AI could leave people less able to do their jobs not through a dramatic takeover, but through the gradual loss of skills and decision-making as systems take on more work. A CBS News report explores that risk, and the choices researchers say could help keep people capable and involved.
Watch Desk analysis
What happened
The report describes “gradual disempowerment”: a scenario in which AI takes on more economic work and decision-making, leaving people increasingly dependent on systems they have less ability to oversee. The idea comes from a paper published in early 2025, the report says. Researchers interviewed by CBS News argue that the pressure is not only about jobs: when people stop practising skills, they may lose the ability to recognise or fix problems.
Hugging Face AI research scientist Margaret Mitchell says people who no longer practise work performed by agents can become reliant on those systems. Jan Kulveit, co-author of the paper, says economic and political incentives could encourage organisations to shift more work to AI and robots. The report also discusses possible effects on democratic governance if states rely less on citizens’ labour and participation.
Why it matters
This is a forecast, not evidence that society has already crossed that line. But it reframes a familiar question about automation: not just which tasks AI can do, but what happens to people’s ability to understand and challenge decisions when they no longer do the work themselves.
The researchers’ suggested responses are practical rather than magical. They include tracking what work is being handed to AI, keeping people involved at deliberate points in a process, and designing systems to explain and support human judgement rather than simply deliver answers. Mitchell calls those deliberate pauses “strategic friction”. In a world forever trying to remove friction, that is an unusually useful reason not to smooth every bump away.
Our read
The argument deserves attention because it links everyday automation choices to long-term human capability. It does not prove that using an assistant makes people less capable; the concern is about what happens when systems replace practice and meaningful oversight at scale. Organisations should ask not only whether AI makes a task faster, but whether people can still understand, improve and take responsibility for the result.
What to watch
- Whether organisations measure which work and decisions are being transferred to AI.
- Whether AI tools are designed to build users’ understanding, not just produce answers.
- Whether people retain real authority and the skills to intervene when systems fail.
Discussion spark: When does AI assistance become harmful dependence: only when people lose the ability to do the work, or earlier, when they lose the chance to practise and question it?
Sources and evidence
- Source update (19 October 2026, 15:33 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.