{"id":1033,"date":"2026-05-18T00:30:24","date_gmt":"2026-05-17T15:30:24","guid":{"rendered":"https:\/\/itexplore.org\/jp\/columns\/openai-finance-ai-llm-memory-reorg\/"},"modified":"2026-05-18T00:30:24","modified_gmt":"2026-05-17T15:30:24","slug":"openai-finance-ai-llm-memory-reorg","status":"publish","type":"post","link":"https:\/\/itexplore.org\/jp\/columns\/openai-finance-ai-llm-memory-reorg\/","title":{"rendered":"OpenAI\u306e\u91d1\u878dAI\u3001LLM\u306e\u52b9\u7387\u7684\u30e1\u30e2\u30ea\u3001\u7d44\u7e54\u518d\u7de8\u306e\u52d5\u5411"},"content":{"rendered":"<p>\u672c\u65e5\u306e\u6ce8\u76eeAI\u30fb\u30c6\u30c3\u30af\u30cb\u30e5\u30fc\u30b9\u3092\u3001\u5c02\u9580\u7684\u306a\u5206\u6790\u3068\u5171\u306b\u304a\u5c4a\u3051\u3057\u307e\u3059\u3002<\/p>\n<div class=\"wp-block-vk-blocks-alert vk_alert alert alert-warning has-alert-icon\">\n<div class=\"vk_alert_icon\">\n<div class=\"vk_alert_icon_icon\"><i class=\"fa-solid fa-triangle-exclamation\" aria-hidden=\"true\"><\/i><\/div>\n<div class=\"vk_alert_icon_text\"><span>Warning<\/span><\/div>\n<\/div>\n<div class=\"vk_alert_content\">\n<p>\u3053\u306e\u8a18\u4e8b\u306fAI\u306b\u3088\u3063\u3066\u81ea\u52d5\u751f\u6210\u30fb\u5206\u6790\u3055\u308c\u305f\u3082\u306e\u3067\u3059\u3002AI\u306e\u6027\u8cea\u4e0a\u3001\u4e8b\u5b9f\u8aa4\u8a8d\u304c\u542b\u307e\u308c\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u305f\u3081\u3001\u91cd\u8981\u306a\u5224\u65ad\u3092\u4e0b\u3059\u969b\u306f\u5fc5\u305a\u30ea\u30f3\u30af\u5148\u306e\u4e00\u6b21\u30bd\u30fc\u30b9\u3092\u3054\u78ba\u8a8d\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/div>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">OpenAI\u3001\u500b\u4eba\u91d1\u878d\u5411\u3051ChatGPT\u3092\u767a\u8868\u3001\u9280\u884c\u53e3\u5ea7\u9023\u643a\u3082\u53ef\u80fd\u306b<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> OpenAI launches ChatGPT for personal finance, will let you connect bank accounts | TechCrunch<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<div class=\"ai-summary-content\">\n<p>TechCrunch\u306e\u8a18\u4e8b\u306f\u76f4\u63a5\u30a2\u30af\u30bb\u30b9\u3067\u304d\u307e\u305b\u3093\u3067\u3057\u305f\u304c\u3001\u30bf\u30a4\u30c8\u30eb\u304b\u3089\u5224\u65ad\u3059\u308b\u3068\u3001<strong>OpenAI<\/strong>\u304c\u500b\u4eba\u91d1\u878d\u7ba1\u7406\u306b\u7279\u5316\u3057\u305f<strong>ChatGPT<\/strong>\u306e\u65b0\u30d0\u30fc\u30b8\u30e7\u30f3\u3092\u767a\u8868\u3057\u305f\u3053\u3068\u304c\u5831\u3058\u3089\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\u3053\u306e\u65b0\u30b5\u30fc\u30d3\u30b9\u306f\u3001\u30e6\u30fc\u30b6\u30fc\u304c\u81ea\u8eab\u306e\u9280\u884c\u53e3\u5ea7\u3092\u63a5\u7d9a\u3057\u3001<strong>AI\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8<\/strong>\u304c\u8ca1\u52d9\u30c7\u30fc\u30bf\u306e\u5206\u6790\u3084\u7ba1\u7406\u3092\u652f\u63f4\u3059\u308b\u3053\u3068\u3092\u53ef\u80fd\u306b\u3059\u308b\u3067\u3057\u3087\u3046\u3002\u3053\u308c\u306b\u3088\u308a\u3001<strong>\u751f\u6210AI<\/strong>\u304c\u91d1\u878d\