The LLM vs human distribution chart on picking a number from 1 to 10
khademinori · x · 2026-10-15
A widely shared chart shows that when asked to pick a number between 1 and 10, LLMs produce an extremely sharp spike around 7, while the human distribution is much broader and covers lower numbers, and random choice is flat. The reshare notes this is unsurprising given how gradient descent works: training pushes models toward the mode of the data, inherently lacking randomness.
Related event: Ask an LLM to pick 1-10 and it almost always says 7(2 posts)→
More from Fun
- Peeqo, the Desktop Bot That Only Replies in GIFs, Wins Hearts Online — shekitup · 2026-10-15
- Gamer says new game footage is first jaw-dropper since OG Soul Calibur on Dreamcast — ID_R_McGregor · 2026-10-15
- The Harry Potter Galleon-to-Sickle 17:1 exchange rate hides a real arbitrage loophole — mgostIH · 2026-10-15
- "First contact with an alien life form, and it's made of us": Chris Albon on AI — chrisalbon · 2026-10-15
- Schmidhuber renews residual connections credit claim: "LSTM is a ResNet rotated 90 degrees" — SchmidhuberAI · 2026-10-15
- OpenAI's Will Brown Has an LLM Draft Joke Tweets About Google's Gemini Interviews — Obijuan_cube · 2026-10-15