Question: In a concise analytical response, explain how comparing a model to expert-level performance can help a machine learning team diagnose problems and move faster.

Sample answer: Expert-level performance gives the team a useful reference point for judging how much error is still unavoidable and how much is due to the current model. That comparison helps set a realistic target for what the system should eventually achieve. If the model is still far behind expert performance, the gap can reveal avoidable bias and point to the most promising fixes. On tasks where people perform well, the team also benefits from easier labeling, clearer error analysis, and more intuitive judgments about what kinds of mistakes matter.

Key points:

• Expert-level performance provides a reference for unavoidable error.
• It helps define a realistic target for system performance.
• A clear target can speed up development by focusing the team.
• A large gap can indicate avoidable bias.
• Human or expert intuition can support labeling and error analysis on tasks people can do well.

Rubric: A strong response explains how expert-level performance helps estimate unavoidable error, why it is useful for setting a practical target, and how a large gap can reveal avoidable bias. It may also mention easier labeling and more effective error analysis on human-solvable tasks.