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: Which LLM-powered search do you use daily?

2 点作者 9woc大约 1 年前
There is a plethora of LLM-powered search, from perplexity, phind, copilot, meta.ai and more. Which ones do you use on a daily basis?

3 条评论

swatcoder大约 1 年前
None.<p>My web use is basically: tracking down a technical reference that is almost always at the vendor&#x27;s website or github repo; looking up an idle curiosity on imdb or wikipedia; searching reddit for a specific discussion topic with multiple perspectives, or lingering around here on HN to kill time.<p>I barely even use a search engine, and when I do it&#x27;s usually tagged with &quot;wiki&quot;, &quot;reddit&quot;, &quot;github&quot;, &quot;imdb&quot; etc because I know the source I want to explore and know how to interpret that source when presented to me in its familiar format.<p>For deeper understandings of things, I collect and read books about them or use something like JSTOR for narrower or fresher interpretations and perspectives.<p>LLM-powered searches just put an obscuring layer between me and the specific, attributed, source information that I find valuable.
vouaobrasil大约 1 年前
None. Using AI search is a mistake and we should not support the further anonymization of sharing into an LLM.
bpiche大约 1 年前
None, I use search engines with booleans when I want to find something. LLMs might provide useful citations at the end of a &#x27;search&#x27; but there&#x27;s too much hallucination in the summaries in my personal experience