A fascinating art social experiment unfolded on social media this week after someone shared an actual Monet painting as an AI-generated artwork and asked people to explain what makes the “AI image” inferior to a genuine Monet piece. There was no shortage of “sharp-eyed” critics eager to chime in.
It all started after X user @SHL0MS posted the painting and wrote: “I just generated an image in the style of a Monet painting using AI. Please describe, in as much detail as possible, what makes this inferior to a real Monet painting.”
The user even marked the post with X’s “Made with AI” label to add to the deception.
In reality, the painting is one of the 250 oil paintings in the renowned French Impressionist painter Claude Monet’s Water Lilies series in which he depicted scenes from his home flower garden over the final 31 years of his life.
The painting is one of the 250 or so paintings in the famous “Water Lilies” series by French painter Claude Monet.
Critics, however, were eager to point out all kinds of “obvious” details that show why the “AI” Monet can’t hold a candle to a genuine Monet. One person even took the time to write out an 850-word breakdown of the AI work’s shortcomings.
“I’m disappointed I have to even point it out,” writes @egg_oni. “There is no cohesion to the depth and color choices. The reflection of the tree bleeds into the lilypads with no regard for spatial depth or contrast. The background lilypad-algae amalgam is egregiously vague, like most AI art.”
“The reflection in AI art is just noise splattered right,” writes @jordoxx. “Monet actually understood how light behaves on water.”
“The choice of color in places e.g. the purple around the lily pads sticks out to me as decidedly worse than most Monet,” writes @0xchiefyeti. “I get a sense that the artist failed to connect their eyes to the brush/palette […]”
“No frame, no sense of the threshold between subject and object, just colors,” writes @robertjett_.
“I would say that the allegedly real one here is superior in the sense that it carries, and conveys more information than the artificial one,” writes @artprograce. “The dark cold reflection of the trees triggers my attention. They strike me as slightly off, too dirty, and too pronounced to be natural.”
“I’m no artist but a real Monet actually looks like a real place…” writes @amaldorai. “the further back you get in this picture the less it looks like anything at all.”
“Depth, contrast, and cohesion are the most obvious,” writes @para_dim3. “There’s also no clear focal point.”
“It feels less lively,” writes @AzuriSplashes. “It lacks the texture, the rugged edges, the folds, the crevices and creases and bevels and topology of plastic arts. The fine, calculated highlights. The AI version is granulated pixelation, and it looks that way, it lacks the mess of humanity.”
“The fact that it looks like s**t and is s**t,” writes @RDL0013 in a since-deleted reply. “Slop. Doesn’t look anywhere near like a Monet. Looks exactly like somebody trying to replicate style and achieving like 20% of it. Not as vibrant as Monet’s typical choice of colors. Looks dull.”
“There’s no coherent composition,” writes @HundtRichard. “The eye is drawn to the 1/3rd from bottom, 1/3rd from left region and there’s nothing really to focus on. The lilly’s contrast is too low and the negative space around it too cluttered. The surface texture in the water regions are too vertical.”
“[T]here is no consistency in colour choice,” writes @Polymind_. “The view looks obscured perspective wise and feels like there is too much detail in the AI version, which if I am thinking correctly comes back again to the colours being so distinct and contrasty.”
“As an amateur art enjoyer, the only criticism I can offer is that the AI generated image does not make me feel anything,” writes @ThrosturTh. “It does not conjure emotion, thought or wonder. It’s just a colorful wallpaper pattern. If you look up ‘monet painting’ in Google images, you feel something.”
“There’s a certain harshness, no soft blending of colors, no depth, no symbiosis of the elements,” writes @JesTer396.
“The AI seems to be unable to distinguish plant reflections and submerged plants, for one,” writes @DavyRogue27930. “It’s combining tokens from the two randomly and the result is an incoherent muddle of inconsistently saturated greens.”
“Spatial coherence,” writes @enfilmigult. “The phony gen-AI pic isn’t getting it right and the reflections look like they’re growing out of the water. You look at the painting and instantly see the angle of the water surface. Also those lily pads are hideous, looks like someone drew on them.”
“Because it’s crap. That simple,” writes @nightingale9181. “This ain’t no painting. No talent to it. AI needs to go.”
“I present you with my eye lines, thickness denotes how quickly my eye moved,” writes @KEMOS4BE in a since deleted post, which included helpful illustrations. “One has a sensible, meandering composition that fits the subject.”
As the post began to go viral, many of the insightful critics began deleting their replies, but thankfully @SHL0MS and other users such as @Jediwolf took screenshots of some of the best replies before they disappeared.
People are pointing out that results of this silly experiment are in line with what studies have shown about how people perceive art differently in light of how it was produced. The famous 2004 Kruger study into something called the effort heuristic found that people liked and valued artworks more if they believe they took more time and effort to create.
There is also a natural human bias against AI. A 2024 study published in Nature found that while people generally prefer AI-generated artworks over human-made ones when they didn’t know they were AI-generated, they preferred AI art less after finding out that AI was behind it.
“Participants were unable to consistently distinguish between human and AI-created images,” write researchers Simone Grassini and Mika Koivisto in the article titled “Understanding how personality traits, experiences, and attitudes shape negative bias toward AI-generated artworks”. “Furthermore, despite generally preferring the AI-generated artworks over human-made ones, the participants displayed a negative bias against AI-generated artworks when subjective perception of source attribution was considered, thus rating as less preferable the artworks perceived more as AI-generated, independently on their true source.
“Our findings hold potential value for comprehending the acceptability of products generated by AI technology.”
It would be interesting for someone to now conduct the same experiment with photographs, perhaps with an obscure photo by Ansel Adams, for example. Given what science is showing about negative human feelings toward AI artwork, the results would presumably be just as hilarious.