Topic: bias

A new approach aims to make AI models used for detecting hate speech more transparent by aligning their reasoning with human rationales, addressing the challenge of culturally coded multilingual hate speech online that conventional systems can miss but lack explainability. This matters because it could help reduce bias and improve moderation accuracy without over-censorship or under-moderation, especially concerning Muslim communities.

arxiv.org↗

A study finds that foundation models in astronomy, trained on survey data including incomplete catalogues, inherit biases related to pixel-level incompleteness, potentially skewing mean redshift measurements. This matters because it highlights systemic issues in how astronomical data is processed and analyzed, impacting the accuracy of cosmological observations.

arxiv.org↗

Large Vision-Language Models (LVLMs), while effective in various tasks, can exhibit biased behavior due to social biases in their training data. Researchers propose a method called Counterfactual Ensemble Decoding to mitigate these biases, highlighting the importance of addressing fairness in AI systems.

arxiv.org↗

The study examines whether language models that pass behavioral bias tests still hold internal biases related to occupational competence, finding that they do retain such biases internally. This matters because it highlights the need for more comprehensive evaluation methods beyond surface-level behavior to ensure unbiased AI systems.

arxiv.org↗