How Good Is AI, Really?: teacher key

Artificial Intelligence, grade 11 edition, written for 11th grade. What is AI, and how do people measure it? Follow tests from Turing's game and Deep Blue to the 2026 AI Index, plus bias, hallucination, rules and power use.

Read the book. Print this page for the key.

Questions and answers

1. What is a false positive in a spam filter?
   • A good email wrongly flagged as spam (answer)
   • A spam email that gets through
   • A test that holds data back
2. How is a learning AI tested?
   • By asking it to rate itself
   • By holding some data back from training and checking accuracy on it (answer)
   • By training it twice
3. What did the 2026 AI Index say happened to tests meant to stay challenging for years?
   • They were never used
   • They became easier to read
   • They were saturated in months (answer)
4. What does the clock-reading result show about a top AI model?
   • It is better than people at clocks
   • It can win a math medal yet be unreliable at telling time (answer)
   • It cannot do math
5. Why did the 2014 Turing test result not count as reliable?
   • The trial was very small (answer)
   • The judges were machines
   • The test was banned
6. What happened to safety performance when people tried deliberate jailbreak prompts?
   • It rose for all models
   • It did not change
   • It dropped for all models tested (answer)

Words

• hallucination: confident mistake
• nodes: simple parts
• saturated: no longer challenging
• jailbreak: trick prompts
• neural: brain-like