Amelia Earhart Day: ChatGPT, Gemini, Claude & Perplexity tackle the mystery - Yahoo Tech
tech.yahoo.com
Four AI systems given the same Earhart disappearance prompt — do they converge on one theory or reveal how differently machines weigh the same cold-case evidence?
Occam's RazorConfirmation BiasEpistemic HumilityInformation Cascade
Theory Briefing
- ChatGPT, Gemini, Claude, and Perplexity were each asked the same question about Earhart's 1937 disappearance to compare how AI systems handle unresolved historical mysteries.
- The experiment surfaces whether AI models echo the most popular theory — navigation error and crash at sea — or give weight to fringe alternatives like capture or survival on a remote island.
- Because all four draw on overlapping training data, their agreements may reflect shared source bias rather than genuine independent reasoning.
- The Earhart case is a stress test for AI confidence: the evidence is genuinely ambiguous, so how each model hedges — or doesn't — reveals its design philosophy around uncertainty.