Lecture 11

Question 1

Algorithms

Answer 1

Any evidence-based forecasting formula or rule

Question 2

Why do we use algorithms?

Answer 2

• People can be irrational and make mistakes (cognitive biases)
• People don’t know what they don’t know (Dunning-Kruger effect)
• Experts might have incentives to be entertaining, not only accurate

Question 3

Trust in algorithms..

Answer 3

• Has reduced for tasks that matter (consequential tasks)
• Has increased when consumers are familiar with the algorithm
• Has increased for objective tasks

Question 4

Algorithmic bias

Answer 4

Algorithms can lead to unintentional discrimination

Question 5

Examples of algorithmic bias:

Answer 5

• Biased healthcare decisions; discrimination against black patients
• Biased hiring decisions; person’s CV
• Biased language; “Female doctors don’t exist”

Question 6

Algorithms can be unfair with special circumstances:

Answer 6

Evaluation of teachers –>
- Changes in student’s circumstances
- Teachers of top students receive worse scores
- Incentivizes teachers to help students cheat

Question 7

Moral machine experiment

Answer 7

• 3 clusters; western, eastern and southern
• Data shows preference for sparing human lives, more lives and young lives