In my articles up to now I have been discussing AI and art only in the sense that AI is the subject of much art when I consider entertainment media including books, films, plays and TV shows. Now it’s time to wade gingerly into the subject of AI producing art, and I realize there may be many of you who have already jumped up and shouted at the page “AI cannot produce art! It only copies and regurgitates what a person has already done”. I’m not saying that’s wrong, but I’m also saying it’s not entirely right. To properly address this subject we can define art. Looking at Merriam-Webster we find the word “skill” used in four of the six definitions of art. I have no skill in the fine arts, so if I make a bad drawing/painting/sculpture — OK, when I make a bad painting, is that art? If skill is required then mine would not qualify. However, I’ve seen many works of art that seem to show the same level of skill as I have. Is that my fault? Is it my perception of art which renders me incapable of seeing it, maybe because I don’t have the skill to produce it? Do we need skill to produce AND to perceive art? Another definition reads “the works produced by artists”. Do I have to define myself as an artist for my work to be considered art even without skill? If AI produces the same thing I do but it looks great, is that art? I’ve seen quite a bit of AI art depicting humans with too many fingers, arms or legs, or legs and arms coming out of weird places, or bending limbs in impossible ways in a video. This seems to show a lack of skill for sure, but art isn’t required to be realistic.
In the 1984 movie Electric Dreams, a PC wrote music that a woman adored, and she fell in love with the owner of the PC thinking he had written it. However, when she discovered that the PC wrote the music she shed a tear, feeling a tenderness toward the PC as well. This seems to me the way a person who has already been touched by the art would react. Maybe she was a little sad, maybe confused, but whatever she was feeling, it brought a tear to her eye.
Without defining art, let’s consider the controversy over AI and art. The first issue is copyright and ownership. It is widely stated that AI models have been trained on works of art — music, literature, images of sculptures, paintings and drawings — scraped from the internet, including copyrighted material, without consent, without payment, and without acknowledgment. [1] This has led to lawsuits and debates over who owns AI-generated art: the user who wrote the prompt, the AI itself, the developers of the technology, or the original artists who claim it was built from their work? [2] The narrow question has now been settled. In Thaler v. Perlmutter the D.C. Circuit held in March 2025 that the Copyright Act requires a human author and that an AI system cannot be the author of a copyrighted work; the Supreme Court declined to review the case in March 2026, leaving that ruling in place. [3] But the ruling is narrower than the headlines suggest. Thaler had told the Copyright Office his image was generated autonomously, with no human contribution at all. The decision does not bar copyright for works made with AI assistance — it holds that works lacking human creative input aren’t eligible. Where the line falls for AI-assisted work is still unsettled: the Copyright Office’s January 2025 report concluded that prompts alone don’t confer authorship, because prompts function as instructions conveying unprotectable ideas, and a pending case, Allen v. Perlmutter, tests whether hundreds of refining prompts change that. [4] The law has answered who can’t be an author. It hasn’t answered how much of a human has to be in the work.
In full disclosure, I recently wrote a book with the help of AI. The story was mine — the characters, the scenes, and most of the dialogue — but the prose was AI. I don’t know how to write a novel and wouldn’t have done it on my own, but with the help of AI the story has come out and is now in the collective. I have been upfront and direct about this collaboration and credit it directly in the foreword to the book.
The Writers Guild of America (WGA) strike against the Alliance of Motion Picture and Television Producers (AMPTP) in 2023 was the culmination of many grievances about changes in the film and television industry, and one of the main points of the strike was protection for writers from generative AI taking their jobs, rewriting their work, or editing it. In an industry long known for finding ways to cut costs, there was immense fear that the studios would replace humans with AI. The 2023 strike ended with an agreement providing writers protections against AI encroachment. But that isn’t where the story ends. In April 2026 the WGA ratified a new four-year deal, and this time the AI safeguards came out notably weaker than the actors’: the studios agreed to notify the guild if they license writers’ work for AI training and to hold ongoing consultation meetings, but there is no requirement to pay writers when their scripts are used to train AI models. Writers appear to have traded AI enforcement for an urgently needed $321 million infusion into their health plan. [5]
Screen Actors Guild – American Federation of Television and Radio Artists (SAG-AFTRA) joined the WGA in the 2023 strike, which helped them secure AI protections for actors, including that their digitized images, voices, or performances could not be used without prior consent and that such use would be fairly compensated. Last month, on June 4, 2026, SAG-AFTRA members ratified their own four-year agreement with the AMPTP by a vote of 91.42% in favor, on a 19.25% turnout, running from July 1, 2026 through June 30, 2030. It restricts wholly synthetic AI performers rather than resolving them: producers may use an AI performer only if it brings “significant additional value” compared to a live actor or that actor’s digital avatar. The catch is that SAG-AFTRA and the AMPTP did not include an agreed-upon definition of “significant additional value” in the contract. The line exists, but nobody has said where it is. [6]
We’ve considered the legal framework in America, but we still haven’t resolved the verdict in the people’s court. The ability to make a firm stand is difficult when the ground you’re standing on keeps shifting. Let’s revisit perception of art. It would intuitively seem that humans can detect AI generated art and reject it — makes me think about drum machines from the 1980s. Sure, they were accurate and fast, but we could always tell when a song was using a drum machine instead of a human drummer. Humans seem to be better at recognizing human generated art, at a rate of 66%, than they are at detecting AI generated art, at a rate of 56%, as discovered in a 2020 study. [7] 56% is pretty close to saying that we aren’t any better at recognizing AI art than flipping a coin…maybe I wasn’t so good at detecting the drum machines.
