别再叫我去问人工智能了
Stop Telling Me to Ask an LLM

原始链接: https://blog.yaelwrites.com/stop-telling-me-to-ask-an-llm/

当作者向资深专家寻求非教科书式的深度建议时,他们越来越多地收到同样的推脱:“去问 Claude 吧。” 作者认为,这是一种令人沮丧的替代方案,无法取代人类的洞察力。他们指出,在求教专家之前通常已经咨询过人工智能模型,这意味着他们的问题已经“经受住”了大型语言模型(LLM)的局限性。当专家建议使用人工智能时,这往往被视为一种礼貌的回绝方式,尽管作者真正寻求的是只有经过几十年的反复试验才能获得的“伤疤”与实践经验。 虽然作者承认专家往往很忙,并非每个问题都值得深思熟虑的回复,但他们认为“问 Claude”远不足以替代同行之间的判断。当一个问题复杂到连用户和人工智能都束手无策时,这种敷衍忽视了人类直觉的价值。作者总结道,虽然大语言模型在信息检索方面很有用,但它们无法替代通过多年专业经验积累而来的具体且细致的智慧;将用户推向人工智能,往往剥夺了他们真正寻求的专业见解。

Hacker News 上关于《别再让我去问大模型》一文的讨论,凸显了人们在寻求帮助时被推向人工智能工具后日益增长的挫败感。 评论者认为,“去问大模型”已成为现代版的“自己去谷歌”,它既可能是一种敷衍的打发,也可能是一种鼓励独立解决问题的初衷。许多参与者强调了“工作量证明”的重要性:如果提问者在不展示自身研究过程或背景的情况下提出问题,往往更容易得到敷衍式的回应。通过提供方案、具体的发现或先前的尝试,提问者更容易获得人类专家的积极回应。 然而,这场辩论的很大一部分焦点在于指导他人的职业责任。一些人认为,拒绝与初级员工交流(无论其付出了多少努力)是领导力的缺失;而另一些人则坚持认为,在人工智能工具充裕的时代,建议学习者在占用资深工程师的时间之前,先利用这些资源探索复杂课题是完全合理的。归根结底,这种矛盾源于对高效、以人为本的知识共享的渴望,与技术环境中管理时间和精力的实际需求之间的张力。
相关文章

原文

I'm a million times more likely to send an email or a text than to pick up the phone. But I had a question I thought was worth an actual call, so I scheduled one with someone senior enough to have real scar tissue, the kind you only get from watching a decision go sideways in a boardroom. I asked him where he'd look, personally, for the answer to a hard question I was chasing, one without industry consensus. Not what the textbook says. If five studies conflicted, which would he trust? I wanted the thing 30 years had taught him that a search engine couldn't.

"Honestly? Ask Claude."

That stung a little, but it wasn't the first time I'd heard it. Once it was a data problem I'd been stuck on. I'd approached it a half dozen different ways and could explain in some detail why none of them had worked. I reached out to a few people who do this kind of thing for a living, people I text with regularly, trading questions and working through whatever we're stuck on. All but one gave me the same redirect.

Each time this happened, I had already asked Claude. That wasn't the step I'd skipped. Before I ever reached out to a person, I'd spent a couple of hours (and sometimes way too many tokens) going back and forth with a large language model, and I still had a question that had survived all of that.

I'm old enough to remember people sending LMGTFY links to folks who didn't seem to know how to use a search engine and expected strangers to do unpaid research for them. But this isn't that. It's closer to what happens when I ask a friend for a food recommendation and get a top-10 list back. I'm not asking what Eater thinks is the best kind-of-quiet spot for late-night drinks, or for a great coffee shop in the city where they used to live. I'm asking what they think, because we have similar taste and a shared history, and because I know they have opinions about where the lists go wrong. I trust their experience over the expert consensus.

It's possible "ask the model" has become the polite version of "I don't know," or "I don't have time for this right now," or "I'd have to think about it." Maybe it's an easy way to decline giving an answer. But I'd take almost anything over the redirect. "I'm busy" is a real answer. "I can't think of anything you haven't already tried" is an answer too. What "ask Claude" doesn't give me is the person's specific, lived experience. That's the thing that's hard to write down and even harder to search for.

There's a real cost to being the person other people call, and it's not fair to expect everyone to bear it. It takes close attention and actual thought, and not everyone has that to spare on a day full of deadlines and fires to put out. Plenty of questions really can be answered by an LLM or a search engine. But when the question is one that already survived the model, "ask Claude" doesn't save anyone a step. It just withholds the thoughtful answer decades of experience could have given.

联系我们 contact @ memedata.com