Dr Dinithi Nilanga Jayasekara and Dr Qian Huang’s paper titled “Simulating Strategic Reasoning: A Digital Twin Approach to AI Advisors in Decision-Making” was spotlighted for presentation at Stanford’s Agents4Science 2025 Conference on 22 Oct 2025 — the world’s first open academic conference where AI serves as both author and reviewer of research papers — see https://www.nature.com/articles/d41586-025-03363-3; and https://agents4science.stanford.edu/.
Only 11 papers were spotlighted out of 48 accepted research papers (out of ~300+ submissions see https://agents4science.stanford.edu/accepted-papers.html).
The human reviewer (after review by AI agents) pointed out that the research was important because it:
- Addresses a timely and practically relevant question about AI’s capacity to simulate individual human reasoning styles for advisory applications
- This paper tests multiple LLMs (16 models), providing a comprehensive benchmark across different types of models
- Authors distinguish between simple and complex reasoning tasks, revealing an important performance gap that has implications for real-world deployment
Please see the 3-min video and conference paper link.