A British psychologist wrote a paper in 1983 that predicted every single problem we’re having with AI in 2026. It's called “Ironies of Automation”
The more advanced and reliable an automated system… | Patrick Slavenburg | 31 comments
The post argues that Lisanne Bainbridge’s 1983 paper, *Ironies of Automation*, anticipated major risks associated with modern AI systems. Its central warning is that as automation becomes more reliable, human operators become responsible mainly for rare, complex, and potentially dangerous exceptions—the very situations least amenable to routine procedures. Meanwhile, frequent reliance on AI can erode people’s practical skills and mental models. When an AI system fails or “hallucinates,” humans may be expected to intervene despite being out of practice and confronted with an urgent, difficult problem.
The discussion also challenges the idea that simply placing a human “in the loop” ensures safety. Humans are poor at maintaining vigilance for rare failures, and they may be unable to evaluate systems operating faster and more complexly than they can follow. Oversight therefore requires more than approving outputs: it needs accountability, independent safeguards, clear operational boundaries, fault recovery, and mechanisms for detecting missing reports or silent failures.
Commenters broaden the issue beyond AI, comparing it with Socrates’ concerns about writing and with decades of industrial automation experience. They emphasize that effective use of agents depends on curiosity, asking informed questions, challenging outputs, and maintaining engagement rather than accepting results uncritically. AI may expand human capabilities, but current processes remain immature. The main risk is not automation alone, but systems that preserve human responsibility while degrading human judgment, creating an illusion of control and potentially concentrating blame on weakened supervisors.
Article timeline
Build this article’s timeline
Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.
Comments
Login to view and post comments.