Pith. sign in

REVIEW 6 cited by

Should we be going MAD? A Look at Multi-Agent Debate Strategies for LLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2311.17371 v3 pith:ANJLWVU3 submitted 2023-11-29 cs.CL cs.AI

classification cs.CLcs.AI
keywords strategiesdebatingllmsmulti-agentotherprotocolsbenchmarkdebate
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Recent advancements in large language models (LLMs) underscore their potential for responding to inquiries in various domains. However, ensuring that generative agents provide accurate and reliable answers remains an ongoing challenge. In this context, multi-agent debate (MAD) has emerged as a promising strategy for enhancing the truthfulness of LLMs. We benchmark a range of debating and prompting strategies to explore the trade-offs between cost, time, and accuracy. Importantly, we find that multi-agent debating systems, in their current form, do not reliably outperform other proposed prompting strategies, such as self-consistency and ensembling using multiple reasoning paths. However, when performing hyperparameter tuning, several MAD systems, such as Multi-Persona, perform better. This suggests that MAD protocols might not be inherently worse than other approaches, but that they are more sensitive to different hyperparameter settings and difficult to optimize. We build on these results to offer insights into improving debating strategies, such as adjusting agent agreement levels, which can significantly enhance performance and even surpass all other non-debate protocols we evaluated. We provide an open-source repository to the community with several state-of-the-art protocols together with evaluation scripts to benchmark across popular research datasets.

Discussion (0). Sign in to comment.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. MARGIN: Runtime Confidence Calibration for Multi-Agent Foundation Model Coordination

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    MARGIN is an online per-agent per-band calibration method using symmetric exponentially weighted moving averages with Bayesian shrinkage that reduces calibration error 3-6x under distribution shift and improves multi-...

  2. Multi-Agent LLMs Fail to Explore Each Other

    cs.MA 2026-07 conditional novelty 6.5 of 10

    Modern multi-agent LLM systems fail to explore peers effectively; explicit LinUCB-style peer selection (MACE) cuts regret and lifts task performance, with gains scaling in agent diversity.

  3. Beyond a Joke: Multi-Angle Reasoning for Detecting and Explaining Harmful Humor in Memes

    cs.AI 2026-07 conditional novelty 6.0 of 10

    MAR-12 improves humor and hate detection in memes by prompting a VLM through twelve reasoning perspectives, attention-weighting them, and generating explanations from the weighted evidence.

  4. When Agents Disagree: The Selection Bottleneck in Multi-Agent LLM Pipelines

    cs.MA 2026-03 conditional novelty 6.0 of 10

    In a 42-task controlled comparison, selecting the best candidate with judge panels beats MoA-style synthesis in every task, and a crossover threshold explains when team diversity helps.

  5. How to Train a Leader: Hierarchical Reasoning in Multi-Agent LLMs

    cs.MA 2025-07 conditional novelty 5.0 of 10

    A leader LLM trained with a GRPO variant that conditions on frozen agent responses improves both collaborative and zero-shot accuracy on BBH, MATH, and MMLU.

  6. Inverse Reinforcement Learning Meets Large Language Model Post-Training: Basics, Advances, and Opportunities

    cs.LG 2025-07 unverdicted novelty 1.0 of 10

    A tutorial reviewing LLM alignment through the lens of inverse reinforcement learning, arguing that neural reward models learned from human data are central to post-training.

Pith tools