{"id":7943,"date":"2025-12-22T05:42:00","date_gmt":"2025-12-21T20:42:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=7943"},"modified":"2025-12-21T18:17:42","modified_gmt":"2025-12-21T09:17:42","slug":"evaluating-large-language-models-in-scientific-discovery","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=7943","title":{"rendered":"Evaluating Large Language Models in Scientific Discovery"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Evaluating Large Language Models in Scientific Discovery&nbsp;<\/strong>[91.7]<br>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb (LLMs) \u306f\u79d1\u5b66\u7814\u7a76\u306b\u307e\u3059\u307e\u3059\u5fdc\u7528\u3055\u308c\u3066\u304d\u3066\u3044\u308b\u304c\u3001\u79d1\u5b66\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u306f\u975e\u6587\u8108\u5316\u3055\u308c\u305f\u77e5\u8b58\u3092\u63a2\u7d22\u3057\u3066\u3044\u308b\u3002 \u751f\u7269, \u5316\u5b66, \u6750\u6599, \u7269\u7406\u306b\u307e\u305f\u304c\u308bLSM\u3092\u8a55\u4fa1\u3059\u308b\u30b7\u30ca\u30ea\u30aa\u30b0\u30e9\u30a6\u30f3\u30c9\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3092\u63d0\u6848\u3059\u308b\u3002 \u3053\u306e\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306f\u3001(i)\u30b7\u30ca\u30ea\u30aa\u30bf\u30a4\u30c9\u30a2\u30a4\u30c6\u30e0\u306e\u8cea\u554f\u30ec\u30d9\u30eb\u7cbe\u5ea6\u3068(ii)\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30ec\u30d9\u30eb\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u306e2\u3064\u306e\u30ec\u30d9\u30eb\u3067\u30e2\u30c7\u30eb\u3092\u8a55\u4fa1\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2512.15567v1\">\u8ad6\u6587<\/a>&nbsp;&nbsp;<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2512.15567v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>&nbsp; &nbsp;(Wed, 17 Dec 2025 16:20:03 GMT)<\/li>\n\n\n\n<li>AI\u306b\u79d1\u5b66\u7684\u767a\u898b\u306f\u3067\u304d\u308b\u306e\u304b\uff1f\u3068\u3044\u3046\u554f\u3044\u306b\u5bfe\u3059\u308b\u8a55\u4fa1\u3002\u30af\u30a4\u30ba\u306e\u3088\u3046\u306a\u5f62\u5f0f\u3067\u306f\u306a\u304f\u3001\u7814\u7a76\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u3001\u73fe\u5834\u306e\u7814\u7a76\u30b7\u30ca\u30ea\u30aa\u306b\u57fa\u3065\u304f\u8a55\u4fa1\u3002\u300cLarge performance variation in research scenarios leads to changing choices of the best performing model on scientific discovery projects evaluated, suggesting all current LLMs are distant to general scientific \u201csuperintelligence\u201d.\u300d\u3068\u306e\u3053\u3068\u3067\u306f\u3042\u308b\u304c\u3001\u6709\u52b9\u6027\u3082\u611f\u3058\u308b\u5370\u8c61\u3092\u6301\u3063\u305f\u3002<\/li>\n\n\n\n<li>\u300cwe observe striking exceptions to the positive correlation between question- and project-level performance.&nbsp;\u300d\u300cThis suggests that rigorous knowledge of explicit structure-property relationships is not a strict prerequisite for LLM-driven discovery. Rather, the capacity to discern optimization directions and facilitate serendipitous exploration appears more critical.\u300d\u3068\u3044\u3046\u6307\u6458\u304c\u9762\u767d\u3044\u3002\u3069\u306e\u30e2\u30c7\u30eb\u304c\u826f\u3044\u304b\u3082\u554f\u984c\u306b\u3088\u3063\u3066\u304b\u306a\u308a\u72b6\u6cc1\u304c\u5909\u308f\u308b\u3088\u3046\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[387,623],"class_list":["post-7943","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-survey","tag-623"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7943","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=7943"}],"version-history":[{"count":2,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7943\/revisions"}],"predecessor-version":[{"id":7957,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/7943\/revisions\/7957"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7943"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7943"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7943"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}