Understanding Bias in AI Algorithms

What was the purpose of Artea’s A/B test?

• To test a new website layout

• To evaluate the effect of discounts on inactive users

• To compare Facebook and Instagram marketing costs

• To collect demographic data for legal compliance

Which customer traits were most responsive to the coupons?

• Customers acquired through Google and with no purchase history

• Customers who left bad reviews and spent less than $20

• Customers from Instagram/Facebook and those who abandoned shopping carts

• Customers who previously purchased over $100

What was the main concern that led Alex Campbel to request a demographic audit?

• Declining coupon redemption rates

• Reports on algorithmic bias in marketing practices

• Customer complaints about ad targeting

• Legal requirement from the FTC

How did Artea initially segment customers for coupon targeting?

• Based on gender and race

• Based on income level and job title

• Based on acquisition channel and shopping behavior

• Based on location and age

Which demographic group was disproportionately acquired via Google?

• Non-male customers

• Minority customers

• Majority male customers

• Customers who spent over $100

The A/B test showed that coupons increased revenue significantly.

• True

• False

Artea originally collected demographic data from users at sign-up.

• True

• False

Algorithmic bias can occur even if the algorithm doesn’t explicitly use race or gender.

• True

• False

Customers acquired via Facebook were mostly minority customers.

• True

• False

Which of the following best describes algorithmic bias in artificial intelligence?

• The tendency of an algorithm to perform better on certain types of data.

• The systematic and repeatable errors in a computer system that create unfair outcomes.

• The ability of an algorithm to learn from new data.

• The speed at which an algorithm processes data.