AI productivity gains can fuel a workplace rat race, UCLA working paper suggests

AI can help every employee produce more without helping anyone get ahead. A UCLA Anderson working paper, described November 11 by UCLA Anderson Review, suggests that rewards tied to beating colleagues can turn productivity gains into greater effort and burnout risk. Changing those incentives may offer more relief than wellness spending alone.

The paper, “The rat race at work: Technology and burnout,” is by UCLA Anderson’s Auyon Siddiq. It uses game theory—a way of examining choices when each person’s outcome depends on what others do. This is a theoretical explanation of how workplace incentives might operate, not a measurement of burnout caused by an AI rollout.

When everyone’s output rises, so does the bar

The model starts with employees choosing how hard to work. Better technology makes each hour more productive, while greater exertion raises their risk of burning out. Part of their pay depends on their performance relative to colleagues. That combination gives workers a reason to push harder even when the tools themselves make tasks easier.

When AI lifts everyone’s productivity, it also raises the standard needed to climb the rankings. Employees increase their effort to keep pace. But if everyone pushes equally hard, nobody gains a relative advantage. The employer gets more output; workers face a harder grind. Collectively, they would be better off working less, but individual incentives keep the competition going.

A formal ranking system is not required. Siddiq observes that scarce raises, bonuses and promotions can create similar pressure: employees know that advancing may mean outperforming someone else. His model also allows a different response as the bar keeps rising. Workers may pace themselves while others overextend and drop out. Surviving that competition can eventually ease the pressure.

Better conditions can invite tougher expectations

The next step concerns management’s response to healthier working conditions. In the model, improvements such as flexible hours initially reduce burnout risk. Employers can then treat that resilience as room to increase the stakes attached to performance rankings. Where the underlying burnout risk was already low, that response can more than erase the original benefit, leaving workers worse off.

Siddiq explores the tension by comparing two kinds of employer: one focused on profit and another that also values employee well-being. When technology produces relatively modest productivity gains, the model separates spending on prevention from success at preventing burnout.

Big productivity gains test the gentler approach

That comparison changes when AI delivers large gains. Each hour now produces much more, making reduced work intensity more costly for the employer. Even the employee-minded firm increasingly turns to prevention rather than easing pressure. Under those conditions, it may spend more on burnout prevention than the profit-focused firm. A generous wellness budget therefore need not signal a less demanding workplace.

Siddiq’s proposed changes target the reward system itself. As UCLA Anderson Review explains, companies adopting high-quality technology could:

• Relax quotas requiring managers to classify a fixed share of employees as low performers.
• Give more weight to other measures of performance, such as collaboration.
• Pair better working conditions with a commitment not to raise performance targets.