TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Leading AI models fail new test of artificial general intelligence

4 点作者 Brajeshwar大约 2 个月前

1 comment

jqpabc123大约 2 个月前
Current AI is largely an attempt to use either memory or trial and error&#x2F;probability as a substitute for logic and reason.<p>In other words, as soon as someone publishes all the answers to this new test, AI will produce a genius score. But this still won&#x27;t mean it has &quot;learned&quot; much of anything.<p>Retrieving an existing answer is a database problem. Trial and error can be terribly inefficient. Neither of these demonstrate much in the way of real &quot;intelligence&quot;.