# Does Model Uncertainty Track Human Ambiguity? Evidence from Multi-Annotator Vision Benchmarks

arXiv: https://arxiv.org/abs/2609.34506 · HTML: https://mlsanity.com/paper/2609.34506/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 25 of its 31 references that are in the map.
- First posted on arXiv 28 Sep 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Human values and AI feedback alignment](https://mlsanity.com/alignment/human-values-and-ai-feedback-alignment/)

## Builds on (most cited first, 25 of 25)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://mlsanity.com/paper/1905.11946/) (2019)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles](https://mlsanity.com/paper/1612.01474/) (2016)
- [Searching for MobileNetV3](https://mlsanity.com/paper/1905.02244/) (2019)
- [Aleatoric and Epistemic Uncertainty in Machine Learning: An Introduction to Concepts and Methods](https://mlsanity.com/paper/1910.09457/) (2019)
- [A Survey of Uncertainty in Deep Neural Networks](https://mlsanity.com/paper/2107.03342/) (2021)
- [ResNet strikes back: An improved training procedure in timm](https://mlsanity.com/paper/2110.00476/) (2021)
- [SelectiveNet: A Deep Neural Network with an Integrated Reject Option](https://mlsanity.com/paper/1901.09192/) (2019)
- [Dealing with Disagreements: Looking Beyond the Majority Vote in Subjective Annotations](https://mlsanity.com/paper/2110.05719/) (2021)
- [Challenges in Representation Learning: A report on three machine learning contests](https://mlsanity.com/paper/1307.0414/) (2013)
- [Human uncertainty makes classification more robust](https://mlsanity.com/paper/1908.07086/) (2019)
- [Predictive Multiplicity in Classification](https://mlsanity.com/paper/1909.06677/) (2019)
- [Training Deep Networks for Facial Expression Recognition with Crowd-Sourced Label Distribution](https://mlsanity.com/paper/1608.01041/) (2016)
- [Evaluating Uncertainty Quantification in End-to-End Autonomous Driving Control](https://mlsanity.com/paper/1811.06817/) (2018)
- [Deep Active Learning for Biased Datasets via Fisher Kernel Self-Supervision](https://mlsanity.com/paper/2003.00393/) (2020)
- [Arbitrariness and Social Prediction: The Confounding Role of Variance in Fair Classification](https://mlsanity.com/paper/2301.11562/) (2023)
- [Asking the Right Question at the Right Time: Human and Model Uncertainty Guidance to Ask Clarification Questions](https://mlsanity.com/paper/2402.06509/) (2024)
- [Facial Emotion Recognition: State of the Art Performance on FER2013](https://mlsanity.com/paper/2105.03588/) (2021)
- [Dropout-Based Rashomon Set Exploration for Efficient Predictive Multiplicity Estimation](https://mlsanity.com/paper/2402.00728/) (2024)
- [Uncertainty Quantification in Machine Learning for Biosignal Applications -- A Review](https://mlsanity.com/paper/2312.09454/) (2023)
- [Mind the Uncertainty in Human Disagreement: Evaluating Discrepancies between Model Predictions and Human Responses in VQA](https://mlsanity.com/paper/2410.02773/) (2024)
- [Uncertainty Estimation by Human Perception versus Neural Models](https://mlsanity.com/paper/2506.15850/) (2025)
- [Robust Ambiguity Detection (RAD) From Model- and Feature-Space Consistency](https://mlsanity.com/paper/2608.11541/) (2026)

## Abstract (arXiv)

Human-model alignment is critical for trustworthy AI-assisted decision-making systems. Yet, most work evaluates model predictions against single ground-truth labels, overlooking that humans themselves often disagree on labels, a signal of genuine ambiguity. We investigate whether models struggle on the same instances that humans find difficult. We measure this on two vision datasets (FER+ and CIFAR-10H) where multiple human annotations per image capture human disagreement patterns. We evaluate eight pretrained models across three architectures (ResNet, EfficientNet, MobileNetV3) in two parts: first, whether model uncertainty (softmax confidence, entropy) correlates with human disagreement, and second, whether predictive multiplicity measures (inter-model disagreement, Jensen-Shannon divergence) do. We find that it does not: alignment is weak in both dimensions. At the discrete label level, 50.4% of CIFAR-10H images and 33.5% of FER+ images receive multiple valid classifications from humans, while the models converge on only one. These instances represent a critical failure case where humans perceive ambiguity and would request expert review, yet models decide confidently. At the continuous score level, single-model uncertainty correlates weakly with human disagreement (ρ = 0.24--0.55), and predictive multiplicity provides only modest improvement. Widely-used uncertainty quantification methods do not reliably identify instances humans find ambiguous. Model uncertainty should not be treated as a trustworthy signal by default for decision-making in high-stakes scenarios.