Research on human and AI strategies presented at the "Game Theory and Management" conference

At Saint Petersburg State University, the traditional international scientific conference "Game Theory and Management" (GTM2026) was held. As part of the conference's Young Researchers Day, Corresponding Member of the Russian Academy of Sciences Alexey Savvateev presented the findings of a study titled "Not Yet! Humans Outplayed LLMs in a Colonel Blotto Game Tournament," conducted within the framework of RSF project No. 25-18-00539 "Comparative Analysis of AI-Based Agents and Real Individuals in Economic Decision-Making."

The Young Researchers Day of the conference was dedicated to engaging students and postgraduate students in science by showcasing meaningful and illustrative applications of game theory. Alexey Savvateev's plenary talk was one example of how classical game-theoretic models can be used to assess the capabilities of artificial intelligence.

The talk presented findings of a study co-authored by Alexey Savvateev, Gleb Vasiliev, Dmitry Dagaev, Egor Ivanov, and Petr Parshakov. The authors compared the strategies of humans and large language models in the Colonel Blotto game – a classical model of strategic allocation of limited resources.

In the study, participants competed against each other in round-robin tournaments. The strategies were framed as an electoral campaign: each player was required to distribute 100 resource units across nine states in order to outperform the opponent by winning the majority of them.

A comparison of strategies from more than 200 human participants and several popular large language models revealed that humans consistently outperform AI in this task. The researchers attribute this to a more precise use of multi-level strategic reasoning: participants not only allocated resources but also attempted to anticipate how their opponent would think. Large language models, by contrast, frequently defaulted to the simplest uniform distribution or made counting errors.

"The Young Researchers Day brought together more than 50 students and postgraduate students interested in game theory. The discussion following Alexey Savvateev's talk was lively and opened up several new research questions," noted Dmitry Dagaev.

The presented study continues the work carried out under RSF project No. 25-18-00539, "Comparative Analysis of AI-Based Agents and Real Individuals in Economic Decision-Making."

Dmitry Dagaev

Senior Research Fellow, International Laboratory of Intangible-driven Economy

"The Young Researchers Day of the conference brought together more than 50 students and postgraduate students interested in game theory. The discussion following Alexey Savvateev's talk was lively and opened up several new research questions."