Won't you praise yourself?
Posted: Wed Sep 24, 2025 4:18 pm
AI requires a large data set for training: video, audio, images, text. The type of information depends on the neural network and the human goals. The more content the machine receives, the better the results. You can also add your own proprietary materials to publicly available datasets. But there's a catch.
The sale of the painting
"Portrait of Edmond Belamy" is worth recalling. In 2018, a work created by artificial intelligence was auctioned at Christie's for the first time. The lot european tech industry data fetched $432,000. The portrait was based on works by famous artists—a special algorithm was fed 15,000 paintings painted between the 14th and 20th centuries. The AI studied them and created its own version of the portrait, as close as possible to the style of the historical examples.
It turns out that a neural network can be trained on the works of great people:
scientists, writers, and artists. If a neural network studies all of Rowling's works, it will offer its own version of "The Order of the Phoenix." We won't speculate on who would read a book written by AI. We'll simply emphasize that, by default, people want to gain new knowledge and are inspired by fresh ideas. Any repetition, clichés, or standards don't resonate widely. Yet, even such neural network work could become a cutting-edge tool in the hands of a writer. Or a con artist.
Creativity or circulation?
How can we avoid turning true creativity—unique, unrepeatable, and valuable—into mere mass production—hundreds of thousands of AI works that will soon flood the internet? This is the danger we discussed above. In 2022, the painting "Spatial Theater" was submitted to a fine art competition in Colorado. Artist Jason Allen created it with AI—the work won first place in the competition for emerging digital artists. This event sparked a series of protests and lawsuits.
The sale of the painting
"Portrait of Edmond Belamy" is worth recalling. In 2018, a work created by artificial intelligence was auctioned at Christie's for the first time. The lot european tech industry data fetched $432,000. The portrait was based on works by famous artists—a special algorithm was fed 15,000 paintings painted between the 14th and 20th centuries. The AI studied them and created its own version of the portrait, as close as possible to the style of the historical examples.
It turns out that a neural network can be trained on the works of great people:
scientists, writers, and artists. If a neural network studies all of Rowling's works, it will offer its own version of "The Order of the Phoenix." We won't speculate on who would read a book written by AI. We'll simply emphasize that, by default, people want to gain new knowledge and are inspired by fresh ideas. Any repetition, clichés, or standards don't resonate widely. Yet, even such neural network work could become a cutting-edge tool in the hands of a writer. Or a con artist.
Creativity or circulation?
How can we avoid turning true creativity—unique, unrepeatable, and valuable—into mere mass production—hundreds of thousands of AI works that will soon flood the internet? This is the danger we discussed above. In 2022, the painting "Spatial Theater" was submitted to a fine art competition in Colorado. Artist Jason Allen created it with AI—the work won first place in the competition for emerging digital artists. This event sparked a series of protests and lawsuits.