On Thursday, Google DeepMind introduced that AI programs referred to as AlphaProof and AlphaGeometry 2 reportedly solved 4 out of six issues from this yr’s Worldwide Mathematical Olympiad (IMO), reaching a rating equal to a silver medal. The tech large claims this marks the primary time an AI has reached this degree of efficiency within the prestigious math competitors—however as traditional in AI, the claims aren’t as clear-cut as they appear.
Google says AlphaProof makes use of reinforcement studying to show mathematical statements within the formal language referred to as Lean. The system trains itself by producing and verifying thousands and thousands of proofs, progressively tackling harder issues. In the meantime, AlphaGeometry 2 is described as an upgraded model of Google’s earlier geometry-solving AI modeI, now powered by a Gemini-based language mannequin educated on considerably extra information.
Based on Google, distinguished mathematicians Sir Timothy Gowers and Dr. Joseph Myers scored the AI mannequin’s options utilizing official IMO guidelines. The corporate studies its mixed system earned 28 out of 42 doable factors, simply shy of the 29-point gold medal threshold. This included an ideal rating on the competitors’s hardest drawback, which Google claims solely 5 human contestants solved this yr.
A math contest not like every other
The IMO, held yearly since 1959, pits elite pre-college mathematicians towards exceptionally troublesome issues in algebra, combinatorics, geometry, and quantity concept. Efficiency on IMO issues has grow to be a acknowledged benchmark for assessing an AI system’s mathematical reasoning capabilities.
Google states that AlphaProof solved two algebra issues and one quantity concept drawback, whereas AlphaGeometry 2 tackled the geometry query. The AI mannequin reportedly failed to resolve the 2 combinatorics issues. The corporate claims its programs solved one drawback inside minutes, whereas others took as much as three days.
Google says it first translated the IMO issues into formal mathematical language for its AI mannequin to course of. This step differs from the official competitors, the place human contestants work straight with the issue statements throughout two 4.5-hour periods.
Google studies that earlier than this yr’s competitors, AlphaGeometry 2 might remedy 83 % of historic IMO geometry issues from the previous 25 years, up from its predecessor’s 53 % success price. The corporate claims the brand new system solved this yr’s geometry drawback in 19 seconds after receiving the formalized model.
Limitations
Regardless of Google’s claims, Sir Timothy Gowers provided a extra nuanced perspective on the Google DeepMind fashions in a thread posted on X. Whereas acknowledging the achievement as “properly past what computerized theorem provers might do earlier than,” Gowers identified a number of key {qualifications}.
“The principle qualification is that this system wanted so much longer than the human opponents—for a number of the issues over 60 hours—and naturally a lot quicker processing pace than the poor previous human mind,” Gowers wrote. “If the human opponents had been allowed that kind of time per drawback they’d undoubtedly have scored greater.”
Gowers additionally famous that people manually translated the issues into the formal language Lean earlier than the AI mannequin started its work. He emphasised that whereas the AI carried out the core mathematical reasoning, this “autoformalization” step was achieved by people.
Relating to the broader implications for mathematical analysis, Gowers expressed uncertainty. “Are we near the purpose the place mathematicians are redundant? It is exhausting to say. I’d guess that we’re nonetheless a breakthrough or two wanting that,” he wrote. He steered that the system’s lengthy processing instances point out it hasn’t “solved arithmetic” however acknowledged that “there may be clearly one thing attention-grabbing happening when it operates.”
Even with these limitations, Gowers speculated that such AI programs might grow to be beneficial analysis instruments. “So we is likely to be near having a program that may allow mathematicians to get solutions to a variety of questions, supplied these questions weren’t too troublesome—the form of factor one can do in a few hours. That will be massively helpful as a analysis device, even when it wasn’t itself able to fixing open issues.”
