[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-research-lets-ai-judges-adapt-how-they-gather-evidence":10,"sections":35},{"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":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},8356,"new-research-lets-ai-judges-adapt-how-they-gather-evidence","New Research Lets AI Judges Adapt How They Gather Evidence","BAER lets AI judges switch evidence-gathering methods per task, beating fixed protocols by up to 7.32 points across eight benchmark-backbone tests.","An AI judge that changes its own grading method depending on the job just beat every fixed method tested against it.\n\nResearchers built Backbone-Adaptive Evidence Routing (BAER), a system for pairwise LLM judges - AI models that compare two answers and pick the better one. Judges can gather evidence three ways: direct comparison, step-by-step reasoning, or checking against a reference answer, and no single method wins across every benchmark, so BAER offers three interchangeable heads instead - evidence stacking, routing to whichever method proves most reliable, and reference checking that ignores answer order. That order-blindness matters: swapping which answer comes first can flip which one a judge prefers, but BAER is built so it cannot change how strongly the judge prefers it. Tested on four benchmarks with two 8B-parameter judge backbones, the system picked its head from validation data, locked that choice in, then beat the strongest rival method in all eight test conditions, by 0.87 to 7.32 points.\n\nLLM judges already grade chatbot leaderboards and generate reward signals for reinforcement learning, almost always using one evidence-gathering recipe applied everywhere. BAER's results suggest that assumption is quietly costing accuracy, benchmark by benchmark, with nothing in place to catch it until now.\n\nWorth remembering: a 0.87-point gain is barely a nudge, and 8B backbones are modest by current standards - whether adaptive routing still helps once you swap in a 70B or frontier-scale judge is a question the paper leaves open.","[\"ai judges\",\"llm evaluation\",\"benchmarking\",\"arxiv\"]","2026-09-28T04:00:00.000Z","2026-09-29T02:39:29.253Z","2026-09-29T02:39:35.498Z","published",null,[],"ai",[26,27,28,29],"ai judges","llm evaluation","benchmarking","arxiv",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.30751",0,{"sections":36},[37,41,46,51,56,61,66,71,76,81,86,91,96,101],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",4918,"2026-09-29T05:13:43.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":45},"Security","security",768,"2026-09-29T01:20:00.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Policy","policy",406,"2026-09-28T19:04:09.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Deals","deals",272,"2026-09-28T17:42:46.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Hardware","hardware",191,"2026-09-28T15:45:00.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Science","science",154,"2026-09-28T13:19:18.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Consumer Tech","consumer-tech",139,"2026-09-28T17:09:47.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",91,"2026-09-25T20:55:00.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Dev Tools","dev-tools",87,"2026-09-28T16:11:42.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Startups","startups",80,"2026-09-28T17:50:28.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",49,"2026-09-28T16:44:57.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]