Against All Likelihood
cs.AI
Submitted: 2026-11-06
Updated: 2026-11-06
License: http://creativecommons.org/licenses/by/4.0/
The gist: Maximum likelihood is the dominant training principle for large language models, yet likelihood is neither necessary nor sufficient for truth.
Abstract
Maximum likelihood is the dominant training principle for large language models, yet likelihood is neither necessary nor sufficient for truth. Unlikely events happen every day, and frequently repeated statements can be false. Drawing on a philosophical tradition that ties objectivity to invariance rather than probability (from Nozick's account of objective facts and Wigner's principles of invariance to the structural realism of Poincaré and Cassirer), we argue that the truth of a statement should be judged not by its probability under a single distribution, but by its invariance across relevant transformations of the environment. We connect this view to machine learning literature on out-of-distribution generalization and invariant learning, sketch how invariance could be tested at inference time with language models, and discuss implications for continual learning.
Sources
- On Provable Length and Compositional Generalization
- Invariant Risk Minimization
- Universal Self-Consistency for Large Language Model Generation
- Measuring Faithfulness in Chain-of-Thought Reasoning
- Perplexity Cannot Always Tell Right from Wrong
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents