Comment by JumpCrisscross
1 day ago
Kagi just launched an AI flagging feature. This is something HN needs.
I don’t mind AI content. But I’m not going to read it carefully before commenting, and will double check it with real sources before changing my mind about anything.
Speaking of real sources, the photographer discusses how it was taken and and answers questions on Reddit https://www.reddit.com/r/spaceporn/comments/1ow9mys/the_most...
Has AI content detection become at all competent?
I think we probably need to go the other way, a comment or article needs a ‘faceId’ check before submission to get human stamp.
Of course that brings a whole another set of problems.
Fine, I'll just set up a head made by some SFX type that looks real enough for camera work.
That sort of thing is trivial to bypass and is a massive privacy violation.
“Encryption is easy — authentication is hard.”
From what I have seen, AI can detect AI better than humans. Humans are bringing baggage into it and vilifying legit writing techniques just because AI uses them. The whole concept that AI was trained on human writing, therefore uses well-known and proven writing techniques, but that those techniques are not exclusive to AI is utterly lost on some people.
The "dead giveaways" are not writing patterns, it is depth. AI will stay at a surface level when using argumentative writing patterns, whereas humans will add supporting information and connect the dots across sentences and paragraphs. It is the lack of connective language between thoughts and phrases that flag an AI.
[dead]
HN can't even get it's web front end out of the 90s or meet basic accessibility standards.
*its