Can you tell AI writing from human writing?

AI can be a useful tool when used ethically. As a replacement for creative thinking is not one of those ethical uses, nor to provide work for qualifications. In and of itself, AI has no ethics, and it is for us to use it properly.
It's also geared for us to depend on LLMs for critical thinking and analysis... for everything. Some school kids not wanting to learn anything is and old trait. Might as well just let those ones frolic, mentally unburdened, since it's so easy to subvert education these days.

Remember when we cared about the environment? Water? It must really have been a sign that we didn't have enough problems.
 
Saw this on .com. Mia Ballard of Shy Girl controversary is suing Hachette for libel over the AI accusations. Very interesting. As they say in the video, this is textbook libel and puts the publisher in a position where they have to prove something they might not be able to prove.

Anybody want to be me a dollar that she wins in blowout?

 
Sounds like they got scared by the bad publicity, then dumped it.

The hilarious thing is the allegations ride on the back of how ultimately poor the prose is. Which you'd think a publisher, gatekeeper, would notice.

I believe Ballard used AI too, but Hachette's actions are so clearly based on public opinion panic that they deserve the lawsuit. It passed their own vetting process. They needed random internet personalities, and eventually a Times article, to tell them it's bad / AI writing? They have one job for crying out loud.

There may be a way to prove she used AI for that particular story, via discovery, but I doubt it.
 
There may be a way to prove she used AI for that particular story, via discovery, but I doubt it.
They won't be using AI detectors, that's for sure. Defense team will tear that shit apart. They'd need some kind of smoking gun. As Oscar Wilde found out the hard way, the parachute for the accused in a libel case is to prove that what they wrote--not "said," that's slander, not libel--is true, and not false. When Oscar Wilde sued the minister(?) who wrote that he was a sodomite, he forced the minister to prove that he really was a sodomite, which he was able to do to the satisfaction of the court at the time.
 
They won't be using AI detectors, that's for sure. Defense team will tear that shit apart. They'd need some kind of smoking gun.
I was thinking ISP records cross-referenced with a subpeona of whatever AI company she was allegedly using, but that would depend on the AI company even admitting it retains some level of personal LLM interaction data for a long duration.

I suspect the most evidence they could provide is just the ISP side, which at best, I believe, could indicate she was using an LLM and when, but not what for.
 
Anybody want to be me a dollar that she wins in blowout?
Chances are she'll make out with a decent settlement that involves no admission on either side. Chances are either way she'll never be published again.

In a way I hope she does win, set a precident for not baselessly accusing authors. Having the precident set the other way is a more frightening possibility.
 
Chances are either way she'll never be published again.
I don't know about that. If she gets enough publicity or sympathy or strikes a blow for creatives everywhere, they'll be tripping over each other to sign her up. That's all quite speculative, of course.

Chances are she'll make out with a decent settlement that involves no admission on either side.
Assuming the lawsuit doesn't get tossed for whatever reason, it'll be interesting to see how much stomach Hachette has for a fight. If I was her, I'd take them to the mattresses to make them prove it. I don't know how they could. That all assumes of course that the lawsuit in the libel sense puts them on that burden of proof footing. Difficult to imagine that the publisher didn't think this all the way through, but then again, and by their own admission, they let AI slip past the goalie, so who maybe they are that stupid?
 
Back
Top