[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-ai-framework-claims-it-writes-better-papers-than-humans":10,"sections":34},{"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":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},6772,"new-ai-framework-claims-it-writes-better-papers-than-humans","New AI Framework Claims It Writes Better Papers Than Humans","A new autonomous AI system called ScientistTwo wrote research papers that outscored human authors in automated peer review, per its creators' study.","A new autonomous AI system says it can write scientific papers on its own, and by one measure, AI reviewers rated its work above human-written papers.\n\nThe system, called ScientistTwo, is built by a team of researchers as a fully autonomous multi-agent framework. Given a research problem, it establishes state-of-the-art baselines, forms its own hypotheses, and then coordinates a set of specialized AI agents to run the entire discovery process without a human in the loop. Those agents run experiments across multiple datasets and metrics, refine their methods through automated ablation studies, and check their own conclusions using a simulated peer-review and rebuttal step. The team benchmarked ScientistTwo against problems tied to papers already accepted at ICLR, ICML, and NeurIPS, and it produced full papers along with working, executable code.\n\nThis is part of a broader push to automate not just parts of research but the entire discovery cycle, from spotting the open problem to writing up the result. The standout claim, that ScientistTwo's papers scored higher on average than the human-authored ones, deserves a close look: the graders were automated AI review agents, not the human program committees who actually decide what gets into these conferences.\n\nAn AI system grading another AI system's homework is not the same as a real conference accepting it, and that is the test ScientistTwo has not yet faced.","[\"ai\",\"autonomous-agents\",\"ai-research\",\"arxiv\"]","2026-09-18T04:00:00.000Z","2026-09-18T16:00:58.677Z","2026-09-18T16:01:11.152Z","published",null,[],"ai",[24,26,27,28],"autonomous-agents","ai-research","arxiv",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.19644",0,{"sections":35},[36,39,43,48,53,57,61,66,71,76,81,86,91,96],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",3959,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",652,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",338,"2026-09-11T04:00:00.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",155,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",116,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",99,"2026-09-09T17:27:33.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Dev Tools","dev-tools",76,"2026-09-18T01:04:54.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",75,"2026-09-10T20:41:21.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Startups","startups",55,"2026-09-09T23:14:29.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Gaming","gaming",43,"2026-09-10T12:18:06.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",41,"2026-09-08T01:57:23.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]