u30a2\u30c9\u30d0\u30a4\u30b9\u3084\u4e88\u7b97\u7de8\u6210\u3001\u652f\u51fa\u8ffd\u8de1\u306a\u3069\u306e\u30bf\u30b9\u30af\u3092\u81ea\u52d5\u5316\u3057\u3001\u500b\u4eba\u306e\u8ca1\u52d9\u72b6\u6cc1\u3092\u6700\u9069\u5316\u3059\u308b\u65b0\u305f\u306a\u9053\u304c\u958b\u304b\u308c\u308b\u3068\u4e88\u60f3\u3055\u308c\u307e\u3059\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/techcrunch.com\/2026\/05\/15\/openai-launches-chatgpt-for-personal-finance-will-let-you-connect-bank-accounts\/\" target=\"_blank\" rel=\"noopener\">TechCrunch \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<\/div>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> OpenAI is expanding ChatGPT into personal finance with bank account connectivity, leveraging AI agents for automated financial management.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Ivan Mehta<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\"><p><span>English Summary:<\/span><\/p>\n<p>Although the TechCrunch article was inaccessible, its title suggests that <strong>OpenAI<\/strong> has launched a new version of <strong>ChatGPT<\/strong> specifically designed for personal finance management.<\/p>\n<p>This new service is expected to allow users to connect their bank accounts, enabling <strong>AI agents<\/strong> to assist with financial data analysis and management. It is anticipated that this will open new avenues for <strong>Generative AI<\/strong> to automate tasks such as financial advice, budgeting, and expense tracking, thereby optimizing individual financial situations.<\/p>\n<\/blockquote>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">\u0394-Mem: \u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u305f\u3081\u306e\u52b9\u7387\u7684\u306a\u30aa\u30f3\u30e9\u30a4\u30f3\u30e1\u30e2\u30ea<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> \u0394-Mem: Efficient Online Memory for Large Language Models<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<div class=\"ai-summary-content\">\n<p>\u672c\u8ad6\u6587\u3067\u306f\u3001<strong>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb (LLM)<\/strong> \u304c\u9577\u671f\u7684\u306a\u30a2\u30b7\u30b9\u30bf\u30f3\u30c8\u3084\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8\u30b7\u30b9\u30c6\u30e0\u306b\u304a\u3044\u3066\u5c65\u6b74\u60c5\u5831\u3092\u84c4\u7a4d\u3057\u518d\u5229\u7528\u3059\u308b\u5fc5\u8981\u6027\u306b\u5bfe\u5fdc\u3059\u308b\u305f\u3081\u3001<strong>\u0394-Mem<\/strong>\u3068\u3044\u3046\u8efd\u91cf\u306a\u30e1\u30e2\u30ea\u30e1\u30ab\u30cb\u30ba\u30e0\u304c\u63d0\u6848\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u5f93\u6765\u306e\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u30a6\u30a3\u30f3\u30c9\u30a6\u306e\u62e1\u5f35\u306f\u30b3\u30b9\u30c8\u304c\u9ad8\u304f\u3001\u52b9\u679c\u7684\u306a\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u5229\u7528\u3092\u4fdd\u8a3c\u3067\u304d\u306a\u3044\u3068\u3044\u3046\u8ab2\u984c\u304c\u3042\u308a\u307e\u3057\u305f\u3002<\/p>\n<p><strong>\u0394-Mem<\/strong>\u306f\u