English

Hallucinations Live in Variance

Machine Learning 2026-01-13 v1 Artificial Intelligence

Abstract

Benchmarks measure whether a model is correct. They do not measure whether a model is reliable. This distinction is largely academic for single-shot inference, but becomes critical for agentic AI systems, where a single rephrased prompt can trigger cascading failures in multi-step execution. Yet this form of instability is not captured by existing evaluations. Hallucinations live in variance: they arise when semantically equivalent prompts activate inconsistent internal pathways, producing divergent outputs. Consistent but incorrect outputs reflect bias or missing knowledge; confident guessing reflects calibration failure. Neither constitutes hallucination under this definition. When error is variance-dominated, reducing redundant pathways improves reliability without adding knowledge. We formalize this through Semantic Stability (SS), measured via Paraphrase Consistency (PC@k): generate k paraphrases, greedy decode each, compute mode agreement. SS is a diagnostic for variance-driven unreliability, not a method for improving correctness. We show that a dense Qwen3-0.6B agrees with itself only 23.8% of the time; at 32% sparsity, agreement jumps to 55.9%. A phase diagram reveals the sweet spot where variance reduction outpaces bias accumulation, and regimes where stability collapses onto wrong answers.

Keywords

Cite

@article{arxiv.2601.07058,
  title  = {Hallucinations Live in Variance},
  author = {Aaron R. Flouro and Shawn P. Chadwick},
  journal= {arXiv preprint arXiv:2601.07058},
  year   = {2026}
}

Comments

8 pages, 3 figures

R2 v1 2026-07-01T08:59:49.196Z