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。

How I Use Claude

28 点作者 zora_goron4 个月前

5 条评论

ithkuil4 个月前
&gt; Claude is really good at helping here, mostly because thinking quickly saturates: when you’ve thought about a problem for five minutes, you’ve had all the thoughts you’re gonna have, and it’s time to talk to someone else.<p>It reminds me of:<p>&quot;if you&#x27;re thinking too much, write; if you&#x27;re not thinking enough, read&quot;<p>It&#x27;s as if the act of writing engages yourself in a sort of conversation with the future reader.
treetalker4 个月前
&gt; Ironically the one situation I’ve found where Claude noticeably hallucinates is when I ask questions about a large and novel text that’s entirely in context. Which is the opposite of what I would expect to see. …<p>Sigh.<p>Accurate responses in such situations would be useful for busy professionals in low-stakes scenarios.<p>But LLMs cannot replace <i>the effect on the human mind</i> that results from actually reading, understanding, and thinking about a text. There is no substitute: we must do our own thinking, because it is the work that matters — the journey, not the destination, yields the benefits.
评论 #42575027 未加载
评论 #42574443 未加载
latexr4 个月前
&gt; thinking quickly saturates: when you’ve thought about a problem for five minutes, you’ve had all the thoughts you’re gonna have, and it’s time to talk to someone else.<p>Perhaps consider you suffer from incredibly short attention span. Thinking for five minutes <i>does not</i> exhaust all the thoughts you are going to have on a topic, and if you spend five minutes considering the implications of such a though you quickly realise how absurd it is.<p>History is <i>filled</i> with stories of “shower thoughts” and bolts of inspiration which came from thinking on a topic long and hard and immersing yourself in it. If your idea were true, humanity would still believe the Earth is the center of the Universe and we wouldn’t have computers. Yours is precisely the type of mentality which leads to the proliferation of scams and conspiracy theories. It’s also a worrying trend with LLMs, that people are so willing to turn off their brains sooner and sooner.
评论 #42575088 未加载
retskrad4 个月前
Which one is superior for explaining topics and going through practice problems related to the SAT test: Claude 3.5 or ChatGPt 4o?
评论 #42574543 未加载
satisfice4 个月前
I am utterly repelled by LLMs. I don’t know why otherwise thoughtful people use them, and this piece doesn’t explain the attraction either, except apparently what strikes me as creepy and pointless doesn’t strike everyone that way.<p>I notice little evidence of testing the information that he gets from Claude. From my own testing, which I repeat every so often, I find I cannot rely on anything I get from LLMs. Not anything. Have you tried AI summaries of documents or meetings that you know well? Are you happy with the results? I have not yet seen a summary that was good enough, personally.<p>Also a lot of example use cases he offers sound like someone who is not very confident in his own thinking, but strangely super-confident in whatever an LLM says ($2000&#x2F;hr consultant? really?).<p>Claude cannot perform an inquiry. No LLM can. These tools do not have inquiring minds, nor learning minds. He says hallucinations have reduced. How can he know that, unless he cross-checks everything he doesn’t already know?<p>I find LLMs exhausting and intellectually infantilizing. From this piece I cannot rule out that there is something very nice about Claude. But I also can’t rule out that there is a certain kind of addictive or co-dependent personality who falls for LLMs for unhealthy reasons primarily.