Company

The origins of LLM Council

The concept behind LLM Council, that structured peer deliberation produces superior outcomes compared to single-source judgment, is grounded in over a decade of empirical research.

2013

The journey began at Arizona State University’s W. P. Carey School of Business, where research with Dr. Timothy J. Richards established a foundational finding: peer networks are three times more influential than anonymous single-source reviews in driving consumer preferences.

This research, published as “Social Networks and Restaurant Ratings” in Agribusiness in 2016, demonstrated that when humans make complex decisions under uncertainty, structured peer feedback consistently outperforms anonymous, single-source judgment, even after controlling for endogeneity.

Peer review superiority: structured feedback from multiple peers produces better decisions than single-source evaluation
Negative feedback impact: critical reviews have greater influence than positive reviews
Network diversity: diverse perspectives improve collective judgment
Endogeneity control: the effect holds even after accounting for selection bias

2015

Research with Dr. Yueming (Lucy) Qiu, published in Energy Efficiency, showed that peer network effects are not limited to human-to-human influence. They extend to non-living entities.

The paper “The Diffusion of Voluntary Green Building Certification: A Spatial Approach” demonstrated that buildings influence other buildings through certification diffusion.

This established a universal principle: human networks influence decisions, building networks show peer effects, and synthetic networks can let language models influence other language models through structured deliberation.

Publications

Tiwari, A. (2013). “Anonymous Social Networks versus Peer Networks in Restaurant Choice.” Arizona State University.

Qiu, Y. L. & Tiwari, A. (2013). “Voluntary Green Building Certification: Economic Decision or Following the Trend? A Spatial Approach.” 32nd USAEE/IAEE North American Conference.

Qiu, Y., Tiwari, A. & Wang, Y. D. (2015). “The Diffusion of Voluntary Green Building Certification: A Spatial Approach.” Energy Efficiency, 8(3), 449-471. Cited 39 times.

Tiwari, A. & Richards, T. J. (2016). “Social Networks and Restaurant Ratings.” Agribusiness, 32(2), 153-174. Cited 16 times.

2019-2021

At Queensland University of Technology, the research focus shifted to language models: how to take fragmented, noisy information from multiple sources and synthesize it into coherent, meaningful insights.

The language model research explored ELMo, BERT, GPT-2, and GPT-3; used attention mechanisms for signal weighting; coverage mechanisms for completeness; and multi-model quality assessment with methods such as BERTScore, BLEURT, and OpenMEVA.

The bridge insight is direct: fragmented social media signals become fragmented model perspectives; attention weights become peer-review scoring; coverage becomes consensus and dissent preservation; coherent synthesis becomes the chairman answer.

2023-2025

The broader AI field then validated the same direction through multi-agent debate, ChatEval, panels of LLM evaluators, Language Model Council research, the “Wisdom of the Silicon Crowd,” and work on how LLMs reshape collective intelligence.

LLM Council productionizes the principle: structured deliberation across multiple models can improve reliability, factuality, reasoning, and user trust compared with single-model judgment.

Attribution

We welcome contributions to the field. The intellectual foundations for LLM Council were established through research beginning in 2013, proven extensible in 2015, applied to language models in 2019-2021, and developed into a production platform in 2025.

Collaborators

Research collaborators include Dr. Timothy J. Richards, Dr. Yueming (Lucy) Qiu, Dr. Alan Woodley, and Dr. Richi Nayak.