analiza aiqbench
This research addresses the critical limitation of current LLM simulators, which often flatten population heterogeneity into average responses rather than modeling distinct individuals. By introducing Anacreon—a system that clusters users and trains dedicated adapters on a Gemma 12B base—the authors achieve state-of-the-art individual-level accuracy while mitigating prompt brittleness and social bias through specific training distributions. This advancement represents a significant step forward in Machine Learning for creating faithful population simulators capable of capturing nuanced human responses without inheriting the central tendency errors of standard statistical or generative methods.
Mind the Gaps: Mixture-of-Minds for Human Simulation
arXiv:2608.06115v1 Announce Type: new Abstract: Predicting how a population will answer a new question is a long-standing goal. Statistical methods succeed at the level of the mass but falter at the level of the individual. Large language model simulators inherit this gap. They recover a population's central tendencies while flattening its heterogeneity, and they carry social biases and prompt brittleness that distort individual predictions. This paper introduces Anacreon, an audience simulation model that targets the individual level within a narrow, well-specified domain. Anacreon learns an authorship embedding that separates individuals, clusters a real qualitative corpus around seed people, and trains a dedicated adapter for each cluster, a mixture of minds, on a Gemma~4 12B base. It harvests demographics, psychological traits, and survey responses from public text, and augments each record with a chain-of-emotion. It reduces prompt brittleness by shuffling response options and reduces positive bias by balancing the training distribution. On a large, externally sourced survey, Anacreon reaches a state-of-the-art ordinal alignment of 0.775, the individual-level accuracy measure on which the field has converged, with a small residual bias. The work is a step toward drawing aggregate insight from faithfully simulated individuals.
Abstract:Predicting how a population will answer a new question is a long-standing goal. Statistical methods succeed at the level of the mass but falter at the level of the individual. Large language model simulators inherit this gap. They recover a population's central tendencies while flattening its heterogeneity, and they carry social biases and prompt brittleness that distort individual predictions. This paper introduces Anacreon, an audience simulation model that targets the individual level within a narrow, well-specified domain. Anacreon learns an authorship embedding that separates individuals, clusters a real qualitative corpus around seed people, and trains a dedicated adapter for each cluster, a mixture of minds, on a Gemma~4 12B base. It harvests demographics, psychological traits, and survey responses from public text, and augments each record with a chain-of-emotion. It reduces prompt brittleness by shuffling response options and reduces positive bias by balancing the training distribution. On a large, externally sourced survey, Anacreon reaches a state-of-the-art ordinal alignment of 0.775, the individual-level accuracy measure on which the field has converged, with a small residual bias. The work is a step toward drawing aggregate insight from faithfully simulated individuals.
Artificial Intelligence (cs.AI)
arXiv:2608.06115 [cs.AI]
(or arXiv:2608.06115v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2608.06115 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 281 cuvinte
Subiecte conexe
AXRP Episode 50 - Eli Lifland on AI 2027
LessWrong · 3 aug. 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · 13 oct. 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · 25 sept. 2026
[AINews] not much happened today
Latent Space · 10 sept. 2026