If you are suspicious, look at comment history. It's usually fairly obvious because all comments made by LLM spambots look the same, have very similar structure and length. Skim ten of them and it becomes pretty clear if the account is genuine.
I'm more worried about how many people reply to slop and start arguing with it (usually receiving no replies — the slop machine goes to the next thread instead) when they should be flagging and reporting it; this has changed in the last few months.
If you are suspicious, look at comment history.
I'm never suspicious though. One of the strange, and awesome, and incredibly rare things about HN is that I put basically zero stock in who wrote a comment. It's such a minimal part of the UI that it entirely passes me by most of the time. I love that about this site. I don't think I'm particularly unusual in that either; when someone shared a link about the top commenters recently there were quite a few comments about how people don't notice or how they don't recognize the people in the top ranks.
The consequence of this is that a bot could merrily post on here and I'd be absolutely fine not knowing or caring if it was a bot or not. I can judge the content of what the bot is posting and upvote/downvote accordingly. That, in my opinion, is exactly how the internet should work - judge the content of the post, not the character of the poster. If someone posts things I find insightful, interesting, or funny I'll upvote them. It has exactly zero value apart from maybe a little dopamine for a human, and actually zero for a robot, but it makes me feel nice about myself that I showed appreciation.
This makes me think a tool that lets me know how much of the engagement I was seeing was from bots would be huge.