{"id":6600,"date":"2025-04-21T04:10:00","date_gmt":"2025-04-20T19:10:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6600"},"modified":"2025-04-21T04:10:00","modified_gmt":"2025-04-20T19:10:00","slug":"can-llm-feedback-enhance-review-quality-a-randomized-study-of-20k-reviews-at-iclr-2025","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6600","title":{"rendered":"Can LLM feedback enhance review quality? A randomized study of 20K reviews at ICLR 2025\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Can LLM feedback enhance review quality? A randomized study of 20K reviews at ICLR 2025\u00a0<\/strong>[115.9]<br>Review Feedback Agent\u306f\u3001\u3042\u3044\u307e\u3044\u306a\u30b3\u30e1\u30f3\u30c8\u3001\u30b3\u30f3\u30c6\u30f3\u30c4\u306e\u8aa4\u89e3\u3001\u30ec\u30d3\u30e5\u30a2\u30fc\u3078\u306e\u5c02\u9580\u7684\u3067\u306a\u3044\u767a\u8a00\u306b\u5bfe\u3059\u308b\u81ea\u52d5\u7684\u306a\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3092\u63d0\u4f9b\u3059\u308b\u3002 ICLR 2025\u3067\u5927\u898f\u6a21\u306a\u30e9\u30f3\u30c0\u30e0\u5316\u5236\u5fa1\u7814\u7a76\u3068\u3057\u3066\u5b9f\u88c5\u3055\u308c\u305f\u3002 \u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u3092\u53d7\u3051\u305f\u30ec\u30d3\u30e5\u30a2\u30fc\u306e27%\u304c\u30ec\u30d3\u30e5\u30fc\u3092\u66f4\u65b0\u3057\u3001\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u304b\u3089\u306e12,000\u4ee5\u4e0a\u306e\u30d5\u30a3\u30fc\u30c9\u30d0\u30c3\u30af\u63d0\u6848\u304c\u30ec\u30d3\u30e5\u30a2\u30fc\u306b\u3088\u3063\u3066\u53d6\u308a\u5165\u308c\u3089\u308c\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2504.09737v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2504.09737v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Sun, 13 Apr 2025 22:01:25 GMT)<\/li>\n\n\n\n<li>ICLR\u306b\u3088\u308bReview Feedback Agent\u306e\u52b9\u679c\u691c\u8a3c\u3001\u300cThis suggests that many reviewers found the AI-generated feedback sufficiently helpful to merit updating their reviews. Incorporating AI feedback led to significantly longer reviews (an average increase of 80 words among those who updated after receiving feedback) and more informative reviews, as evaluated by blinded researchers.\u300d\u3068\u80af\u5b9a\u7684\u306a\u7d50\u679c\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/zou-group\/review_feedback_agent\">GitHub &#8211; zou-group\/review_feedback_agent<\/a><\/li>\n\n\n\n<li>\u672c\u8ad6\u3068\u306f\u95a2\u4fc2\u306a\u3044\u304c\u300cAuthors at AI conferences increasingly report receiving short, vague reviews with criticisms like \u2018not novel\u2019 or \u2018not state-of-the-art (SOTA)\u2019 \u300d\u3068\u3044\u3046\u306e\u306f\u5927\u5909\u305d\u3046\u306a\u30fb\u30fb\u30fb<\/li>\n<\/ul>\n\n\n\n<p>\u4f3c\u3066\u975e\u306a\u308b\u8ad6\u6587\u3067\u306f\u3042\u308b\u304c\u3001\u300cWe evaluated The AI Scientist-v2 by submitting three fully autonomous manuscripts to a peer-reviewed ICLR workshop.  Notably, one manuscript achieved high enough scores to exceed the average human acceptance threshold, marking the first instance of a fully AI-generated paper successfully navigating a peer review.\u300d\u3068\u3044\u3046AI Scientist-v2\u3082\u8208\u5473\u6df1\u3044\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search\u00a0<\/strong>[16.9]<br>AI Scientist-v2\u306f\u3001AI\u304c\u751f\u6210\u3057\u305f\u6700\u521d\u306e\u30d4\u30a2\u30ec\u30d3\u30e5\u30fc\u53d7\u3051\u5165\u308c\u30ef\u30fc\u30af\u30b7\u30e7\u30c3\u30d7\u7528\u7d19\u3092\u751f\u7523\u3067\u304d\u308b\u30a8\u30f3\u30c9\u30c4\u30fc\u30a8\u30f3\u30c9\u306e\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30b7\u30b9\u30c6\u30e0\u3067\u3042\u308b\u3002 \u79d1\u5b66\u7684\u306a\u4eee\u8aac\u3092\u53cd\u5fa9\u7684\u306b\u5b9a\u5f0f\u5316\u3057\u3001\u5b9f\u9a13\u3092\u8a2d\u8a08\u3057\u3001\u5b9f\u884c\u3057\u3001\u30c7\u30fc\u30bf\u3092\u5206\u6790\u3057\u3001\u8996\u899a\u5316\u3057\u3001\u79d1\u5b66\u7684\u306a\u539f\u7a3f\u3092\u81ea\u5f8b\u7684\u306b\u4f5c\u6210\u3059\u308b\u3002 \u3042\u308b\u5199\u672c\u306f\u3001\u5e73\u5747\u7684\u306a\u4eba\u9593\u306e\u53d7\u3051\u5165\u308c\u95be\u5024\u3092\u8d85\u3048\u308b\u5341\u5206\u306a\u30b9\u30b3\u30a2\u3092\u9054\u6210\u3057\u3001\u5b8c\u5168\u306aAI\u751f\u6210\u8ad6\u6587\u304c\u30d4\u30a2\u30ec\u30d3\u30e5\u30fc\u3092\u3046\u307e\u304f\u30ca\u30d3\u30b2\u30fc\u30c8\u3057\u305f\u6700\u521d\u306e\u4e8b\u4f8b\u3068\u306a\u3063\u305f\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2504.08066v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2504.08066v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 10 Apr 2025 18:44:41 GMT)<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/SakanaAI\/AI-Scientist-v2\">GitHub &#8211; SakanaAI\/AI-Scientist-v2: The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>\u4f3c\u3066\u975e\u306a\u308b\u8ad6\u6587\u3067\u306f\u3042\u308b\u304c\u3001\u300cWe evaluated The AI Scientist-v2 by submitting three fully autonomous manuscripts to a peer-revi &hellip; <a href=\"https:\/\/devneko.jp\/wordpress\/?p=6600\" class=\"more-link\"><span class=\"screen-reader-text\">&#8220;Can LLM feedback enhance review quality? A randomized study of 20K reviews at ICLR 2025\u00a0&#8221; \u306e<\/span>\u7d9a\u304d\u3092\u8aad\u3080<\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[223,623],"class_list":["post-6600","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-llm","tag-623"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6600","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6600"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6600\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6600"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6600"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6600"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}