I doubt that. Current models have been trained using curated datasets and fine tuned by humans. Going forward, I doubt it would be trained on random web data, b

I doubt that. Current models have been trained using curated datasets and fine tuned by humans. Going forward, I doubt it would be trained on random web data, but only from good sources. Even if the "source" uses text generated by a LLM, it's fine as long as it has been edited by humans - think news sources.