Researchers have introduced INCLUDE-BENCH, a new benchmark designed to evaluate disability-related biases in text-to-image (T2I) models. The benchmark, which includes 119,000 generated images, reveals that T2I models often depict individuals with disabilities, particularly those with mobility impairments, predominantly in wheelchairs. The study also found that disability-conditioned generations tend to have less diversity and that stereotypical portrayals align more strongly with disability-related text prompts. A new metric, the Stereotype Content Model (SCM) Score, was developed to quantify these real-world stereotypical associations reflected in T2I outputs. AI
IMPACT Highlights the need for more nuanced evaluation of AI models to prevent the perpetuation of harmful stereotypes towards people with disabilities.
RANK_REASON The cluster describes a new academic paper introducing a benchmark for evaluating bias in AI models.
Read on Hugging Face Daily Papers →
- alphaXiv
- arXiv
- CatalyzeX
- DagsHub
- Gotit.pub
- Hugging Face
- INCLUDE-BENCH
- PWD
- ScienceCast
- Sophia Lichtenberg
- Stereotype Content Model (SCM) Score
- T2I models
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →