AGI<\/a>The key path\u3002<\/p>\nHashabis points out that the law of scale (scaling laws) is the core principle of AI's progress, which is that \"the larger the model, the larger the data, the better the math, the higher the intelligence.\"\u3002<\/p>\n
WE HAVE TO SCALE THE CURRENT AI TO THE POINT WHERE IT WILL AT LEAST BECOME A KEY COMPONENT OF UNIVERSAL ARTIFICIAL INTELLIGENCE AND MAY EVEN CONSTITUTE THE ENTIRE AGI SYSTEM\u3002<\/p>\n
AGI IS SEEN AS A THEORETICAL SMART SYSTEM THAT CAN REASON AND PLAN LIKE HUMAN BEINGS AND IS THE GOAL OF GLOBAL TECHNOLOGY COMPANIES COMPETING\u3002<\/p>\n
HOWEVER, HASHABIS ALSO ACKNOWLEDGED THAT SIZE LAWS ALONE MAY NOT BE SUFFICIENT TO FULLY ACHIEVE AGI, AND THAT THERE MAY BE \u201cONE OR TWO ADDITIONAL BREAKTHROUGHS\u201d IN THE FUTURE\u3002<\/p>\n
He stressed that there were real constraints to scaleing: limited availability of data and increased calculus meant that more data centres would not only be costly but also put pressure on the environment\u3002<\/p>\n
At the same time, there were different voices in industry\u3002<\/p>\n
The former Meta chief AI scientist Yann LeCun (Yan Liqun) believed that the law of scale was not everything. In his speech at the National University of Singapore in April this year, he said, \"Most of the really interesting issues are extremely bad under the laws of scale, and you can't simply think that stacking data and compute power can produce a smarter AI.\" I'm sorry<\/p>\n
Before that, LeCun had left Meta to start a new company dedicated to the development of a \"world model\" based on spatial data, aimed at creating a new generation of AI systems capable of understanding the physical world and of lasting memory and complex reasoning\u3002<\/p>","protected":false},"excerpt":{"rendered":"
On December 9th, Google DeepMind CEO Demes Hassabis stressed at the recent Axios AI+ summit in San Francisco that the scale-up of artificial intelligence (AI) must be \"extremely driven\" as a key path to achieving universal artificial intelligence. Hashabis points out that the law of scale (scaling laws) is the core principle of AI's progress, which is that \"the larger the model, the larger the data, the better the math, the higher the intelligence.\" We have to scale the current AI to the point where it will at least become a key component of universal artificial intelligence and may even constitute the entire AGI system. A<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[146],"tags":[151,593,7974],"collection":[],"class_list":["post-47127","post","type-post","status-publish","format-standard","hentry","category-news","tag-agi","tag-deepmind","tag-demis-hassabis"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/47127","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=47127"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/47127\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=47127"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=47127"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=47127"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=47127"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
![\"c30e6ed0j00t6zgs5001sd000umim\"](\"https:\/\/www.1ai.net\/wp-content\/uploads\/2025\/12\/c30e6ed0j00t6zgs5001sd000u000mim.jpg\" "\\"c30e6ed0j00t6zgs5001sd000umim\\"")
<\/p>\n