3001\u51cd\u7d50\u3055\u308c\u305f\u30d5\u30eb\u30a2\u30c6\u30f3\u30b7\u30e7\u30f3\u30d0\u30c3\u30af\u30dc\u30fc\u30f3\u3092\u3001\u9023\u60f3\u30e1\u30e2\u30ea\u306e\u30b3\u30f3\u30d1\u30af\u30c8\u306a\u30aa\u30f3\u30e9\u30a4\u30f3\u72b6\u614b\u306b\u3088\u3063\u3066\u62e1\u5f35\u3057\u307e\u3059\u3002\u904e\u53bb\u306e\u60c5\u5831\u3092\u56fa\u5b9a\u30b5\u30a4\u30ba\u306e\u72b6\u614b\u884c\u5217\u306b\u5727\u7e2e\u3057\u3001\u30c7\u30eb\u30bf\u30eb\u30fc\u30eb\u5b66\u7fd2\u306b\u3088\u3063\u3066\u66f4\u65b0\u3057\u3001\u305d\u306e\u8aad\u307f\u51fa\u3057\u3092\u7528\u3044\u3066\u30d0\u30c3\u30af\u30dc\u30fc\u30f3\u306e\u30a2\u30c6\u30f3\u30b7\u30e7\u30f3\u8a08\u7b97\u306b\u4f4e\u30e9\u30f3\u30af\u88dc\u6b63\u3092\u751f\u6210\u3057\u307e\u3059\u3002<\/p>\n<p>\u308f\u305a\u304b8x8\u306e\u30aa\u30f3\u30e9\u30a4\u30f3\u30e1\u30e2\u30ea\u72b6\u614b\u3067\u3082\u3001<strong>\u0394-Mem<\/strong>\u306f\u51cd\u7d50\u3055\u308c\u305f\u30d0\u30c3\u30af\u30dc\u30fc\u30f3\u306e\u5e73\u5747\u30b9\u30b3\u30a2\u30921.10\u500d\u3001\u6700\u5f37\u306e\u975e<strong>\u0394-Mem<\/strong>\u30e1\u30e2\u30ea\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u306e1.15\u500d\u306b\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002\u7279\u306b\u30e1\u30e2\u30ea\u8ca0\u8377\u306e\u9ad8\u3044\u30d9\u30f3\u30c1\u30de\u30fc\u30af\u3067\u306f\u3001<strong>MemoryAgentBench<\/strong>\u30671.31\u500d\u3001<strong>LoCoMo<\/strong>\u30671.20\u500d\u3068\u5927\u304d\u306a\u30b2\u30a4\u30f3\u3092\u9054\u6210\u3057\u3064\u3064\u3001\u4e00\u822c\u7684\u306a\u80fd\u529b\u3092\u7dad\u6301\u3057\u307e\u3059\u3002<\/p>\n<p>\u3053\u308c\u3089\u306e\u7d50\u679c\u306f\u3001\u52b9\u679c\u7684\u306a\u30e1\u30e2\u30ea\u304c\u3001\u5b8c\u5168\u306a\u30d5\u30a1\u30a4\u30f3\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3001\u30d0\u30c3\u30af\u30dc\u30fc\u30f3\u306e\u7f6e\u304d\u63db\u3048\u3001\u307e\u305f\u306f\u660e\u793a\u7684\u306a\u30b3\u30f3\u30c6\u30ad\u30b9\u30c8\u62e1\u5f35\u306a\u3057\u306b\u3001\u30a2\u30c6\u30f3\u30b7\u30e7\u30f3\u8a08\u7b97\u306b\u76f4\u63a5\u7d50\u5408\u3055\u308c\u305f\u30b3\u30f3\u30d1\u30af\u30c8\u306a\u30aa\u30f3\u30e9\u30a4\u30f3\u72b6\u614b\u3092\u901a\u3058\u3066\u5b9f\u73fe\u3067\u304d\u308b\u3053\u3068\u3092\u793a\u3057\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/arxiv.org\/abs\/2605.12357\" target=\"_blank\" rel=\"noopener\">arXiv \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<\/div>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> \u0394-Mem offers an efficient, lightweight online memory solution for LLMs, significantly improving performance on memory-intensive tasks without extensive fine-tuning or context expansion, by using a compact associative memory state.