I don’t think that’s an architectural requirement. We could train LLMs that are better at reporting differences of opinion about history, areas of ignorance, areas where archival material needs to be consulted, &c. I also think *good* (underline) historical language models would increase curiosity.

Agreed on the genie point & on @grimalkina.bsky.social‘s inspiring thread. But I think you're underestimating the basic literacy, thinking skills, and comfort w cognitive load that’s needed to use these tools productively. Anyone reading this ALREADY has that foundation +