Reddit is doomed anyway. People are using AI to start threads, and other people are using AI to comment on these threads. You can never know what you're interacting with.

▲

_fs 11 hours ago | parent | next [-]

Do you think that this is not happening here?

▲

gorgoiler 10 hours ago | parent | next [-]

Worse, I am repeatedly being accused nowadays of being an LLM. It probably doesn’t help that I riff-write with only a rough outline of what I want to say, not how to say it.

If the accusation is that I am an inference engine pumping out words based on a trailing context window then I am guilty as charged. It’s just that I run on Fe + C6H12O6 + O2 (a bloodstream charged with lunch and air) instead of y/C/N2 -> Si+e- (sunlight, coal, and wind turned into silicon electrons.)

▲

lelanthran 10 hours ago | parent [-]

> If the accusation is that I am an inference engine pumping out words based on a trailing context window then I am guilty as charged. It’s just that I run on Fe + C6H12O6 + O2 (a bloodstream charged with lunch and air) instead of y/C/N2 -> Si+e- (sunlight, coal, and wind turned into silicon electrons.)

This sort of tells me that you are pro-LLM, and most pro-LLM people mostly paste the contents of their ChatGPT output and try to pass it off as their own.

Given that you say you aren't, the most likely explanation might be that you are spending a lot of time reading LLM prose, and are starting to write like it now too.

	▲	huhkerrf 8 hours ago \| parent [-]
		I think you're proving OP's point.

▲

siva7 11 hours ago | parent | prev [-]

Got any proof?

▲

charcircuit 11 hours ago | parent [-]

Check a larger thread. It is pretty clear since there are people doing nothing to hide the writing style.

▲

21asdffdsa12 10 hours ago | parent | next [-]

You are absoawesomeamazingaffirmativeabundantauthenticabsolutely right!

▲

JumpCrisscross 10 hours ago | parent | prev [-]

> Check a larger thread. It is pretty clear

It tends to get downvoted and flagged.

▲

siva7 9 hours ago | parent [-]

I got a pretty interesting ai detection demo run through historic HN data, would love to see if HN could make use of this tech for HN. But i have no clue how to reach out or who might be a contact here.

	▲	JumpCrisscross 8 hours ago \| parent [-]
		Make it into a browser extension. Or, honestly, just a page with outputs and a tip jar. If there are interesting findings, highlight them-maybe blog about it and post that.

▲

altairprime 10 hours ago | parent | prev [-]

If you email comment links to the mods that you believe are AI-assisted, they’ll review and act on that. Footer contact link. It’s not hopeless.