Hacker News Reader

Comment by KennyBlanken | Hacker News Reader

If we wanted to find out what an LLM thought we'd ask the LLM. Don't be a meat proxy.

https://gruhn.me/blog/2026-08-03/

But you weren't going to ask the LLM, and now you know something you weren't aware of. Seems no different than someone reporting what they found out in a Google Search, as long as they disclose that they used an LLM (as long as they used one that actually searches the web, but pretty sure even ChatGPT Free does that these days).

> but pretty sure even ChatGPT Free does that these days).

It can do, but won't do so reliably; and the guessing it does if it's not grounding with a search is still like someone in a pub saying "I recon…". Admittedly, a Cambridge pub like the Carlton Arms used to be when I lived there (sometimes I'd be the only one at the table who wasn't studying for a PhD), but still, pub.

One of the more annoying pre-llm behaviours was responding with the first link of google when a question was asked (but not of them specifically).

Even if you ignore my LLM numbers, "what percent of a reactor's output must be reinvested to enrich its own fuel?" is a meaningful contribution to the discussion.

But are your numbers reliable? By your own admission you don't know. So you aren't contributing expertise but rather the brief use of a tool that everyone has ready access to and which doesn't necessarily return reliable answers. AKA noise, at least from the perspective of a community like HN.

Relax dude.

This site is intended to be a discussion between people. If we wanted to consult some other resource, we would go do that. Injecting unverified information is not a contribution to a discussion between people, especially if it isn't coming from a human source.

No _book_ knowledge please! Hard to know if an author actually knew their subject and was making up numbers.

I ONLY want to hear first person _human_ experiences.

Jfc mate

Are you a meat proxy for that blog post?

Reply on news.ycombinator.com