Important information about tourism in Costa Rica - Florida

There has been an avalanche of advances in generating images from simple text prompts. In particular several people have asked me what I think about the question of name emulation using autonomous art tools. 

So here are some notes:

Since we’re in an age of innovation in machine-learning-driven art, we need to develop a bill of rights for the practice, especially when it comes to name emulation and use of trademarked properties.

Prompt: "A stunning photograph of a Pikachu wearing a cape, 8K HD, incredibly detailed"

by Cybertroniss using Dall-e2. Based on the Pokémon media franchise. 

Pikachu originally designed by Atsuko Nishida and Ken Sugimori,

Let’s assume that generative art via machine learning will eventually be able to perfectly emulate the style of a named artist, creating a pushbutton corpus of work that may be indistinguishable from samples that the artist labored to create by hand or voice.

For the purposes of play and experimentation, some artists will enjoy seeing their name / style emulated (I’m one of them), but others won’t. Living artists should be invited and should be able to opt out. The styles of artists who are dead / public domain / Creative Commons should be fair game.

Artist Studies by Remi Durant. Follow the link to explore various artist prompts

If a machine-generated artwork using name emulation is offered for sale, a living artist should share in the proceeds, and there should be an agreement in advance governing the venture that addresses approvals, distribution, etc.

We need to work this out. Machine learning tech is advancing rapidly. It may soon be able to invent new songs by Bob Dylan and new paintings by Norman Rockwell. To avoid confusion, infringement, and fraud, autonomous works should be clearly labeled so people know whether they are synthetic or authentic.