OpenAI announces unverified scientific discoveries amid growing criticism
In brief
- OpenAI publishes AI-generated « scientific discoveries » without peer review.
- 722 math preprints and three recent science claims remain unverified.
- External experts are consulted, but researchers criticize methods and lack of collaboration.
OpenAI is increasingly making claims of scientific « discoveries » generated by its AI models, but struggles to get them peer-reviewed. According to Next, the company has published 722 preprints in mathematics generated by AI, alongside three new announcements in mathematics, computer science, and physics, none of which have been validated by experts.
These results, shared via blog posts or GitHub, remain unconfirmed hypotheses. OpenAI has involved external experts for evaluation, but researchers like Levent Alpöge from Anthropic criticize these practices, including ignoring existing work and publishing without prior collaboration.
These efforts align with a broader strategy where OpenAI and Anthropic aim to position themselves as major funders of global research, despite lacking internal expertise and confirmed scientific credibility.
This summary gives you the essentials. The full article, its figures and quotes are on Next.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error