Did Two Rival AI Labs Just Overturn a Nearly 30-Year-Old Law of Computer Science on the Exact Same Day?

For nearly three decades, computer scientists built hundreds of foundational proofs on a single bedrock assumption: the Exponential Time Hypothesis, or ETH (OpenAI, The Daily Star). Proposed in the late 1990s by researchers Russell Impagliazzo and Ramamohan Paturi, ETH states that solving complex logical puzzles known as 3-SAT Boolean satisfiability problems inevitably requires computing time that explodes exponentially as the problem grows (Emergent, The Daily Star).

Now, that 27-year-old pillar of theoretical computer science is facing an unprecedented challenge—not from a human mathematician, but from two competing artificial intelligence models that independently arrived at the same breakthrough on the exact same day, October 16 (The Daily Star, Synapse Negar).

Both OpenAI and Anthropic have published AI-generated papers—spanning 67 pages and 49 pages respectively—presenting subexponential-time algorithms that appear to refute the Exponential Time Hypothesis (Emergent, The Daily Star). Even more remarkably, these aren't vague chatbot hallucinations. Both labs formalized their proofs using Lean, a strict mathematical verification system that checks every single line of logic for errors (OpenAI, Emergent).

When Russell Impagliazzo learned that AI systems had challenged the conjecture he co-created in 1998, he called the development "very shocking news" (Emergent). Meanwhile, MIT complexity theorist Ryan Williams, who is helping lead an independent academic review alongside Impagliazzo and Paturi, noted that the proofs contain 

"two or three alien-looking parts" that require the entire scientific community to unpack together (OpenAI, Emergent). 

While the review team emphasized in a joint statement that (Emergent): "our involvement should not be taken as certification of the claim"

the implications are already sending shockwaves through universities worldwide (Synapse Negar). Here is what this breakthrough actually means for the real world:

• Rewriting the textbooks: According to an analysis by Anthropic, more than 800 published scientific papers across graph theory, databases, computational biology, and geometry rely on ETH being true (OpenAI, Synapse Negar).

• Keeping your bank passwords safe for now: Despite viral rumors online, this does not prove that P equals NP or break modern encryption overnight, because the speedups only kick in at astronomically massive problem sizes exceeding $10^{2,389}$ variables (Emergent, Synapse Negar).

• Unlocking future chip and algorithm design: Researchers say studying the unorthodox mathematical techniques inside these AI proofs could inspire faster chip architectures and new approaches to problems scientists had given up on years ago (Emergent, Synapse Negar).

We are witnessing a historic shift: AI is moving beyond summarizing existing human knowledge to discovering—and formally proving—new frontiers in pure mathematics and physics, including OpenAI's parallel formalizations on the Kakeya conjecture and Yang–Mills quantum field theory (OpenAI).

How do you feel about AI discovering mathematical techniques that look "alien" even to top human scientists? Share your take in the comments!

#TechNews #ArtificialIntelligence #ComputerScience #Mathematics #OpenAI #Anthropic #FutureOfScience #TecZgyan