How an AI doom scenario could gradually erode human skills, influence and control
A gradual loss of control
The term “gradual disempowerment” comes from a paper published in early 2025. Experts say the concern is not limited to job losses: people may also lose knowledge and critical-thinking skills as they hand more tasks and decisions to AI.
Chooi described the danger as a civilization-scale problem and said it is as concerning as engineered pandemics and more destructive cyberattacks that capable AI could accelerate. Chooi also said many people feel powerless because responding will take substantial effort and society-wide action.
AI competes across work
AI models may become better than humans at many thinking tasks, including mathematics, science, coding, writing and creating art or music. Advances in robotics could also put machines to work in physical jobs, from construction and mining to packing boxes.
The article imagines a company using AI agents and robots to make products faster and at lower cost than human workers. If that pattern spread across industries, businesses that kept human employees could face pressure from competitors that automated more of their work.
Dependence on automated decisions
As AI takes on complex decisions, people may become less able to understand or oversee the systems, or challenge their choices. One concern raised in the article is what happens when an automated system fails and no person can fix the problem.
The risk extends beyond businesses. Although humans still make most or all decisions in war today, AI’s growing capabilities could shift that balance, the article says.
Citizens’ leverage at risk
Jan Kulveit, a co-author of the paper and co-founder of ACS Research Group, said states depend on citizens for taxes and military service. If governments no longer need citizens in those ways, he said, they may be less likely to remain democratic.
The article says people could lose leverage over those in power if they are no longer central to work or warfare. Kulveit and his colleagues wrote that gradual disempowerment need not result from a deliberate plan by AI systems; economic, cultural and state incentives could help drive it.
Key points
- Experts warn that AI could gradually erode people’s skills, influence and oversight.
- Companies may face pressure to replace human workers with cheaper AI and robots.
- Greater reliance on AI could leave people unable to understand or fix system failures.
- Jan Kulveit warns that states may be less democratic if they no longer need citizens.