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。

Meta Uses LLMs to Improve Incident Response

138 点作者 wilson0906 个月前

11 条评论

beoberha6 个月前
You’re better off reading the original post instead of this ad: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41326039">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41326039</a><p>On the general topic - I’m interested in people sharing their experiences using LLMs in livesite scenarios.
bagels6 个月前
I found something important was broken at Facebook, created a SEV, but the hardest part was figuring out what team to page. I unfortunately roped in one non-responsible team&#x27;s oncall who was making dinner at home, but thankfully, he did know what team to reach out to, based on the description of the problem.<p>Would be nice if there was better tooling for going from observed problem to responsible team.
评论 #42191739 未加载
评论 #42195699 未加载
pzh6 个月前
So if you look at the flip side, the on call engineer is being misled by AI ~60% of the time. The question is does this slow them down more or less than the speedup they get when the AI is right the other 40% of the time.
whiplash4516 个月前
I would expect HN to flag product marketing posts when the original post is available instead
netik6 个月前
Great idea but yet another blog post, which is actually marketing, which ends with “they did it buy our product so you can too”, which is probably not what Meta did.
tkgally6 个月前
Previous discussion:<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41326039">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=41326039</a>
nyellin6 个月前
I know there are already a number of comments here about proprietary solutions.<p>If you&#x27;re looking for something open source: <a href="https:&#x2F;&#x2F;github.com&#x2F;robusta-dev&#x2F;holmesgpt&#x2F;">https:&#x2F;&#x2F;github.com&#x2F;robusta-dev&#x2F;holmesgpt&#x2F;</a>
evnsio6 个月前
&gt; Humans aren&#x27;t great at incident response, and we all hate waking up at 2am to resolve an issue.<p>Agree that most people hate being woken at 2am, but disagree that humans aren&#x27;t great at incident response. Speaking generally, I think we&#x27;re about as good as it gets when it comes to adaptability and the kind reasoning that&#x27;s necessary to investigate complex issues.<p>That said, I also think AI can play a massive role aiding humans, especially in undifferentiated tasks like checking deployments, code changes, past incidents, and when it comes to spotting patterns.<p>IMO the sweet spot is going to come from highly ergonomic AI products that enable collaborative incident response, rather AI incident management or any other marketing BS.
评论 #42192318 未加载
评论 #42194030 未加载
jolarkin6 个月前
of course they do, why would this be interesting at all???
评论 #42209064 未加载
trod12346 个月前
Yeah this is just brilliant. &#x2F;s<p>Reading between the lines&#x2F;Translation: We need ~50% less IT operations staff, we can have LLMs do this instead.<p>The LLMs are coming for your white-collar work, and this is likely driving the white-collar recession.<p>Here&#x27;s a thought, incident response is one of the areas where entry-level SA&#x27;s sharpen their teeth and skills to become able to complete intermediate and senior level roles.<p>What happens when all the low-hanging fruit, the entry-level jobs are now replaced by LLMs and there is no short-term business need to hire such people.<p>No jobs means labor pool finds something else outside their profession, or they sit around on food stamps homeless, agitating with the homies, until a critical mass occurs.<p>Your intermediate and senior people naturally age and die, so how do you find replacements for them? ...<p>When there is no economic advantage for developing a skill set, no one goes into the field, it acts like a sieve, and eventually the skills involved becomes lost knowledge. Those that have the skills that are unable to find work seek work elsewhere, and rarely return. They were burned severely, enough times that it becomes a bad bet to try again in that field.<p>These things aren&#x27;t rocket science, and yet people seem to be so slothful or greedy, that they are unable to see or act to prevent what naturally happens next.<p>When people can&#x27;t find jobs to feed themselves or loved ones, where the only future which has been imposed on them is slavery or death, these people will organize and do the only thing they can once they are desperate enough; and that is violence. These same dynamics were present in the decades leading up to 1776, according to historic record.<p>It is so extremely short-sighted.
jedberg6 个月前
Wild Moose (W23) does this for everyone.<p><a href="https:&#x2F;&#x2F;www.wildmoose.ai">https:&#x2F;&#x2F;www.wildmoose.ai</a><p>Edit: Lol at myself, I thought this was a blog post from Meta and I was pointing out that there is a YC company that does this for everyone.<p>Now I realize that this was an ad for a different YC company that also does (although WM is a year older).
评论 #42193540 未加载
评论 #42190782 未加载