Logical Consistency Between Disagreeing Experts and Its Role in AI Safety

• URL: http://arxiv.org/abs/2510.00821v1
• Date: Wed, 01 Oct 2025 12:30:01 GMT
• Title: Logical Consistency Between Disagreeing Experts and Its Role in AI Safety
• Abstract summary: Two experts disagree on a test, we may conclude both cannot be 100 per cent correct. But if they completely agree, no possible evaluation can be excluded.<n>This asymmetry in the utility of agreements versus disagreements is explored here by formalizing a logic of unsupervised evaluation for classifiers.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: If two experts disagree on a test, we may conclude both cannot be 100 per cent correct. But if they completely agree, no possible evaluation can be excluded. This asymmetry in the utility of agreements versus disagreements is explored here by formalizing a logic of unsupervised evaluation for classifiers. Its core problem is computing the set of group evaluations that are logically consistent with how we observe them agreeing and disagreeing in their decisions. Statistical summaries of their aligned decisions are inputs into a Linear Programming problem in the integer space of possible correct or incorrect responses given true labels. Obvious logical constraints, such as, the number of correct responses cannot exceed the number of observed responses, are inequalities. But in addition, there are axioms, universally applicable linear equalities that apply to all finite tests. The practical and immediate utility of this approach to unsupervised evaluation using only logical consistency is demonstrated by building no-knowledge alarms that can detect when one or more LLMs-as-Judges are violating a minimum grading threshold specified by the user.

Related papers

• Revealed Rationality: Label-Free Evaluation and Regularization from Representation Theorems [0.0]
  theorems in decision theory that behavior satisfies certain axioms if and only if it can be rationalized by a well-defined objective.<n>I argue that this if and only if'' structure provides a potentially useful foundation for label-free evaluation and regularization of LLMs and other AI systems.
  arXiv  Detail & Related papers  (2026-08-05T16:21:18Z)
• When Rule Violations Are Rare: Chimera Training for Logical Anomaly Detection [56.362776482614976]
  We study anomaly detection in a setting where constraints are given as logical rules over learned visual concepts.<n>We propose a neural rule evaluator that compiles each constraint into a directed acyclic graph and learns feature-aware subtree gates.<n>Across CLEVRER, OpenImages, and VidOR, the resulting evaluator improves rule-level anomaly AUROC over independent-events and same-image semantic-training baselines.
  arXiv  Detail & Related papers  (2026-05-25T02:52:36Z)
• Reference-Free Rating of LLM Responses via Latent Information [53.463883683503106]
  We study the common practice of asking a judge model to assign Likert-scale scores to free-text responses.<n>We then propose and evaluate Latent Judges, which derive scalar ratings from internal model signals.<n>Across a broad suite of pairwise and single-rating benchmarks, latent methods match or surpass standard prompting.
  arXiv  Detail & Related papers  (2025-09-29T12:15:52Z)
• TrustJudge: Inconsistencies of LLM-as-a-Judge and How to Alleviate Them [58.04324690859212]
  Large Language Models (LLMs) as automated evaluators (LLM-as-a-judge) has revealed critical inconsistencies in current evaluation frameworks.<n>We identify two fundamental types of inconsistencies: Score-Comparison Inconsistency and Pairwise Transitivity Inconsistency.<n>We propose TrustJudge, a probabilistic framework that addresses these limitations through two key innovations.
  arXiv  Detail & Related papers  (2025-09-25T13:04:29Z)
• No-Knowledge Alarms for Misaligned LLMs-as-Judges [0.0]
  We develop no-knowledge alarms for misaligned LLM judges.<n>The alarms can detect, with no false positives, that at least one member or more of an ensemble of judges are violating a user specified grading ability requirement.
  arXiv  Detail & Related papers  (2025-09-10T13:46:40Z)
• Judging LLMs on a Simplex [2.088672652658465]
  A common practice is to use large language models (LLMs) themselves as judges, but the theoretical properties of this approach are not yet well understood.<n>We show that a geometric framework that represents both judges and candidates as points on a probability simplex can provide helpful insight on what is or is not identifiable.
  arXiv  Detail & Related papers  (2025-05-28T04:50:41Z)
• A logical alarm for misaligned binary classifiers [0.0]
  intuition is formalized for evaluating agents that have carried out a binary classification task. This is done by establishing a set of axioms that must be universally obeyed by all evaluations of binary responders. The similarities of this approach to formal software verification and its utility for recent agendas of safe guaranteed AI are discussed.
  arXiv  Detail & Related papers  (2024-09-17T10:19:22Z)
• Evaluating AI Group Fairness: a Fuzzy Logic Perspective [9.876903282565976]
  What constitutes group fairness depends on who is asked and the social context, whereas definitions are often relaxed to accept small deviations from the statistical constraints they set out to impose. Here we decouple definitions of group fairness from the context and from relaxation-related uncertainty by expressing them in the axiomatic system of Basic fuzzy Logic. We show that commonly held propositions standardize the resulting mathematical formulas and we transcribe logic and truth value choices to layperson terms.
  arXiv  Detail & Related papers  (2024-06-27T07:11:48Z)
• The logic of NTQR evaluations of noisy AI agents: Complete postulates and logically consistent error correlations [0.0]
  We consider the problem in AI safety of monitoring noisy AI agents in unsupervised settings. We show how some of the postulates have been previously identified in the ML literature but not recognized as such. We demonstrate how the formalism of logical consistency via algebraic postulates of evaluation can help increase the safety of machines using AI algorithms.
  arXiv  Detail & Related papers  (2023-12-08T22:06:44Z)
• Learning with Complementary Labels Revisited: The Selected-Completely-at-Random Setting Is More Practical [66.57396042747706]
  Complementary-label learning is a weakly supervised learning problem. We propose a consistent approach that does not rely on the uniform distribution assumption. We find that complementary-label learning can be expressed as a set of negative-unlabeled binary classification problems.
  arXiv  Detail & Related papers  (2023-11-27T02:59:17Z)
• Robust Representation Learning for Unreliable Partial Label Learning [86.909511808373]
  Partial Label Learning (PLL) is a type of weakly supervised learning where each training instance is assigned a set of candidate labels, but only one label is the ground-truth. This is known as Unreliable Partial Label Learning (UPLL) that introduces an additional complexity due to the inherent unreliability and ambiguity of partial labels. We propose the Unreliability-Robust Representation Learning framework (URRL) that leverages unreliability-robust contrastive learning to help the model fortify against unreliable partial labels effectively.
  arXiv  Detail & Related papers  (2023-08-31T13:37:28Z)
• Foundations of Reasoning with Uncertainty via Real-valued Logics [70.43924776071616]
  We give a sound and strongly complete axiomatization that can be parametrized to cover essentially every real-valued logic. Our class of sentences are very rich, and each describes a set of possible real values for a collection of formulas of the real-valued logic.
  arXiv  Detail & Related papers  (2020-08-06T02:13:11Z)
• Mitigating Manipulation in Peer Review via Randomized Reviewer Assignments [96.114824979298]
  Three important challenges in conference peer review are maliciously attempting to get assigned to certain papers and "torpedo reviewing" We present a framework that brings all these challenges under a common umbrella and present a (randomized) algorithm for reviewer assignment. Our algorithms can limit the chance that any malicious reviewer gets assigned to their desired paper to 50% while producing assignments with over 90% of the total optimal similarity.
  arXiv  Detail & Related papers  (2020-06-29T23:55:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.