What AI Does—and Does Not Do
Understand models as pattern-driven systems, not databases or people.
By the end
You will be able to
- Explain the difference between a model and a search database.
- Recognize why confident output can still be wrong.
- Choose when verification is required.
Models produce likely continuations
A language model learns patterns from large collections of examples. Given instructions and context, it generates a useful continuation one piece at a time.
That makes models excellent at explaining, transforming, drafting, classifying, and reasoning over supplied material. It does not make every output a stored fact.
Verification should match the stakes
Brainstorming can tolerate uncertainty. Medical, legal, financial, security, and production decisions require current authoritative evidence.
Good AI use combines a clear task, relevant context, constraints, and a way to verify the result.
Evidence
Sources and verification
- Artificial Intelligence Risk Management FrameworkNIST · verified 2026-07-23
Knowledge check
Make it stick.
Choose the strongest answer for each question. Your attempts become part of your device-local transcript.