TL;DR
A report led by researchers from Imperial College London and Microsoft warns that AI may shift work from routine tasks to supervising autonomous systems, increasing cognitive and emotional burdens. The paper urges measuring supervision demands and including them in job descriptions, while evidence and investment trends raise questions about whether automation will reduce overall worker workload or pay.
What happened
Researchers writing in the Society of Occupational Medicine's journal Occupational Medicine, led by Dr Lara Shemtob, say the primary effect of AI adoption may not be mass job losses but a redefinition of many roles. As AI takes over routine activities, humans are likely to become stewards of agentic systems—briefing models, checking outputs, correcting mistakes and managing integration across workflows. That shift could replace manual drudgery with oversight, problem-solving and emotional labor, all of which carry psychological costs. The report cites prior evidence that AI coding tools slowed developers because of time spent verifying and fixing generated code, and warns that increasing autonomy can raise risks such as hallucinations—false or inaccurate outputs that are difficult to spot. Authors recommend quantifying the supervision workload and putting it into job descriptions to avoid hidden burdens. They also note many AI projects cost firms tens of billions with limited returns, leaving the pace and extent of adoption uncertain.
Why it matters
- Work could become more mentally demanding even as routine tasks are automated, raising stress and burnout risks.
- Employers may increase worker responsibility for system oversight without adjusting pay or job expectations.
- Unaccounted supervision time could negate efficiency gains promised by automation.
- Wider rollout risks depend on real-world deployment complexity and uncertain returns on AI investments.
Key facts
- The report appears in the Society of Occupational Medicine's journal Occupational Medicine.
- Research team included contributors from Imperial College London and Microsoft; Dr Lara Shemtob led the work.
- Authors predict a shift from task execution to stewardship: briefing AI, reviewing outputs and correcting errors.
- A 2024 study is cited showing AI coding tools sometimes slowed developers because of verification and correction work.
- The paper highlights 'hallucinations'—false or inaccurate AI outputs—as a growing supervision hazard.
- Authors urge employers to quantify AI supervision demands and include them in job descriptions to avoid hidden workloads.
- The report notes companies have invested tens of billions in generative AI with limited returns and many failed projects.
What to watch next
- Research quantifying the time and cognitive load required to supervise agentic AI systems.
- Whether employers update job descriptions and staffing models to include AI supervision duties.
- Corporate reporting on AI project returns and evidence of reduced or increased worker workload over time.
Quick glossary
- Agentic AI: AI systems capable of autonomous actions in a workflow, such as generating outputs, making recommendations or initiating tasks.
- Hallucination: When an AI system produces an output that is false, fabricated or not supported by its training data.
- Occupational health: The field concerned with the physical and mental well-being of people in the workplace.
- Stewardship (in AI context): Human responsibilities that involve overseeing, validating and correcting the outputs and behaviors of AI systems.
Reader FAQ
Will AI replace most jobs?
The report argues the likely effect is role change toward AI supervision rather than wholesale job elimination.
Will AI reduce workers' workloads?
Not necessarily—researchers warn automation can create hidden supervision tasks that may offset time savings.
Are pay cuts for supervised roles confirmed?
The paper warns of downward pressure on compensation but does not offer empirical proof of widespread pay cuts; not confirmed in the source.
What should employers do now?
The authors recommend measuring supervision demands and incorporating them into job descriptions to avoid hidden burdens.
AI + ML AI's grand promise: Less drudgery, more complexity, same (or lower) pay Workers face new mental health pressures as they shift from doing tasks to babysitting agentic AI…
Sources
- AI's grand promise: Less drudgery, more complexity, same (or lower) pay
- AI has promise to reshape payroll, but it still needs human …
- How AI Will Rewire Organizational Structure and the Future …
- AI's revenue promise – and the hidden cost of inefficient …
Related posts
- YouTube Music adds AI-generated backgrounds to lyrics sharing
- Why AI Agents Didn’t Join the Workforce as Promised in 2025: A Postmortem
- Scientific production in the era of large language models — PDF release