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。

It's not intelligence, it's "summarization and research"

2 点作者 profsummergig大约 1 个月前
I was looking at my prompt histories.<p>I think a lot of hand-wringing about AI (&#x2F; LLMs) could be mitigated if one were to call it what it really is (so far): a really good summarization and research tool. Not artificial intelligence per se.<p>Or at least that&#x27;s what I&#x27;m calling it from now on. In my mind. To reduce the guilt induced by the hand-wringers.<p>Discuss.

3 条评论

overu589大约 1 个月前
There is a matter of perspective.<p>Abstract intrigues symbolized, iconified, codified, correlated.<p>Am I describing what you are doing or what the LLM is doing?<p>LLM is an “intelligent” tool, generated from “intelligence” as both information and design by intelligent others.<p>Are you intelligent for using it?<p>Are you intelligent for relying upon it?<p>Does it reduce your uncertainty? Empower your capacity for resolve?<p>As far as “intelligence” is concerned you are referring to a process and an applicable actor of that process. You are not for instance referring to a natural language search card catalog.
keernan30 天前
I view it as a sophisticated retrieval tool performing lookups in history&#x27;s largest encyclopedia and returns the information nicely collated.<p>It has no independent &#x27;understanding&#x27; of the information it retrieved. LLM&#x27;s are incapable of thinking of something new... of something that has never existed or been imagined before (eg something not known or implied in its data).
delichon大约 1 个月前
It seems to me that to be able to usefully summarize natural language is a clear sign of intellect. It is how we test natural students after all. Describing it with another phrase doesn&#x27;t change that.