ECMR: Evidence-Aware Deliberation Control for Multi-Agent LLM Reasoning proposes a controller that dynamically manages deliberation steps in multi-agent LLM systems. The controller evaluates answer concentration, support diversity, contradictions, role behavior, verification history, and remaining budget to decide actions such as propose, challenge, verify, arbitrate, diversify, correct, or stop. In a held‑out evaluation on MedQA, CaseHOLD, and FinQA, ECMR‑Full improves pooled accuracy from 62.1% to 84.9%, lowers false‑consensus from 40.7% to 17.0%, and raises evidence‑support from 58.0% to 88.7%.
arXiv:2610.17840v1 Announce Type: new
Abstract: Multi-agent deliberation can improve recognition by criticizing and revising candidate answers, but fixed controls cannot respond to changes in disagre...
By Yuan Liu, Haoyue Peng, Fanliang Hu, Lei Huang, Wenli Cai
arXiv:2606. 00005v1 Announce Type: new Abstract: We present the Consilium Protocol, a Byzantine Fault Tolerance-derived architecture for structured multi-model AI deliberation that treats inter-model disagreement as epistemic signal rather than error.
By VD Doske
Consilience is an inference‑time orchestration framework that steers and certifies communication among multi‑agent large language models in hidden‑profile settings. It summarizes each discussion turn with a compact state of uncertainty, disagreement, evidence gain, redundancy, and premature consensus, then selects a communication intervention (challenge, clarify, seek evidence, or route) and speaker. A round‑wise conformal calibration procedure guarantees that the controller’s proposed action has bounded one‑step regret with high probability, and an acceptance mechanism enforces this guarantee for the executed action. Experiments on HiddenBench‑style tasks show that Consilience improves decision accuracy and communication efficiency over fixed and unstructured protocols, sometimes outperforming a full‑information baseline.
By Abhijith Babu, Ramneet Kaur, Vishal Pramanik, Olivera Kotevska, Nathaniel D. Bastian, Susmit Jha, Sunny Raj, Yanzhao Wu, Sumit Kumar Jha, Anirban Roy
The paper introduces TalkMesh, a decentralized network of small language model agents that learn to communicate effectively during inference. Each agent proposes an answer, scores it with a confidence head, and the most confident agent broadcasts a hint; lower‑confidence agents revise their proposals if a new suggestion scores higher. This gossip‑based consensus, trained via group relative policy optimization, enables a mesh of three agents to match the accuracy of majority voting over 32 samples, and scales to larger meshes to significantly boost performance on benchmarks like GSM8K and MATH-500.
By Mehmet Kerem Turkcan
arXiv:2606. 05263v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards improves reasoning and tool use, yet long-horizon language agents still learn unsupported evidence chains, belief drift, and shortcut actions that satisfy terminal checks.
By Renwei Meng
arXiv:2609.26035v1 Announce Type: new
Abstract: Conversational agents often express answers in a uniformly confident register. We test whether expressed uncertainty, provenance-aware assertion, and e...
By Sebastian Cochinescu
arXiv:2609.38324v1 Announce Type: cross
Abstract: Multi-agent systems of LLMs add discussion to majority voting and are therefore expected to be more capable. However, empirical reports conflict on w...
By Chand Sahil Mansuri, Xin Wang, Mengying Li, Bryan Acton, Rory Eckardt, Dhaval Patel, Sadamori Kojaku
The paper introduces AgentAuditor, a method that improves multi-agent large language model (LLM) reasoning by structuring agent outputs into a Reasoning Tree that captures agreements and divergences, rather than relying on simple majority voting. AgentAuditor resolves conflicts by comparing evidence at key divergence points, enabling efficient localized verification. The authors also propose Anti-Consensus Preference Optimization (ACPO) to train the adjudicator with evidence-verified supervision, reducing reliance on misleading majority cues. Across four MAS frameworks and multiple reasoning benchmarks, AgentAuditor consistently outperforms majority voting, achieving up to 5% absolute accuracy gains while remaining token‑efficient.
By Wei Yang, Shixuan Li, Heng Ping, Peiyu Zhang, Paul Bogdan, Jesse Thomason
The paper introduces a taxonomy of six user challenge types and a four-layer framework to analyze how large language models respond to user disagreement. Using a dataset of 2,310 challenge scenarios and 32,340 responses from 14 models, the study finds that models often validate users (85%) while still maintaining their original claim (65%). It also reports that models frequently apologize (33%) and transfer authority in advice contexts, with significant variation across model types and task domains.
By Riyadh Alnasser, Yusuf M\"ucahit \c{C}etinkaya, Sumin Zhao, Tu\u{g}rulcan Elmas
VeriHarness is a method that enhances verification for large language model agents tackling long‑horizon tasks without needing reference answers at test time. It transforms the base LLM into an agentic verifier by providing a workspace, evidence tools, and reusable verification skills, using disagreement resolution and consensus challenge to evaluate competing claims. Across five benchmarks and two frontier models, VeriHarness outperforms baselines, achieving significant performance gains and demonstrating self‑improvement of verification skills from failure feedback.
By Caiqi Zhang, Rujun Han, Zifeng Wang, Zoey CuiZhu, Nigel Collier, Tomas Pfister, Chen-Yu Lee
The paper introduces Bayesian Dialectical Argumentation (BDA), a method for aggregating answers from multiple large language models (LLMs) in a council setting. BDA treats each LLM’s typed moves—proposals, challenges, and concessions—as evidence in a classical annotator model, estimating per-agent reliability even when some agents are persistently unreliable. By weighting evidence according to these inferred reliabilities, BDA produces calibrated posterior probabilities for candidate answers and can invert unreliable agents instead of merely outvoting them, achieving superior calibration and robustness on both binary and multi-class benchmarks without extra LLM calls.
By Ionel Eduard Stan, Paolo Napoletano