Andrew Chen: novelty becomes a retention constraint

Andrew Chen argues that AI excels when outputs are verifiable, while consumer products compete for novelty, parasocial connection, and cultural freshness. He calls this adversarial creativity: originality resists evaluation against a model’s recent training data, and direct AI copy-pasting can make messaging feel inauthentic. Why it matters: As digital products become easier to ship, retention may increasingly resemble content: an initial spike followed by the need to earn attention again. The risk is not only weak AI output; it is a product or message signaling automation and breaking trust.

Try this: For one consumer-facing AI feature, run a two-week cohort review: compare return use after a human-edited experience with a copy-pasted AI version, then record qualitative reasons for drop-off. Use the retention curve and feedback as the decision record.