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

Manya Singh; Arjun Pakrashi

cs.AISep 28, 2026

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

Authors: Manya Singh, Arjun Pakrashi

Organizations: School of Computer Science, University College Dublin, Ireland

Abstract

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.

Figures & tables

Fig. 1 : Representative FER+ instances for the four human–model disagreement profiles. Each row shows the test image with the distribution of its $10$ human votes (red) and of the $8$ model predictions (grey) over the emotion classes: Neu(tral), Hap(piness), Sur(prise), Sad(ness), Ang(er), Dis(gust), Fea(r), Con(tempt).

Term	Definition
Human disagreement	Uncertainty arising from disagreement among human annotators when assigning labels to the same instance, as defined in [ 10 , 3 ] (there termed human uncertainty ). It captures the inherent ambiguity or subjectivity of the ground-truth label.
Model uncertainty	Uncertainty in a single model’s prediction, typically quantified using predictive entropy or another uncertainty score derived from its predicted class probabilities [ 12 , 13 ] . It captures how confident a model is about its prediction.
Predictive multiplicity	The disagreement in model predictions from a set of nearly equivalent accuracy models, or the Rashomon set of models [ 11 ] . The extent to which equivalent models reach conflicting predictions for the same instance.
Disagreement profile	The joint certainty status of an instance: whether the annotators and the models are each internally unanimous or divided, yielding the four profiles MC, MoD, HoD, and MD (Sec. III-F).
Alignment	The degree to which model uncertainty or disagreement corresponds to human disagreement. How well model uncertainty or multiplicity reflects the disagreement observed among human annotators.

TABLE I : Definitions of key terms used in this work.

	ResNet			EfficientNet			MobileNetV3
	18	34	50	B0	B1	B2	Small	Large
CIFAR-10H	94.34	96.31	96.05	96.13	94.96	96.20	92.95	95.10
FER+	77.38	77.58	78.44	78.28	80.32	77.41	77.86	78.72

TABLE II : Test accuracy (%) of the 8 fine-tuned models (three families) on CIFAR-10H and FER+ .

	Models Agree	Models Disagree	Total
Humans Agree	MC: $815$ ( $22.8$ %)	MoD: $186$ ( $5.2$ %)	$1001$ ( $28.0$ %)
Humans Disagree	HoD: $1195$ ( $33.5$ %)	MD: $1376$ ( $38.5$ %)	$2571$ ( $72.0$ %)
Total	$2010$ ( $56.3$ %)	$1562$ ( $43.7$ %)	$3572$

TABLE III : Human-model agreement on the FER+ test set (

n{=}3572

). Cells show counts with percentages of the test set in parentheses.

	Models Agree	Models Disagree	Total
Humans Agree	MC: $3310$ ( $33.1$ %)	MoD: $255$ ( $2.6$ %)	$3565$ (35.7%)
Humans Disagree	HoD: $5044$ ( $50.4$ %)	MD: $1391$ ( $13.9$ %)	$6435$ (64.4%)
Total	$8354$ ( $83.5$ %)	$1646$ ( $16.5$ %)	$10,000$

TABLE IV : Human–model agreement on the CIFAR-10H test set (

n{=}10{,}000

). Cells show counts with percentages of the test set in parentheses.

Fig. 4 : Per-instance human disagreement ( $H^{\mathrm{norm}}_{\mathrm{human}}$ ) vs. model disagreement for each dataset and model-side measure; colour gives image counts on a log scale. Even in the best-aligned setting ( FER+ , average predictive entropy ( $\bar{H}$ ), $\rho=0.572$ ), instances are widely dispersed off the diagonal. On CIFAR-10H the weak alignment holds for the continuous measure as well ( $\rho=0.363$ ). For inter-model vote entropy ( $H^{\mathrm{norm}}_{\mathrm{vote}}$ , $\rho=0.315$ ), the dense band at zero, which is the $83.5\%$ of images on which all eight models agree, spans the full range of human disagreement.

