Tech companies use A/B testing to optimize for short-term engagement metrics like time spent, which can lead to addictive product features that harm users long-term. The author draws parallels between social media's documented negative effects and emerging concerns about AI chatbots, arguing that optimizing for engagement without considering broader consequences repeats harmful patterns seen in tobacco and processed food industries. The piece advocates for product builders to take responsibility for understanding and mitigating long-term user harm, rather than blindly following metrics that increase engagement.
Table of contents
How big tech launches features through A/B testingSkepticism of experimentationHacking human psychology for engagementWill we let the pattern repeat with chatbots?Root cause mattersWe can do betterSort: