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。

Ask HN: Socratic LLM?

3 点作者 lahvak12 个月前
Anybody tried to train a LLM, or seen a LLM that instead of providing "authoritative" answers to queries, would lead a user to a discovery through a series of questions and hints? Would such thing even be possible?

1 comment

exz12 个月前
It&#x27;s definitely possible. The founder of Khan Academy demonstrated something like this with GPT-4o: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=IvXZCocyU_M" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=IvXZCocyU_M</a>