When an AI agent destroyed PocketOS's production data, many observers dismissed it as the result of irresponsible users. This post argues that reaction is a well-documented cognitive trap called 'distancing through differencing,' coined by resilience engineering researchers Richard Cook and David Woods. By focusing on how 'those people' are different from us, we convince ourselves an incident couldn't happen to us and forfeit the opportunity to learn from it. The same pattern appears in workplace accidents: workers on a different shift blamed each other rather than examining shared systemic risks. As AI-related incidents become more common, the instinct to call out others as bozos will actively prevent organizations from improving their own safety practices.
Table of contents
Share this:Sort: