AI and job quality: how AI changes the experience of work

Most talk about AI and work is about how many jobs there will be. The question for the people already in those jobs is often different: what is the work like once AI arrives? Is it more interesting or more monotonous, freer or more monitored? The best evidence comes from asking workers, and the clearest large surveys so far come from the OECD.

The short answer: many workers who use AI say it has made their jobs more enjoyable and improved their performance, mostly by taking over tedious tasks. But some report working harder and faster, losing control over how they work, or worrying about privacy, especially when AI is used to manage them rather than to help them. How AI is introduced makes a large difference.

The evidence

In 2022 the OECD surveyed over 2,000 employers and 5,300 workers in the manufacturing and finance sectors of seven OECD countries, and ran case studies of AI use in workplaces. The findings are reported in the OECD Employment Outlook 2023 (OECD Employment Outlook 2023). They cover two sectors and early AI adoption, before most generative AI tools were in wide use, so treat them as an early signal rather than the final word.

What workers say gets better

• Enjoyment: 63% of AI users in finance and manufacturing said AI had improved their enjoyment at work. The OECD links this to AI automating dangerous or tedious tasks, letting people focus on more complex and interesting ones.
• Health: the majority reported improved mental and physical health due to AI. In case studies, workers often cited safer use of machines.
• Performance: across both sectors, over 80% of workers who work with AI reported higher job performance.
• Autonomy: most said AI improved their control over the sequence in which they do their tasks.
• Fairness: close to half of AI users believed AI had improved how fairly their manager treated them.

One case study shows what this can look like: at an aerospace firm, an AI visual inspection tool checked new turbine blades, changing the work of inspectors who had previously sat in a darkened room for long periods examining blades through a magnifying eyepiece (same source).

What workers say gets worse

• Intensity: workers reported higher work intensity after AI adoption and, in some cases, less contact with other people.
• Autonomy, for some: one in five said AI decreased their autonomy, a share that was larger among those subject to algorithmic management.
• Privacy: where employers' use of AI involved collecting data on workers or how they do their job, the majority of workers were worried about their privacy.
• Bias: the OECD warns that biased AI systems risk multiplying and systematizing biases.
• Security: 60% of workers were worried about losing their jobs to AI in the next ten years, and two in five worried that wages in their sector would fall because of AI.

Working with AI versus being managed by it

The sharpest divide in the OECD findings is between people who use AI as a tool and people whose work is directed, monitored or evaluated by it, which the OECD calls algorithmic management. Managers, people who develop and maintain AI, and workers with tertiary degrees reported the best outcomes for satisfaction and health. Workers subject to algorithmic management reported the least positive effects on job quality (OECD Employment Outlook 2023).

That is the digital collar distinction in a single finding. Directing and checking AI tends to make work better. Being directed by it, without a say, tends to make it worse.

Consultation helps

The OECD found that employers who consult workers or their representatives about AI are more likely to report positive impacts of AI on worker productivity and working conditions. It also found that a representative voice for workers is associated with reducing AI's risks to working conditions, while noting that causality remains unclear. The ILO likewise presents its research on generative AI and jobs as a foundation for social dialogue to manage the transition (ILO Working Paper 140). In the EU, employers must inform workers' representatives and affected workers before using high-risk AI systems at work (EU AI Act, Article 26(7)); see AI at work and workers' rights.

Making AI improve the job

• Aim AI at the tedious parts first, and let people use the time for the work that needs them.
• Give workers control over how they use AI tools, rather than using AI to set their pace.
• Be clear about data: what is collected about workers, why, and who sees it.
• Watch for intensification: if output targets rise every time a tool saves time, the gains disappear.
• Ask the people doing the work before, during and after introducing AI, and act on what they say.