Last February, I asked different AIs (ChatGPT, Gemini, DeepSeek) to investigate if composers have historically died younger than other professions. Two of them couldn’t answer adequately. The third, after much additional prodding, suggested that maybe, but without being able to provide any firmer answer than that.
A year has passed, and AIs have improved a lot. I asked the same question to Claude Code last week, using almost the same prompt than I used a year ago. Claude Code is optimized for processing complex prompts on its own: It looks for the right type of data, fetches it, builds an analysis pipeline, runs it, and interprets the results.
This time round, the answer was unambiguous: Composers between the years 1700 and 1900 on average died 5.3 younger than scientists (p=0.0003) and 3.8 years younger than statesmen (p=0.0066). They may also have died 1.0 years younger than writers, but this isn’t statistically significant (p=0.254). P were calculated using a one-tailed t-test and aren’t corrected for multiple testing.
The more important message is the massive improvement in AI capabilities within the last 12 months.
Nehaveigur 