Ai actually has huge problems with this. If you feed ai generated data into models, then the new training falls apart extremely quickly. There does not appear to be any good solution for this, the equivalent of ai inbreeding.
This is the primary reason why most ai data isn’t trained on anything past 2021. The internet is just too full of ai generated data.
This is why LLMs have no future. No matter how much the technology improves, they can never have training data past 2021, which becomes more and more of a problem as time goes on.
Doesn’t look like we’ll have much of a choice. They’re not going back into the bag.
We definitely need some good AI content filters. Fight fire with fire. They seem to be good at this kind of thing (pattern recognition), way better than any procedural programmed system.
Ai actually has huge problems with this. If you feed ai generated data into models, then the new training falls apart extremely quickly. There does not appear to be any good solution for this, the equivalent of ai inbreeding.
This is the primary reason why most ai data isn’t trained on anything past 2021. The internet is just too full of ai generated data.
This is why LLMs have no future. No matter how much the technology improves, they can never have training data past 2021, which becomes more and more of a problem as time goes on.
You can have AIs that detect other AIs’ content and can make a decision on whether to incorporate that info or not.
can you really trust them in this assessment?
Doesn’t look like we’ll have much of a choice. They’re not going back into the bag.
We definitely need some good AI content filters. Fight fire with fire. They seem to be good at this kind of thing (pattern recognition), way better than any procedural programmed system.
last time i’ve checked ais are pretty bad at recognizing ai-generated content
anyway there’s xkcd about it https://xkcd.com/810/