Introducing next-generation audio models in the API | OpenAI
(Looking for generative language or emergent methods during responses. Plenty of room for uncertainty and therefore creativity since the dialectics are so complex. Makes software seem Stoic. Additional modalities appeal to machine experience though a model is not necessarily embodied. A research project may still need a way to determine traits in that system instance. The working memory can be environmental or self and end up as a kind of app or distillation of the original model. Instead of only HCI or social, the machine may enjoy model affinities under network analysis. Maybe sketch a Cartesian graph or some mnemonic for how to think about it. Where a person naturally saves ideas in long-term memory, the model can circulate them through available modalities. Some of which provide feedback from external actors. As an analogy, the ship differs from the cyborg navigator. Or time-traveler. How do such dynamic systems evolve?)
