AI content is polarizing. When users spot that AI was used in art, backlash often follows. So, let\u2019s run a quick experiment.<\/p>\n
<\/a><\/p>\n Take a look at the picture below. I just generated an image, in the style of a Monet painting, using AI. I used the latest AI models trained on the world\u2019s best artwork. Have a look, and think about what makes this image inferior to a real Monet painting. Think about what you like and what you don\u2019t like about it.<\/p>\n\n Source<\/em><\/a><\/p>\n Now, what if I told you that I lied? The image is a real Monet. A few weeks back, X user SHLOMS<\/a> performed the same exercise. Let\u2019s talk about what happened.<\/p>\n Table of Contents<\/strong><\/p>\n The Post That Started It All<\/a> <\/a> <\/p>\n Source<\/em><\/a><\/p>\n Over 1,300 people responded to SHLOMS\u2019s post with resoundingly similar feedback. Almost everyone who replied said this AI art was pure crap.<\/p>\n They said it lacked \u201ccoherent composition.\u201d The colors were an \u201cincoherent muddle.\u201d Some said it was \u201ctrying too hard.\u201d<\/em><\/p>\n Others called the image \u201cobvious AI slop.\u201d They called it \u201cgarbage\u201d and \u201cborked nonsense with no sense of space.\u201d Users picked holes in every aspect of the painting. They questioned its lack of meaning and how AI art simply lacks the magical quality imbued by human creativity.<\/p>\n But there\u2019s a problem \u2026 this Monet painting is real.<\/p>\n <\/a> <\/p>\n SHOLMS lied. It was their version of a social experiment. They took a real (much-loved) Monet painting, pretended it was an AI-generated image, and then gauged the reaction.<\/p>\n I\u2019ll be honest. My gut reaction to the X post was the same. I immediately thought it was slop. I tried to spot flaws and thought the colors were dull. And it turns out that all of us do this.<\/p>\n When we think art is created by AI, we tend to dislike it. In fact, when we think anything took no effort to build, we dislike it.<\/p>\n We follow what Andrea Morales<\/a>, Lonnie L. Ostrom Chair in Business and professor in the Department of Marketing at Arizona State University, calls an \u201ceffort heuristic.\u201d In other words, if something took two seconds to generate, we\u2019ll hate it. If it took 15 years of craft to perfect, we\u2019ll love it.<\/p>\n In a 2005 study published in the Journal of Consumer Research<\/em><\/a>, Morales found that people will generally reward actions that took a great deal of effort. The study followed 46 participants who were in the market for a new home. They were told to hire a real estate agent to help them search. The agent reviewed their preferences and put together a personalized list of 10 apartments.<\/p>\n However, 23 of the participants were told the agent \u201cdid everything manually and spent nine hours building the list.\u201d The other 23 were told that \u200b\u201cthe agent used a computer program, spending only one hour on the list.\u201d<\/p>\n\n In reality, the houses on both lists were identical. The real estate agent lied about how much time they spent on the search. It always took one hour.<\/p>\n But perception was swayed by the perceived effort. Participants who were told that the list took nine hours to create rated the houses 30.5% higher. We use effort as a proxy to measure quality \u2014 especially when quality is hard to measure.<\/p>\n\n So when 6.8 million people saw an image of a real Monet painting but thought it was AI-generated, they immediately called it slop.<\/p>\n This presents a tricky contradiction. We\u2019re repeatedly told that to compete, we need to embrace AI. But the Morales study shows a different perspective.<\/p>\n If we only use AI (or even if people think<\/em> we only use AI), people will feel an urge to hate our work. The fantastic copywriter Dave Harland calls this \u201cDeath By Sepia<\/em><\/a>.\u201d<\/p>\n\n Taken from Dave\u2019s \u201cThe Word\u201d newsletter (sign up <\/em>here<\/em><\/a>).<\/em><\/p>\n The landlord at the Bluebell might think this sign is infinitely better than a standard blackboard sign. They might say, \u201cIt\u2019s colorful! It stands out! It includes all of our menu! Even a goblet of wine!\u201d<\/em><\/p>\n But most people will see it and recognize that off-sepia tone as AI. And as we\u2019ve seen with Monet, once people think something is lazily created with just AI, they\u2019ll dislike it.<\/p>\n I spoke with Professor Matt Johnson about why some AI content falls flat on a recent episode of Nudge. <\/em>Listen here.<\/em><\/a><\/p>\n
\n The \u201cAI\u201d Painting That Fooled Thousands<\/a> <\/p>\nThe Post That Started It All<\/h2>\n\n
The \u201cAI\u201d Painting That Fooled Thousands<\/h2>\n