Why does AI answer differently every time?
Temperature isn't a knob that makes AI smarter — it's a knob for how bold it gets.
1 · Think first
When AI picks the next piece, it gives each candidate a probability, then does a lucky draw using those odds.
Temperature is the knob that decides how flat that probability table gets — turn it down and only #1 gets drawn; turn it up and lower candidates get drawn often too.
So with a low temperature the answers come out pretty much the same, and with a high one they get more varied, but the risk of going off the rails grows too.
2 · Experiment · Probability slider
Try it — turn the temperature knob
So far you only looked at the candidates. Today you change the probability table itself. Move the knob and the bars really change length.
The percentages are made-up example numbers to explain the idea. A real AI does the same math, but with tens of thousands of candidates.
Figure
In one picture
Same candidates, same ranking — change only the temperature and the bars look this different.
3 · Quiz
Quick check
Three questions. Pick an answer and it's marked right away, with the reason why.
Today in one line
Today in one line
Temperature isn't a knob that makes AI smarter —
it's a knob for how bold it gets.
Learn more
- This trick of flattening the probability table with temperature became widely known through Hinton's team's 2015 distillation paper — Distilling the Knowledge in a Neural Network
- A 2026 experiment ran things 4,705 times and found that big AIs rarely give the same answer even at temperature 0 (models of 120B and up were under 20% reproducible) — Replayable Financial Agents