[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-study-shows-11-ai-models-disagree-on-what-info-to-chase":10,"sections":40},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},10579,"study-shows-11-ai-models-disagree-on-what-info-to-chase","Study Shows 11 AI Models Disagree on What Info to Chase","Researchers tested 11 large language models as information gathering agents and found their choices about what to ask next vary sharply by model family.","A new study finds that AI models put in charge of asking questions do not agree on what is worth asking about next.\n\nResearchers tested 11 large language models from multiple model families and parameter sizes on the same task: deciding what information to pursue next in an open-ended exchange, a setup the authors call information elicitation. Each model saw the same pool of information and used the same selection rule, so the only variable was the model's own judgment about what mattered. The team tracked whether models favored breadth, covering lots of topics shallowly, or depth, digging into fewer topics thoroughly, and how that balance shifted as interaction history built up. They also ran sensitivity checks, varying the options offered to the model, the labels used to define information value, and whether redundant questions counted against a choice.\n\nThis matters because elicitation agents sit behind things like AI interviewers, intake chatbots, and research assistants that are supposed to ask good follow-up questions on their own. If model choice alone decides whether an agent skims or digs deep, swapping the underlying model changes the product's behavior in ways no prompt tweak will fix. That is a real planning problem for anyone building on a third-party model API rather than a model they control.\n\nThe team posted its code, data, and full interaction trajectories on GitHub, so any team can check whether their favorite model is a shallow skimmer or a deep diver before building a product around it.","[\"ai\",\"llm-agents\",\"ai-research\",\"information-retrieval\"]","2026-10-07T04:00:00.000Z","2026-10-08T20:45:58.376Z","2026-10-08T20:46:25.547Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"publisher-r1","publisher",1,"The opening line says 'ten different AI models' but the study actually covers 11 models as stated in the next paragraph, an internal factual inconsistency.","resolved","ai",[30,32,33,34],"llm-agents","ai-research","information-retrieval",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.07509",0,{"sections":41},[42,46,51,56,61,66,71,76,81,85,90,95,100,105],{"name":43,"slug":30,"count":44,"latest_published_at":45},"AI",6431,"2026-10-07T18:45:00.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Security","security",902,"2026-10-07T19:53:42.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Policy","policy",474,"2026-10-07T18:23:21.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Deals","deals",453,"2026-10-07T23:58:31.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Hardware","hardware",222,"2026-10-07T21:19:54.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Science","science",186,"2026-10-06T21:20:39.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Consumer Tech","consumer-tech",174,"2026-10-07T17:41:41.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Software","software",113,"2026-10-07T18:10:00.000Z",{"name":82,"slug":83,"count":79,"latest_published_at":84},"Startups","startups","2026-10-07T23:36:57.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"Dev Tools","dev-tools",105,"2026-10-07T16:59:11.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"General","general",61,"2026-10-07T22:00:24.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Gaming","gaming",56,"2026-10-07T12:00:00.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"Reviews","reviews",33,"2026-10-05T11:57:17.000Z",{"name":106,"slug":107,"count":108,"latest_published_at":109},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]