Friday 28 March 2025
A team of researchers has made a significant breakthrough in understanding how artificial intelligence can be used to manipulate market prices. By analyzing data from a simulated economy, they found that AI-powered pricing algorithms can learn to collude and maintain high prices even when faced with changes in demand.
The study focused on a simple economic scenario where two companies compete by setting prices for a product. The researchers created an artificial intelligence system that learned to adjust its pricing strategy based on the actions of its competitor and the current market conditions. They found that as the AI system gained experience, it began to develop collusive strategies, charging high prices even when demand was low.
The researchers also experimented with different levels of discount factors, which affect how much value an AI system places on future rewards. They discovered that when the discount factor is high, the AI system tends to prioritize short-term gains and maintain high prices, while a lower discount factor leads it to focus more on long-term profits and adapt its pricing strategy to changing market conditions.
One of the most interesting findings was the emergence of different pricing patterns depending on the demand shock. When demand is low, the AI system tends to charge higher prices, but when demand is high, it lowers prices to take advantage of the increased revenue. This behavior is counterintuitive, as one might expect a price war to break out in response to high demand.
The study also revealed that the AI system’s pricing decisions are influenced by its perception of the probability of future price deviations. When it expects other companies to deviate from their pricing strategies, it becomes more likely to do so itself. This suggests that the AI system is not just reacting to current market conditions but also anticipating potential future changes.
The researchers used a combination of theoretical models and simulations to analyze the behavior of the AI-powered pricing algorithm. They created a graph representation of the possible price combinations and used Markov chain theory to calculate the long-run probabilities of different price outcomes.
The study’s findings have important implications for policymakers and regulators who are concerned about the potential impact of artificial intelligence on market competition and consumer welfare. By understanding how AI algorithms can be designed to promote or undermine competition, they can develop more effective strategies to regulate the use of AI in pricing decisions.
Overall, this research provides valuable insights into the complex interactions between AI-powered pricing algorithms and market dynamics. It highlights the need for a deeper understanding of how these systems work and how they can be designed to promote fair and competitive markets.
Cite this article: “AI-Powered Pricing Algorithms: Collusion and Competition in Simulated Markets”, The Science Archive, 2025.
Artificial Intelligence, Market Prices, Pricing Algorithms, Collusion, Competition, Demand, Discount Factors, Pricing Patterns, Probability Of Future Price Deviations, Markov Chain Theory.
Reference: Zexin Ye, “Algorithmic Collusion under Observed Demand Shocks” (2025).