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Jingdi Lei, Di Zhang, Junxian Li, Weida Wang, Kaixuan Fan, Xiang Liu, Qihan Liu, Xiaoteng Ma, Baian Chen, Soujanya Poria<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\"><p><span>English Summary:<\/span><\/p>\n<p>This paper introduces <strong>\u0394-Mem<\/strong>, a lightweight memory mechanism designed to address the increasing need for <strong>Large Language Models (LLMs)<\/strong> to accumulate and reuse historical information in long-term assistants and agent systems. Simply expanding the context window is often costly and fails to ensure effective context utilization.<\/p>\n<p><strong>\u0394-Mem<\/strong> augments a frozen full-attention backbone with a compact online state of associative memory. It compresses past information into a fixed-size state matrix, updated by delta-rule learning, and uses its readout to generate low-rank corrections to the backbone's attention computation during generation.<\/p>\n<p>Even with only an 8x8 online memory state, <strong>\u0394-Mem<\/strong> improves the average score to 1.10 times that of the frozen backbone and 1.15 times that of the strongest non-<strong>\u0394-Mem<\/strong> memory baseline. It achieves larger gains on memory-heavy benchmarks, reaching 1.31 times on <strong>MemoryAgentBench<\/strong> and 1.20 times on <strong>LoCoMo<\/strong>, while largely preserving general capabilities.<\/p>\n<p>These results demonstrate that effective memory can be realized through a compact online state directly coupled with attention computation, without the need for full fine-tuning, backbone replacement, or explicit context extension.<\/p>\n<\/blockquote>\n<\/div>\n<div class=\"wp-block-group\" style=\"margin-top:40px;margin-bottom:40px\">\n<h2 class=\"wp-block-heading\">Greg Brockman\u304cOpenAI\u306e\u88fd\u54c1\u90e8\u9580\u3092\u6b63\u5f0f\u306b\u638c\u63e1\u3001\u6700\u65b0\u306e\u7d44\u7e54\u518d\u7de8\u3067<\/h2>\n<ul>\n<li><strong>\u539f\u984c:<\/strong> Greg Brockman Officially Takes Control of OpenAI\u2019s Products in Latest Shakeup<\/li>\n<\/ul>\n<h3 class=\"wp-block-heading\">\u5c02\u9580\u30a2\u30ca\u30ea\u30b9\u30c8\u306e\u5206\u6790<\/h3>\n<div class=\"ai-summary-content\">\n<p>Wired\u306e\u8a18\u4e8b\u306f\u76f4\u63a5\u30a2\u30af\u30bb\u30b9\u3067\u304d\u307e\u305b\u3093\u3067\u3057\u305f\u304c\u3001\u30bf\u30a4\u30c8\u30eb\u304b\u3089\u5224\u65ad\u3059\u308b\u3068\u3001<strong>OpenAI<\/strong>\u306e\u5171\u540c\u5275\u8a2d\u8005\u3067\u3042\u308b<strong>Greg Brockman<\/strong>\u304c\u3001\u540c\u793e\u306e\u88fd\u54c1\u90e8\u9580\u306e\u7d71\u62ec\u3092\u6b63\u5f0f\u306b\u5f15\u304d\u7d99\u3044\u3060\u3068\u3044\u3046\u91cd\u8981\u306a\u7d44\u7e54\u518d\u7de8\u306b\u3064\u3044\u3066\u5831\u3058\u3066\u3044\u307e\u3059\u3002<\/p>\n<p>\u3053\u306e\u52d5\u304d\u306f\u3001<strong>OpenAI<\/strong>\u304c\u7814\u7a76\u958b\u767a\u304b\u3089\u88fd\u54c1\u5316\u3078\u306e\u79fb\u884c\u3092\u52a0\u901f\u3055\u305b\u3001<strong>\u751f\u6210AI<\/strong>\u6280\u8853\u3092\u3088\u308a\u5e83\u7bc4\u306a\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u306b\u5c55\u958b\u3059\u308b\u6226\u7565\u7684\u306a\u30b7\u30d5\u30c8\u3092\u793a\u5506\u3057\u3066\u3044\u307e\u3059\u3002<strong>Brockman<\/strong>\u306e\u30ea\u30fc\u30c0\u30fc\u30b7\u30c3\u30d7\u306e\u4e0b\u3001<strong>AI\u30a8\u30fc\u30b8\u30a7\u30f3\u30c8<\/strong>\u3084\u305d\u306e\u4ed6\u306e<strong>LLM<\/strong>\u30d9\u30fc\u30b9\u306e\u88fd\u54c1\u958b\u767a\u304c\u5f37\u5316\u3055\u308c\u308b\u3068\u4e88\u60f3\u3055\u308c\u307e\u3059\u3002<\/p>\n<p>\ud83d\udc49 <strong><a href=\"https:\/\/www.wired.com\/story\/openai-reorg-greg-brockman-product\/\" target=\"_blank\" rel=\"noopener\">Wired \u3067\u8a18\u4e8b\u5168\u6587\u3092\u8aad\u3080<\/a><\/strong><\/p>\n<\/div>\n<ul>\n<li><strong>\u8981\u70b9:<\/strong> Greg Brockman's official takeover of OpenAI's product division signals a strategic pivot towards accelerated product development and commercialization of Generative AI and LLM-based applications.<\/li>\n<li><strong>\u8457\u8005:<\/strong> Maxwell Zeff<\/li>\n<\/ul>\n<blockquote class=\"wp-block-quote\"><p><span>English Summary:<\/span><\/p>\n<p>Although the Wired article was inaccessible, its title indicates a significant organizational shakeup at <strong>OpenAI<\/strong>, reporting that co-founder <strong>Greg Brockman<\/strong> has officially taken control of the company's product division.<\/p>\n<p>This move suggests a strategic shift for <strong>OpenAI<\/strong>, accelerating its transition from research and development to product commercialization and deploying <strong>Generative AI<\/strong> technologies into broader applications. Under <strong>Brockman's<\/strong> leadership, the development of <strong>AI agents<\/strong> and other <strong>LLM<\/strong>-based products is expected to be significantly bolstered.<\/p>\n<\/blockquote>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>OpenAI\u304c\u500b\u4eba\u91d1\u878d\u5411\u3051ChatGPT\u3092\u767a\u8868\u3057\u3001LLM\u306e\u52b9\u7387\u7684\u306a\u30aa\u30f3\u30e9\u30a4\u30f3\u30e1\u30e2\u30ea\u300c\u0394-Mem\u300d\u304c\u767b\u5834\u3002\u307e\u305f\u3001Greg Brockman\u304c\u88fd\u54c1\u90e8\u9580\u3092\u7d71\u62ec\u3059\u308bOpenAI\u306e\u7d44\u7e54\u518d\u7de8\u3092\u6df1\u6398\u308a\u3002<\/p>\n","protected":false},"author":1,"featured_media":856,"comment_status":"","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"vkexunit_cta_each_option":"","footnotes":""},"categories":[3],"tags":[8,17,16,57,26,15,394,393],"class_list":{"0":"post-1033","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","6":"hentry","7":"category-columns","8":"tag-ai","10":"tag-llm","11":"tag-openai","12":"tag-26","14":"tag-394"},"_links":{"self":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts\/1033","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/comments?post=1033"}],"version-history":[{"count":0,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/posts\/1033\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/media\/856"}],"wp:attachment":[{"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/media?parent=1033"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/categories?post=1033"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/itexplore.org\/jp\/wp-json\/wp\/v2\/tags?post=1033"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}