We Gave 4 AI Models the Same Toast Prompt

We handed the same prompt to DeepSeek Flash, GPT-6.1 Sol, Claude Haiku 5.5, and GPT-3.5 Turbo: write a three-minute episode of this show answering whether toast really lands butter-side down. This is the listenable version of the bakeoff — full scripts read aloud, no rubric, no scores. The physics shape came out roughly right in all four. The problem is what happened in the one spot the question actually asked about: whether anyone has tested it. Three of the four made something up — a floor full of buttered toast, a BBC presenter named Tim Richardson, a 2016 Bristol study with its result reversed. Raw outputs and the prompt are public in the show's eval repository if you want to check our quoting.

Episode 6142 — myweirdprompts.com/n/6142 (permanent short link: myweirdprompts.com/805606)