GPT-5.2 Pro Solves Another Erdős Problem While New Database Reveals Most Attempts Still Fail
1 min read
AI for Software Engineering (Copilots, SDLC, Testing)
-/5
In short
- The recent development that GPT-5.2 Pro may have solved another Erdős problem is noteworthy.
- However, mathematician Terence Tao warns against a skewed perception: the actual success rate of AI on such problems is only one to two percent.
- In this context, it is important to note that despite advancements in AI research, the challenges and risks associated with solving complex mathematical problems remain significant.
The recent development that GPT-5.2 Pro may have solved another Erdős problem is noteworthy. However, mathematician Terence Tao warns against a skewed perception: the actual success rate of AI on such problems is only one to two percent. In this context, it is important to note that despite advancements in AI research, the challenges and risks associated with solving complex mathematical problems remain significant. A final assessment would be premature at this point, as the database documenting success rates indicates that most attempts still fail. The implications of these findings are far-reaching and necessitate a nuanced consideration of the opportunities and limitations of AI in mathematics.
Source: