What should an AI optimize when people want different things?
Imagine a library recommendation tool that could maximize loans, introduce unfamiliar books, or help readers find exactly what they came for. Those goals can point toward different recommendations. Choosing a performance measure also chooses what counts as success, and a high score cannot settle whose priorities deserve weight.
Questions you could explore
- Who should help decide the goal?
- What useful outcome might disappear if the tool optimized only one number?
Start reading
- National Institute of Standards and TechnologyArtificial Intelligence Risk Management Framework (AI RMF 1.0)
Explains why balancing AI goals involves context and values, not only technical measures.
Ready to try your ideas aloud?
Your topic carries over. Record up to 2½ minutes, depending on your plan.