		ResNet			EfficientNet			MobileNetV3
		18	34	50	B0	B1	B2	Small	Large
CIFAR-10H	SC	-0.276	-0.237	-0.243	-0.340	-0.344	-0.317	-0.344	-0.350
CIFAR-10H	PE	0.275	0.235	0.241	0.344	0.346	0.327	0.346	0.347
FER+	SC	-0.470	-0.450	-0.420	-0.438	-0.550	-0.413	-0.488	-0.451
FER+	PE	0.474	0.450	0.421	0.442	0.554	0.416	0.491	0.453

TABLE V : Spearman correlations (

\rho

) between single-model uncertainty measures and normalised human vote entropy on FER+ and CIFAR-10H . All correlations are statistically significant (

p<0.05

). SC: Softmax Confidence; PE: Predictive Entropy.

	$\bar{C}$	$\bar{H}$	$H^{\mathrm{norm}}_{\mathrm{vote}}$	$D_{\mathrm{JSD}}$	$D_{\mathrm{within}}$	$D_{\mathrm{between}}$	$R$
CIFAR-10H	$-0.360$	$0.363$	$0.315$	$0.184$	$0.201$	$0.174$	$-0.184$
FER+	$-0.561$	$0.572$	$0.421$	$0.507$	$0.523$	$0.471$	$-0.314$

TABLE VI : Spearman correlations (

\rho

) between the aggregate model-side measures (Secs. III-D - III-F ) and human disagreement

H^{\mathrm{norm}}_{\mathrm{human}}

. All correlations are statistically significant (

p<0.05

Explore similar work

From Model Uncertainty to Human Attention: Localization-Aware Visual Cues for Scalable Annotation Review
May 12, 2026Moussa Kassem Sbeyti, Joshua Holstein, Philipp Spitzer +2Automatic Labelling Annotations
1Scientific Computing Center, Karlsruhe Institute of Technology. · Institute for Information Systems, Karlsruhe Institute of Technology.
An Assessment of Human vs. Model Uncertainty in Soft-Label Learning and Calibration
May 18, 2026Maja Pavlovic, Silviu Paun, Massimo PoesioArtificial Intelligence Alignment
Queen Mary University London · Amazon · Queen Mary University London - University of Utrecht
The Ghost Annotator: a Framework to Explore Human Label Variation in Content Moderation through Conformal Prediction
Jun 1, 2026Mirko Lai, Alessandra Urbinati, Simona Frenda +2Large Language Model Annotations Human Annotators
Heriot-Watt University, Edinburgh, Scotland · aequa-tech, Torino, Italy · Laboratory for the Modeling of Biological and Socio-technical Systems, Northeastern University, Boston, MA, USA +2

May 12, 2026, cs.HCJ/K move · Enter open · S save

From Model Uncertainty to Human Attention: Localization-Aware Visual Cues for Scalable Annotation Review

Moussa Kassem Sbeyti, Joshua Holstein, Philipp Spitzer, Nadja Klein+1

1Scientific Computing Center, Karlsruhe Institute of Technology. · Institute for Information Systems, Karlsruhe Institute of Technology.

Automatic Labelling Annotations Visual Cues

Open paper AskPDF arXiv

High-quality labeled data is essential for training robust machine learning models, yet obtaining annotations at scale remains expensive. AI-assisted annotation has therefore become standard in large-scale labeling workflows. However, in tasks where model predictions carry two independent components, a class label and spatial boundaries, a model may classify an object with high confidence while mislocalizing it. Existing AI-assisted workflows offer annotators no signal about where spatial errors are most likely. Without such guidance, humans may systematically underinspect subtly misplaced boxes. We address this by studying the effect of visualizing spatial uncertainty via a purpose-built interface. In a controlled study with 120 participants, those receiving uncertainty cues achieve higher label quality while being faster overall. A box-level analysis confirms that the cues redirect annotator effort toward high-uncertainty predictions and away from well-localized boxes. These findings establish localization uncertainty as a lever to improve human-in-the-loop annotation. Code is available at https://mos-ks.github.io/MUHA/.

Abstract

Explore similar work

From Model Uncertainty to Human Attention: Localization-Aware Visual Cues for Scalable Annotation Review

An Assessment of Human vs. Model Uncertainty in Soft-Label Learning and Calibration

The Ghost Annotator: a Framework to Explore Human Label Variation in Content Moderation through Conformal Prediction