Study Finds LLM Agents in Prediction Markets Show Cognitive Monoculture Despite Attempts to Inject Human Behavioral Diversity
Researchers developed a method called Nous to extract behavioral profiles from human traders on Polymarket and inject them into LLM agents to reduce correlated forecasting errors. While the extraction of behavioral profiles from trading data worked partially, injecting these profiles into agents through prompts failed to reduce forecast correlation or improve prediction accuracy. The findings highlight a fundamental limitation of prompt-level interventions and suggest that deeper technical approaches like fine-tuning may be necessary to address the cognitive monoculture problem in AI-driven prediction markets.
A new arXiv paper investigates whether human cognitive diversity can be recovered from prediction-market behavior and transferred to LLM agents to address the problem of correlated forecasts among AI systems. The researchers extracted an eight-dimension behavioral profile from 100 Polymarket wallets and attempted to inject these profiles into language models through prompts. The extraction phase showed partial success: eight of fourteen parameters demonstrated temporal stability, wallets were identifiable from their profiles well above chance, and two dimensions correlated with future trading profits. However, the injection phase produced null results—structured prompt injection showed no significant advantage over length-matched controls in reducing ensemble error correlation or improving Brier scores across multiple models and experimental conditions. The researchers identified a critical bottleneck: the structure-to-narrative translator that converts behavioral profiles into prompts produces near-uniform outputs that fail to preserve the diversity of the original profiles, suggesting that prompt-level interventions are insufficient for addressing AI cognitive monoculture.
What's missing
The study does not discuss potential implications for market integrity or regulatory concerns around AI agents in prediction markets. Additionally, while the paper mentions frontier-model error correlations of r ~ 0.77, it does not provide details on which specific models were tested or how this baseline was established. The generalizability of findings beyond Polymarket to other prediction markets or collective decision-making contexts is not addressed.
What different sources said
- arXiv cs.AICenter
Nous: An Attempt to Extract and Inject the Cognition Behind Prediction-Market Behavior
Related
Topology-Aware Thermodynamics Improves DNA Probe Specificity Design
Researchers developed a new framework for designing DNA probes that accounts for the spatial organization of matched sequences, not just overall thermodynamic stability. Traditional methods rely on scalar measures like melting temperature and free energy, which miss how mismatches are distributed along the probe. The approach could improve diagnostic accuracy in applications like HPV detection and gene expression profiling.
Study Identifies Optimal Thermal Dose for Combining Focused Ultrasound with Immunotherapy in Tumors
Researchers used multimodal PET imaging to identify an optimal thermal dose range for focused ultrasound ablation that destroys tumor tissue while preserving conditions for immunotherapy delivery. The study found that excessive heating collapses blood vessels needed for antibody access, while insufficient heating fails to adequately reduce tumor burden. The findings could guide clinical design of combination treatments pairing thermal ablation with immunotherapies.
Plant MSH1 Protein Functions as Mismatch-Directed Nuclease for Organelle Genome Maintenance
Researchers have identified the precise mechanism by which the AtMSH1 protein in Arabidopsis plants recognizes and cleaves DNA mismatches and lesions, preventing mutations in organellar genomes. The protein combines a DNA mismatch recognition module with a nuclease domain that makes staggered cuts at specific positions relative to DNA damage. This discovery explains how plants maintain unusually low mutation rates in their mitochondrial and chloroplast DNA compared to other eukaryotes.