The Future of AI

Daniel imagines that future AI's are some base model, like GPT-4, adjusted for different use cases. In this scenario, there might be three to ten big AI companies capable of running GPT- 4 style large models. Each might have a flagship product, trained in a slightly different way and with a slightly different starting random seat. Daniel: If every instance of an AI was fine-tuned differently, so that the missile defense AI had different values from the factory running AI or the traffic coordinating AI then maybe AI factions would be smaller and more heterogeneous. This wouldn't bode well for humans.