now learning.
- The legal arguments around fair use may apply to distillation as well, cutting both ways for AI companies.