AI Clinches Gold at the Math Olympiad
Also: Netflix starts using GenAI and SimilarWeb says search traffic fell by 15%
An experimental reasoning model from OpenAI just scored a Gold medal at the International Math Olympiad (IMO), the world’s most prestigious math competition for young mathematicians.
This is the first time an AI model has reached this milestone.
The Math Olympiad is designed to test for creative, problem-solving skills rather than textbook knowledge. Contestants compete by devising mathematical proofs during two 4.5-hour sessions, without a calculator or tools.
Some caveats: Fields Medalist Terence Tao (also the youngest IMO participant on record!) notes that AI-versus-human comparisons can be apples-to-oranges without transparent methodology.
Even so, most observers agree this result is a watershed moment. If true, AI systems can now produce intricate, watertight arguments that rival those of top human mathematicians.
Why this matters:
Marathon-length reasoning
The model can sustain a line of thought for hours, and soon, perhaps, days or weeks. That opens the door to tackling complex, multi-stage problems and workflows.
Proof-grade coherence
IMO judging revolves around multi-page proofs. Clearing that bar means the AI can generate consistent and coherent arguments that can withstand human scrutiny.
Complex reasoning abilities will lead to faster scientific breakthroughs
We might soon see more breakthroughs in research and math aided by models like these, that can handle complex logic and proofs.
If AI is inching towards invention-level reasoning, it could accelerate humanity’s entire timetable for scientific breakthroughs.
And unlike how calculators moved humans away from arithmetic… we might see human researchers move more towards deep questioning, curation, and steering research, instead of working through derivation.
Happy Monday.
-TTYL
The Download —
OpenAI just won gold at the world's most prestigious math competition. (LessWrong)
An experimental LLM from OpenAI won gold at the International Math Olympiad, solving five out of six of the problems correctly, working under the same testing conditions as humans.
Netflix starts using GenAI in its shows and films (Guardian)
Netflix has used artificial intelligence in one of its TV shows for the first time, in a move the streaming company’s boss said would make films and programmes cheaper and of better quality.
“We remain convinced that AI represents an incredible opportunity to help creators make films and series better, not just cheaper,” Ted Sarandos, a co-chief executive of Netflix.
Perplexity AI hits $18 billion valuation in latest funding round (Pyments)
Perplexity AI has secured $100 million in new capital, boosting its valuation to $18 billion, Bloomberg reported. The deal extends a recent funding round that had previously valued the company at $14 billion.
Similarweb estimates that worldwide search traffic fell by about 15% in the year to June (@shashj)
Humans no longer visit the websites from which the information is gleaned. Similarweb, which measures traffic to more than 100m web domains, estimates that worldwide search traffic fell by about 15%.