{"id":41588,"date":"2025-08-21T11:32:43","date_gmt":"2025-08-21T03:32:43","guid":{"rendered":"https:\/\/www.1ai.net\/?p=41588"},"modified":"2025-08-21T11:32:43","modified_gmt":"2025-08-21T03:32:43","slug":"openai-%e7%a7%91%e5%ad%a6%e5%ae%b6%e8%b0%88-agi%ef%bc%9aai-%e6%ad%a3%e5%8a%a0%e9%80%9f%e6%8e%a8%e8%bf%9b%e7%a7%91%e5%ad%a6%e7%a0%94%e7%a9%b6%e8%87%aa%e5%8a%a8%e5%8c%96","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/41588.html","title":{"rendered":"OpenAI Scientist on AGI: AI is Accelerating the Automation of Scientific Research"},"content":{"rendered":"<p>In the latest issue <a href=\"https:\/\/www.1ai.net\/en\/tag\/openai\" title=\"[View articles tagged with [OpenAI]]\" target=\"_blank\" >OpenAI<\/a> In the podcast, the company's chief<a href=\"https:\/\/www.1ai.net\/en\/tag\/%e7%a7%91%e5%ad%a6%e5%ae%b6\" title=\"[Sees articles with [scientific] labels]\" target=\"_blank\" >the scientist<\/a>Yakubu Pachoki and Senior Fellow Simon Siddall delve into the <a href=\"https:\/\/www.1ai.net\/en\/tag\/agi\" title=\"_OTHER ORGANISER\" target=\"_blank\" >AGI<\/a> The current state of development and future direction of the<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-41589\" title=\"81173303j00t1br5h00msd000u000gwp\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/08\/81173303j00t1br5h00msd000u000gwp.jpg\" alt=\"81173303j00t1br5h00msd000u000gwp\" width=\"1080\" height=\"608\" \/><\/p>\n<p>When asked to define AGI, Pachocki admits that two years ago, AI models couldn't even correctly calculate the product of two four-digit numbers, whereas now they show creative thinking rather than rigid formula application in Olympiad questions.<\/p>\n<p>However, Pachocki believes that this point measurement is no longer sufficient, so they have begun to focus on the actual impact of AI on the world. In his view, for AI to truly have a meaningful impact on the world, the<strong>The key lies in its ability to drive the discovery of new technologies and automate production.<\/strong><\/p>\n<p>In discussing AI capability assessment, Pachocki said that<strong>There are a number of problems that are indeed being faced in benchmarking, the most obvious of which is the phenomenon of saturation, whereby models have reached human levels in many standardized tests of intelligence or ability.<\/strong><\/p>\n<p>Another problem is that the field has seen the emergence of many more efficient methods of training on data that can be optimized for specific capabilities and thus train models that are far superior in one area. Such models tend to excel in specific benchmarks, but that doesn't mean they have a higher level of overall intelligence.<\/p>\n<p>In planning the course of OpenAI's research program, Pachocki emphasized that they see the creation of highly generalizable intelligence as a core mission.<strong>They prioritize working on building AI systems that enable automated research, rather than narrowly limiting the technology to applications in specific areas.<\/strong><\/p>","protected":false},"excerpt":{"rendered":"<p>In the latest episode of the OpenAI podcast, the company's Chief Scientist Yakub Pachocki and Senior Researcher Simon Siddall delve into the current state of AGI's development and where it's headed in the future. When asked to define AGI, Pachocki admits that two years ago, AI models couldn't even correctly calculate the product of two four-digit numbers, and now they're showing creative thinking rather than rigid formula application in Olympiad questions. However, Pachocki believes that this kind of point measurement is no longer sufficient, so they have begun to focus on the actual impact of AI on the world. In his view, the key to AI truly having a meaningful impact on the world is its ability to drive the discovery of new technologies and automate production. In discussing the assessment of AI capabilities, Pachocki said that currently<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[151,190,1412],"collection":[],"class_list":["post-41588","post","type-post","status-publish","format-standard","hentry","category-news","tag-agi","tag-openai","tag-1412"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/41588","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=41588"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/41588\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=41588"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=41588"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=41588"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=41588"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}