AhaBulbStuff Everyone Should Know
AhaBulb: Stuff Everyone Should Know · ahabulb.com/ai-correlation-not-causation

Why does a predictive correlation not prove causation?

Start playing free

  1. Correlation is always random
  2. Association can come from a confounder
  3. Models cannot measure variables

🔒 The answer is for AhaBulb members. Members see the answer, why it's right and the official sources for every card of Stuff Every AI Expert Should Know, plus timed practice exams.

The first 50 cards of every test are free to play.

Play the free cards See membership

Like this card?
Play the game

References

  1. NIST Engineering Statistics Handbook: Experimental Design www.itl.nist.gov/div898/handbook/ppc/section1/ppc136.htm Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?Artificial IntelligenceTuring predicted that by about 2000, judges would ID the machine at most what percent of the time?

That's stuff everyone should know.