I think we’re witnessing the death of open source AI. The logical outcome of this is that only large companies will be able to acquire and use the training data necessary to compete with ChatGPT.
So anyone who thinks otherwise will have to answer the question: how are we going to make any datasets?
It’s tempting to think that we can pull together data out of non copyrighted work. But there isn’t enough data. That would mean the model has no knowledge about any books.
I think we’re witnessing the death of open source AI. The logical outcome of this is that only large companies will be able to acquire and use the training data necessary to compete with ChatGPT.
So anyone who thinks otherwise will have to answer the question: how are we going to make any datasets?
It’s tempting to think that we can pull together data out of non copyrighted work. But there isn’t enough data. That would mean the model has no knowledge about any books.