But that’s not the whole story. People somehow prefer human art anyway — even when they can’t tell which is which. From a Bowling Green study: participants typically didn’t know the difference and admitted they couldn’t tell once asked, but reliably said they liked the human images more without even knowing whether it was AI or not. [8] Wait, not so fast. More recent studies found that with advancements in AI image generation capabilities using OpenAI’s DALL·E 2, humans not only showed a preference for AI generated art [9] but also, when it comes to recognizing real pictures of human faces and AI generated pictures of human faces, it was shown that it was nearly impossible for the subjects to distinguish between the two. [10] The ability for humans to recognize and perceive human faces is one of the most deeply-rooted and evolutionary functions of our brains. It is somewhat disconcerting to discover we are able to be fooled by AI even in this matter. To make matters more complicated, humans perceived the AI generated faces as more trustworthy than the actual human faces!
The authors of the art study posited that AI art has evolved to be more pleasing to us because it has evolved to exploit existing biases in human cognition — trained on works that were already broadly popular, it reflects our own preferences back at us. [9] They explicitly connect this to the research on AI generated faces, where the same hyper-realism that fools us also carries that enhanced sense of trust. [10] If that’s right, then AI art isn’t better than ours. It’s aimed at us — the same way an AI partner engineered to know exactly what we need would be (see my article Humans and AI — A Race to a Connected Future).
AI is here and it’s not going away, and humans are adopting it for a variety of uses in an increasing way (see my article AI Reliability: Requires 20 Watts of Vigilance), which includes art and entertainment. Our ability to tell human generated art from AI generated art turns out to be barely an ability at all, and the ground keeps moving as the systems improve. We will have to figure out and collectively determine what is acceptable and what is not from legal, ethical and societal perspectives. Or determine that it doesn’t matter.
References
#
Source
1
Why Artists Hate AI Art: The Meaning Behind the Debate on Creativity and Control — Art Explained Simply.
https://artexplainedsimply.com/why-artists-hate-ai-art-meaning/
2
Behind the Controversy: Why Artists Hate AI Art — TechRepublic.
https://www.techrepublic.com/article/ai-art-controversy/
3
Thaler v. Perlmutter, No. 23-5233 (D.C. Cir., March 18, 2025) — opinion of the U.S. Court of Appeals for the D.C. Circuit. (Certiorari denied by the U.S. Supreme Court, March 2, 2026.)
https://media.cadc.uscourts.gov/opinions/docs/2025/03/23-5233.pdf
4
Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence — U.S. Copyright Office, Federal Register (March 16, 2023).
5
SAG-AFTRA AI deal: what the new contract means — RedShark News (June 2026), reporting the WGA’s April 2026 four-year agreement and AI terms.
https://www.redsharknews.com/sag-aftra-ai-deal-synthetic-actors-2026
6
SAG-AFTRA Members Approve Four-Year Deal With AI Terms and Pension Merger — Variety (June 5, 2026).
https://variety.com/2026/film/news/sag-aftra-ratification-ai-pension-merger-1236767288/
7
Ragot, M., Martin, N., & Cojean, S. (2020). AI-Generated vs. Human Artworks: A Perception Bias Towards Artificial Intelligence? Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems.
https://dl.acm.org/doi/fullHtml/10.1145/3334480.3382892
8
Samo, A., & Highhouse, S. Artificial intelligence and art: Identifying the aesthetic judgment factors that distinguish human- and machine-generated artwork. Psychology of Aesthetics, Creativity, and the Arts.
https://gwern.net/doc/ai/nn/transformer/gpt/dall-e/1/2023-samo.pdf
9
van Hees, J., Grootswagers, T., Quek, G. L., & Varlet, M. (2025). Human perception of art in the age of artificial intelligence. Frontiers in Psychology, 15:1497469.
https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2024.1497469/full
10
Nightingale, S. J., & Farid, H. (2022). AI-synthesized faces are indistinguishable from real faces and more trustworthy. Proceedings of the National Academy of Sciences (PNAS).