{"version":1,"type":"story","url":"https://digestai.news/story/study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order","json":"https://digestai.news/story/study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order.json","markdown":"https://digestai.news/story/study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order.md","slug":"study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order","headline":"Study finds multimodal AI models shift answers based on evidence order","summary":"A new paper on arXiv examines how multimodal AI models process conflicting evidence from images, speech, and text. Researchers found that when models encounter conflicting information, the order of evidence matters: placing visual or auditory data *after* conflicting text shifts answers toward the later modality’s content. This phenomenon, called *cross-modal evidence noncommutativity*, contradicts earlier studies that assumed evidence order was neutral, the authors argue.\n\nThe study tests models on fixed instructions and evidence content, swapping only the order of modalities. Results show consistent bias toward perceptual data when it appears later, suggesting prior research may have misattributed model behavior to modality preference rather than presentation order. The findings challenge assumptions about how AI integrates multimodal inputs and could inform future experimental design.","keyPoints":["Models prioritize later-perceived evidence (images/audio) over conflicting text when placed after it","Prior studies may have misinterpreted modality bias due to fixed evidence order","Researchers propose a paired-comparison method to isolate order effects in multimodal tasks"],"whyItMatters":"This reveals a critical flaw in how AI systems process conflicting inputs, potentially skewing real-world applications like medical imaging or legal analysis where evidence order isn’t controlled.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":[],"models":["multimodal large language models"],"people":[]},"firstPublishedAt":"2026-09-24T04:00:00Z","updatedAt":"2026-09-24T04:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"arXiv cs.AI","title":"Same evidence, different judgments: Evidence noncommutative in vision/speech-text conflicts","url":"https://arxiv.org/abs/2609.26986","publishedAt":"2026-09-24T04:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Study finds multimodal AI models shift answers based on evidence order\", 24 September 2026, https://digestai.news/story/study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order","publisher":"Digest AI","title":"Study finds multimodal AI models shift answers based on evidence order","datePublished":"2026-09-24T04:00:00Z","url":"https://digestai.news/story/study-finds-multimodal-ai-models-shift-answers-based-on-evidence-order"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}