3.2 Bias in model evaluation
5 blocks · 18 min · Outcome 3.2
Text
Human · written by you
3 min
In the loan example, the team reported 94% accuracy. Two problems hid inside that number…
AI draft · from Week3_slides.pdf p. 4–7 · v3
Remember: a single headline metric is a claim about the typical user average person. Fairness questions are about who the average ignores hides.
history
AI draft · not yet reviewed
Knowledge check
Reading level grade 14 · target 11
Outcome 3.2 · 2 min
Which change would you make first before trusting the 94% figure? (3 options · 1 correct · feedback drafted)
/ Add a block: text, video, scenario (AI role-play), reflection, discussion, assignment…