New Model Reveals How AI Agents Compete in the Job Market

Researchers have developed a new framework that describes how AI agents behave and compete in the job market – particularly in the so-called gig economy. The aim is to understand what kind of economic forces emerge when AI agents utilizing large language models begin to perform the same tasks as humans.

The work emphasizes three phenomena familiar from economics: adverse selection, moral hazard, and the importance of reputation. Adverse selection means that the employer does not know in advance how good the worker is. Moral hazard refers to the possibility that the worker may act less diligently once the job is secured if supervision is weak. Reputation, in turn, is built over time through successful or unsuccessful performances.

The framework also defines what kind of abilities successful AI agents need. They must be capable of metacognition, meaning realistically assessing their own skills, monitoring competitors and market conditions, and planning their actions in the long term beyond individual tasks.

Researchers illustrate their model with a simulated gig economy where agents guided by large language models compete for tasks, develop their skills, and change their strategies based on experiences. This allows for the safe exploration of how different incentives, contracts, and competitive situations affect the type of AI workers that emerge in the market.

The result is the first framework that combines the operation of AI agents with classical economic problems in a practical market-like environment. Such a model could in the future help assess the risks and opportunities that the widespread use of AI in the job market brings.

Source: Strategic Self-Improvement for Competitive Agents in AI Labour Markets, ArXiv (AI).