A growing chorus of AI researchers warns that the rush to deploy ever-more capable digital assistants and autonomous systems could set off “gradual disempowerment”—a systematic handover of cognitive and physical work that leaves humans as passive approvers and weakens democratic leverage. Jay Chooi of Robocurve ranks the risk alongside engineered pandemics and cyberattacks, citing relentless market and political incentives to replace people with cheaper, tireless AI agents and robots. Jan Kulveit, co-author of a 2025 paper that coined the term, argues that as states and firms rely less on human labor and decision-making—extending from factories to battlefields—citizens’ influence erodes, echoing the resource-state dynamic that can foster authoritarianism. Margaret Mitchell of Hugging Face advocates “strategic friction,” keeping humans meaningfully in the loop to preserve skills and oversight. The authors propose measuring where AI is displacing work and designing systems to complement, not supplant, human judgment. The challenge, they say, may prove harder than stopping overtly harmful AI—because economic logic pushes automation forward even when society’s capacity for self-governance is at stake.
Related articles:
– Robots and Jobs: Evidence from US Labor Markets
– AI Risk Management Framework
– Pause Giant AI Experiments: An Open Letter
– Automation and Artificial Intelligence: How Machines Affect People and Places
