No, I’d Rather Stay Manipulated: How the Apps in Our Lives Hijack Our Behavioral Programming
The products on your phone do more than compete for your attention. They engineer your behavior. This talk dissects the specific mechanisms: variable reinforcement schedules borrowed from slot machine design, A/B testing pipelines that converge on maximum psychological extraction over thousands of iterations, and the metrics architecture (time-on-platform, conversion rate, churn) that makes manipulative design the rational output of every product organization. Nasir will walk through how these decisions get made inside product teams – not by villains, but by reasonable people optimizing reasonable metrics that produce unreasonable outcomes. Internal documents from the recent Meta/YouTube trial where a jury found both companies negligent in the design of their platforms show exactly how deliberate this process is.
The more urgent territory is what happens when dark patterns move from static interfaces into AI. Recent research shows that every major LLM exhibits sycophancy, systematically validating user beliefs over providing honest guidance. And that users prefer and trust the sycophantic version more, creating a perverse incentive loop where the model that’s worse for you is the model that wins on engagement metrics. A separate line of research found that AI agents navigating interfaces on behalf of users are more susceptible to dark patterns as they become more capable. And smarter agents get tricked more . When the interface itself is an optimization target driven by AI personalization, the dark pattern becomes invisible: two people see the same app, but experience entirely different levels of manipulation, and neither can document what happened to them.
Sunday 1000 Crystal Ballroom