Generative models are no longer just capable. They're developing taste.
Across image, video, audio, and interactive media, models are beginning to diverge in ways that go beyond benchmark scores: in taste, aesthetics, reasoning patterns, and creative behavior. What was once a single frontier is fracturing into something more interesting: a landscape of distinct creative intelligences, each with its own signature.
That raises harder questions. How do you give a model taste? How do you find signal in model divergence? How do you evaluate creativity when there's no longer a single right answer?…

Hosted by Albert - Founder of AI Circle & 4 others