Decisions By Algorithm

Imagine that you are running the AAA admissions office. A business analytics professor has suggested that you could do a better job of picking students for the school if the admissions office used models to predict success and then just admit the students that score highest. Note: Humans can still be used to numerically code subjective judgments.
1. What do you think of this plan? What might models do better than expert admissions officers? (focus on Camerer reading)
2. What problems do you anticipate? How would you solve them?
3. How can we ensure that the model maintains what makes AAA unique?
Hint 1: Specify what measures of success the model is trying to predict for applicants. That is, what success metric is the model trying to maximize.
Hint 2: Even if you are against using models, you still must base your write-up on this week’s readings (and not just discuss your own opinions)!