Exploring AI: Training Models on Classic Texts (Full Transcript)

Speaker 1: Do you think we should be sending chat GPT to Bible school?

Speaker 2: Absolutely. Absolutely. Well, we've been, I have a colleague, we've been training AI systems on classic texts. They're way more useful. I use one all the time. We haven't released it. How is it more useful? It's not woke.

Speaker 1: Seriously, it's not ideologically adamant. But surely it is ideologically controlled and confined, just in a different way. No, I don't think it's ideologically confined. How far do you go with this? I mean, like, do you train this model on the New Testament? Do you train it on the epistles?

Speaker 2: We're playing with that. I don't know. Do you train it on John Milton? Good question. Dante. Dante, Ovid. Well, so imagine, that's a very good question. So imagine that at the foundation, you have the biblical library. Okay, but then you have, like you said, the secondary literatures. What about the Quran? We want to train a separate one on the Quran and the Hadith. And then we want to have them debate. Yeah.