ChatGPT Image Generator vs. Comp Climbing
The general public is often confused about many aspects of our sport, and apparently, so is AI

ChatGPT’s latest image generator tool recently went viral for its ability to turn photos into shockingly detailed recreations in a variety of cartoon styles, from Studio Ghibli and old-school stop animation to Disney and Pixar. We fed the model some photos of the world’s top climbers at the Olympics and last year’s World Cup circuit to see how it would fare in understanding the complex terrain and body positions of lead climbing and bouldering.
The results are part genuinely impressive, part hilarious. The model does an amazing job at recognizing and recreating facial expressions, general body positions, clothing, the colour of surroundings, and surprisingly, chalk bags and climbing holds. It has more trouble recreating functional harnesses, ropes that make sense, quickdraws and carabiners, and chalk. Check out the results below, featuring photographs of Natalia Grossman, Janja Garnbret, Adam Ondra, and Toby Roberts.
Real Natalia Grossman
“Studio Ghibli” Natalia Grossman
Real Janja Garnbret

“Studio Ghibli” Janja Garnbret
“Pixar” Janja Garnbret
Real Adam Ondra

“Studio Ghibli” Adam Ondra
“Rankin/Bass” Adam Ondra
Real Toby Roberts
