AI automation paradox: Workers face more complex responsibilities for lower pay, study warns

AI Adoption Creates Unexpected Workplace Burdens

A new report published in the Society of Occupational Medicine's journal Occupational Medicine challenges the assumption that AI adoption will lighten workplace demands 1

1

. Researchers from Imperial College London and Microsoft warn that the AI automation paradox is emerging: rather than reducing workloads, AI is transforming human roles in ways that may increase stress levels and create novel occupational hazards 1

1

. The real impact on the workforce won't be mass job displacement, but a fundamental shift in how work gets done—and who bears the burden 2

2

.

Overseeing AI Systems Becomes the New Job

As AI absorbs routine tasks, human roles are evolving from performing work to stewarding AI agents across workflows. This includes briefing AI systems, reviewing outputs, and correcting AI-generated errors—essentially transforming workers into managers of artificial intelligence 1

1

. Dr Lara Shemtob, who led the research, explains that "as AI absorbs routine tasks, human roles may shift toward stewardship, problem-solving, or emotional labor, all with their own psychological demands" 1

1

. This shift creates more complex responsibilities that not everyone is suited for, introducing mental health pressures that weren't previously part of job descriptions 1

1

.

Hidden Workloads Negate Automation Benefits

The report warns that AI may "paradoxically increase the knowledge worker's burden of handling complex tasks while simultaneously exerting downward pressure on compensation" 1

1

. This means increased responsibilities for lower pay, because AI supposedly makes work "easier" 1

1

. Evidence already supports this concern. A 2024 study found AI coding tools actually slowed developers down due to time spent checking and correcting errors that wouldn't have existed otherwise 1

1

2

2

. Researchers describe these as hidden workloads that negate the benefit of automating outsourcing tasks 2

2

.

AI Supervision Demands Require Urgent Attention

The research calls for quantifying AI supervision demands and building them into job descriptions to avoid hidden workloads 1

1

. As AI systems become more autonomous, problems like "hallucinations"—false or inaccurate outputs—may escalate and become harder to detect 1

1

. The report emphasizes finding routes to manage where humans sit "in the human-in-the-loop" 2

2

. Poor management may lead to productivity gaps as well as health and unemployment consequences across industries, age groups, and regions 2

2

. A previous MIT study found that around one in nine US jobs could be replaced by artificial intelligence, while Bank of England and ONS data shows youth unemployment rising across the UK 2

2

.

Deployment Complexity Challenges Remain Uncertain

Whether this scenario fully materializes remains uncertain. Recent reports show companies have invested tens of billions in generative AI with little return, and many projects fail due to underestimated deployment complexity 1

1

. Researchers don't yet know the exact impact on human employees from having to work more closely with AI, but they say occupational health should be part of the dialog and analysis of how AI changes expectations of workers 1

1

. The report concludes that greater care should be given to how we handle this transitionary period to ensure that humans continue to play a sustainable role in productivity 2

2

. The question isn't just how AI will change work—it's whether widespread adoption will deliver the promised automation benefits at all 1

1

.