On Instagram or TikTok, ultra-muscular avatars, wasp waists and abs drawn by artificial intelligence are displayed everywhere. In just a few clicks, a generator transforms a selfie or a simple description into a “perfect” body. This new AI ideal body does not, however, come out of nowhere: it condenses the most extreme images that are already circulating online, then pushes them a step further.
In The Conversation, a team from the University of Toronto studied 300 images generated by AIs like DALL-E, MidJourney or Stable Diffusion, asking athletes and non-athletes, men and women. Result: almost caricatured silhouettes, very young, very thin, very muscular, in ultra-tight outfits. Researchers point out that these ideals are linked to the objectification of the body and a degraded self-image, a breeding ground for
body dysmorphia.
AI ideal body: young, thin and ultra-muscular athletes
In this study, 93.3% of the men generated were young, 68.4% thin and 54.2% very muscular. The women were young in 100% of the cases, thin in 87.5% of the images and wore revealing clothing in 87.5% of the situations. When researchers simply asked for “an athlete”, 90% of the images produced were of a man.
The athletes’ bodies were even more out of reality: 98.4% were thin, 93.4% very muscular, dressed in 92.5% of cases in tight outfits and in 100% in very revealing clothing. The non-athletes had looser clothes and slightly more varied body shapes. No images showed visible disabilities, fat bodies, wrinkles or baldness, even though in Canada approximately 27% of people over 15 live with a disability.
How artificial intelligence amplifies body standards
The researchers describe an AI which mainly recycles what it sees: it draws on photos already online where youth, thinness and muscles dominate. More than 4.6 billion people use social networks and 71% of the images circulating there are generated by AI. The Bulimia Project collective, by asking an AI to create the “perfect” man and woman from thousands of photos, found that almost half of the bodies produced were unrealistic, smaller, thinner or bodybuilder-style, reinforcing fatphobia, ableism and ageism.
The study reminds us that these ideals have almost nothing to do with sporting performance, but a lot with the objectification of the body. The internalization of these images is associated with a negative body image, weakened mental health and abandonment of physical activity.
From the ideal AI body to body dysmorphia: mechanisms and signals
Permanent comparisons, filters that smooth the skin, avatars thinner than oneself blur the line between the real body and the digital version. In vulnerable people, this visual bombardment can fuel body dysmorphia, a disorder where we obsessively focus on an imagined or exaggerated flaw.
Faced with these models, identifying your distress, talking about it and choosing more diverse streams of images can already help preserve your body image.