Hacker News Reader

Comment by mpalmer | Hacker News Reader

Why would I want to take advice about keeping humans in the loop from someone who let an LLM write 90% of their blog post?

View on HN

I don't like reading AI text because I feel each word matters a lot less, however the message the author is conveying can be preserved. I read an article like this for the quality of the message not the craftsmen of the medium.

If the author didn't have the good taste and decency to edit the painfully obvious generated text, I just assume the message is low quality.

This is the new world we live in. Writers use AI to balloon a 2 paragraph thought into a full article, readers then use AI to compress the article into something akin to a 2 paragraph easily digestible piece. Everyone much happy. Example:

Key points from The Human in the Loop..

- The author pushes back on the idea that AI has made software developers obsolete, arguing instead that it has shifted where human effort matters.

- AI is increasingly good at producing code quickly, but that doesn’t remove the need for human oversight—especially for correctness, security, edge cases, and architectural fit.

- The “human in the loop” is not a temporary bottleneck but the accountable party who must understand, review, and take responsibility for what ships.

- Senior engineers’ most valuable skill has always been judgment, not typing speed—and AI makes that judgment even more critical.

- The author warns against blaming AI for bugs or bad outcomes; responsibility still lies with the human who approved the result.

- Software practices, team structures, and workflows need to evolve to emphasize review, verification, and intent over raw code production.

On what basis did you make this judgement? I found the article to be reasonable and not excessively padded.

But here's the thing. The LLM house writing style isn't just annoying, it's become unreadable through repeated exposure. This really gets to the heart of why human minds are starting to slide off it.

Not trying to be rude but your very short reply is hard to understand. "Unreadable", "starting to slide off", I honestly don't know what you're saying here.

Pretty sure they are mocking LLM outputs by making their own comment look like as if it came from LLM. It's sarcasm.

Other people might point to more specific tells, but instead I'll reference https://zanlib.dev/blog/reliable-signals-of-honest-intent/, which says that you can tell mainly because of the subconscious uncanny valley effect, and then you start noticing the tells afterwards.

Here, there's a handful of specific phrases or patterns, but mostly it's just that the writing feels very AI-written (or at least AI-edited). It's all just slightly too perfect, like someone's trying to write the perfect LinkedIn post but are slightly too good at it? It's purely gut feeling, but I don't think that means that it's wrong (although equally it doesn't mean that it's proven beyond reasonable doubt either, so I'm not going to start any witch hunts about it).

The human pressed the red button. :)

Reply on news.ycombinator.com