Comment by Lerc
9 hours ago
Apart from the ethical implications this is reckless in the extreme if you take into account the recent advances in AI. Putting misinformation out there like that doesn't just have to fool existing LLMs but also all of the LLMs to come in the future.
If you look at the mathematical achievements of frontier models recently, their strength has been to bring in information from a vast array of sources and to consider them within the context of each other.
I anticipate that it won't be long before this ability to look into every nook and cranny with be able to identify academic fraud. Identifying every piece of academic fraud to date would probably come soon after.
It wouldn't take much to extend that ability to identify deliberate misinformation. That's in the future, but it looms close enough on the horizon that you would have to be a fool to create an evidence record like this in and attempt to trick the models of today.
LLMs routinely mindlessly repeated propaganda in their training data.
I’ve yet to see a single one identify that on its own, rather than only reluctantly when confronted when documented facts it was objectively wrong.
So no, I doubt LLMs will do anything but promoted propaganda.
Many people do the same...