Pith. sign in

REVIEW 5 cited by

The Bias Amplification Paradox in Text-to-Image Generation

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2308.00755 v2 pith:A2QJHKSY submitted 2023-08-01 cs.LG cs.CLcs.CVcs.CY

classification cs.LGcs.CLcs.CVcs.CY
keywords trainingamplificationbiasdatabiasescaptionscomparinggender
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Bias amplification is a phenomenon in which models exacerbate biases or stereotypes present in the training data. In this paper, we study bias amplification in the text-to-image domain using Stable Diffusion by comparing gender ratios in training vs. generated images. We find that the model appears to amplify gender-occupation biases found in the training data (LAION) considerably. However, we discover that amplification can be largely attributed to discrepancies between training captions and model prompts. For example, an inherent difference is that captions from the training data often contain explicit gender information while our prompts do not, which leads to a distribution shift and consequently inflates bias measures. Once we account for distributional differences between texts used for training and generation when evaluating amplification, we observe that amplification decreases drastically. Our findings illustrate the challenges of comparing biases in models and their training data, and highlight confounding factors that impact analyses.

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Understanding and evaluating computer vision models through the lens of counterfactuals

    cs.CV 2025-08 conditional novelty 6.0 of 10

    Counterfactual-based methods for concept attribution in classifiers and for dynamic bias evaluation and mitigation in text-to-image models.

  2. Mitigate One, Skew Another? Tackling Intersectional Biases in Text-to-Image Models

    cs.CV 2025-05 conditional novelty 6.0 of 10

    BiasConnect predicts how mitigating bias on one axis shifts bias on another axis in text-to-image models, and InterMit uses that to guide efficient multi-axis bias mitigation.

  3. Prompting Away Stereotypes? Evaluating Bias in Text-to-Image Models for Occupations

    cs.CL 2025-08 conditional novelty 5.0 of 10

    Diversity prompts shift the gender and race of AI-generated occupational images, but the effect is unstable and model-specific, often overcorrecting.

  4. Inference Time Debiasing Concepts in Diffusion Models

    cs.GR 2025-08 reject novelty 5.0 of 10

    DeCoDi subtracts a biased-concept guidance term during diffusion inference to shift generated images away from targeted stereotypes, with evaluation on gender, ethnicity, and age.

  5. VideoGuard: Protecting Video Content from Unauthorized Editing

    cs.CV 2025-08 unverdicted novelty 5.0 of 10

    VideoGuard adds joint, motion-aware perturbations to videos to block unauthorized diffusion-model editing.

Pith tools