Yes. So add relevant new data along with the older stuff. The problem is not that AI-generated content is magically “poison” somehow. Model collapse happens when you lose rare data from repeated generations of training data generated by AIs.
A simple way to imagine it is training an AI by showing it random coloured marbles out of a bucket and then asking it to fill the next AI’s bucket with new marbles to train on. If there’s just one single blue marble in the first bucket then it’s easily possible that the AI will fail to put a blue marble in the second bucket, after which there will never be a blue marble again if that’s all that subsequent AIs have to train off of. But if each time you train a new AI you reuse half the marbles from the first bucket again, you can have that blue marble show back up again in future AIs.
If LLMs are as revolutionary as the zealots believe, then there will exist less and less blue marbles in the universe with each iteration. So either the bucket gets smaller or the ratio of blue marbles gets smaller.
But if each time you train a new AI you reuse half the marbles from the first bucket again, you can have that blue marble show back up again in future AIs.
The original bucket containing the blue marble isn’t going anywhere. It still exists. The blue marble will always be available to mix into future AIs. All you have to do is make sure you’re using some historical data (or otherwise guaranteed “human-generated”) along with whatever new unvetted stuff you’re using.
All you have to do is make sure you’re using some historical data (or otherwise guaranteed “human-generated”) along with whatever new unvetted stuff you’re using.
Emphasis added. Please read more carefully, this is getting repetitive. You keep assuming that the AI will be trained either entirely with old data or entirely with new data and that’s just not the case.
Then the missing diversity comes from the non-AI-generated stuff that’s included in the mix.
I’m not sure what the problem is here. The cause of model collapse when AIs are fed on the output of previous generations is that the rare “fringes” of the data are lost over time. The training data becomes increasingly monotonous. Adding that fringe data back in should cure that.
Yes. So add relevant new data along with the older stuff. The problem is not that AI-generated content is magically “poison” somehow. Model collapse happens when you lose rare data from repeated generations of training data generated by AIs.
A simple way to imagine it is training an AI by showing it random coloured marbles out of a bucket and then asking it to fill the next AI’s bucket with new marbles to train on. If there’s just one single blue marble in the first bucket then it’s easily possible that the AI will fail to put a blue marble in the second bucket, after which there will never be a blue marble again if that’s all that subsequent AIs have to train off of. But if each time you train a new AI you reuse half the marbles from the first bucket again, you can have that blue marble show back up again in future AIs.
If LLMs are as revolutionary as the zealots believe, then there will exist less and less blue marbles in the universe with each iteration. So either the bucket gets smaller or the ratio of blue marbles gets smaller.
I said:
The original bucket containing the blue marble isn’t going anywhere. It still exists. The blue marble will always be available to mix into future AIs. All you have to do is make sure you’re using some historical data (or otherwise guaranteed “human-generated”) along with whatever new unvetted stuff you’re using.
So then your back to locking LLMs to the year 2023. They’re usefulness is severely limited if you can’t train them on new data.
Emphasis added. Please read more carefully, this is getting repetitive. You keep assuming that the AI will be trained either entirely with old data or entirely with new data and that’s just not the case.
And what happens when “whatever new unvetted stuff” is primarily comprised of AI-generated content?
Then the missing diversity comes from the non-AI-generated stuff that’s included in the mix.
I’m not sure what the problem is here. The cause of model collapse when AIs are fed on the output of previous generations is that the rare “fringes” of the data are lost over time. The training data becomes increasingly monotonous. Adding that fringe data back in should cure that.