[visarga]

Prompted, iterative generation of content is not like that, it is based on a novel experience of the interaction between human and AI, the AI assistant has guidance, feedback and grounding in the human interlocutor. Imagine millions or billions of such creative sessions, all ripe with ideas and feedback, that have been facilitated in part by AI, that is the novel experience that belongs to the model.

Unprompted generative art is interesting to see as a "walk through the latent space" of culture, it is a mirror house of ideas, where you can contemplate on the intricacies of our recorded collective experience. Considering the ingredients that went into cooking the model - all our culture - it represents the emergence of this data as a live interactive agent. How can we say the child of our culture is stripping all context and meaning?

[__loam]

Sounds pretty awful to me bud. If an AI art model generates some images in a latent space, and no one is around to see it, does it make a sound?

I'd prefer we not rely on horrifically inefficient, expensive, and statistically boring corporate software to generate culture. Feel free to use the lens given to you by thieves if that's how you want to experience the heritage of your civilization I guess.