COURSE 06L1100% FREE
Verified 2026-08-10

Uncertainty and Bayesian thinking

**Uncertainty** is not knowing for sure. **Bayesian thinking** is a disciplined way to update beliefs when new evidence arrives—starting from a prior guess, then revising.

What it is

Uncertainty is not knowing for sure. Bayesian thinking is a disciplined way to update beliefs when new evidence arrives—starting from a prior guess, then revising.

Why it matters

AI systems output scores, rankings, and confident prose. Users need to ask: *how sure should I be?* Medical, legal, and financial uses demand calibrated doubt (Course 29 medical overtrust).

How it works (plain)

  1. Start with a prior belief (“this email is probably not a scam”)
  2. Observe evidence (urgent money request from a new domain)
  3. Update toward a new belief

Good systems surface uncertainty. Bad ones hide it behind fluent sentences.

Everyday example

Weather: 30% chance of rain is a probability statement—not a promise it will drizzle lightly.

Try it

Take one AI answer you trusted this week. Write: prior, evidence, remaining uncertainty, how you would verify.

Myths

⚠️ Myth: Probability 0 or 1 is normal for real-world claims.
✓ Reality: Extremes are rare outside closed formal systems.
⚠️ Myth: A model saying “I’m confident” equals calibrated probability.
✓ Reality: Verbal confidence is not a trustworthy probability meter unless evaluated.

Sources