OpenAI and Anthropic each say their AI models have overturned a 27-year-old assumption in computer science, with both claims surfacing on the same day, October 16.
October 16 · OpenAI and Anthropic
Two AI labs claim their models overturned a 27-year-old computer science assumption
Both companies say their systems found a faster approach than the Exponential Time Hypothesis was thought to allow. The claims arrived on the same day, but the technical details and independent verification have not yet been reported.
27
years the Exponential Time Hypothesis has stood as a complexity assumption
2
companies making similar claims on the same date
0
independent verifications reported in the material reviewed
Claims made vs. independent checks
Each block = one claim or one independent verification
Independent checks
0 reported
Takeaway: two claims stand on the table, and no independent verification has been reported yet.
How a claim becomes an established result
1. Claim
Labs say models found a faster approach
→
2. Methods published
Problems, methods and results disclosed
→
3. Peer review
Researchers check the proofs or results
→
4. Scope clear
Size and reach of the finding understood
If the claims hold up
A result challenging a long-standing complexity conjecture would be notable for theoretical computer science.
It would support the argument that AI systems can contribute original research, not only assist existing work.
Open questions
The specific problems addressed are not yet known.
The size of the improvement cannot be assessed yet.
It is unclear whether either result has been peer reviewed.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article...