Key Points

The post discusses a “pelican riding a bicycle” benchmark and the community’s reaction, suggesting that automated test suites should be added to stack these tests over time.
It confirms that ChatGPT Images 2.0 independently generated a “WHY ARE YOU LIKE THIS” road sign, even though the sign was not explicitly requested in the prompt.
The example prompt describes a highly chaotic chain of animals/objects (horse → astronaut → pelican → bicycle), yet the model added an extra visual element on its own.
Overall, it highlights how generative image models can inject additional details, making reliable benchmarking and regression testing important.

Simon Willison’s Weblog

Sponsored by: Sonar — Now with SAST + SCA for secure, dependency-aware Agentic Engineering. SonarQube Advanced Security

25th April 2026

@scottjla on Twitter in reply to my pelican riding a bicycle benchmark:

I feel like we need to stack these tests now

I checked to confirm that the model (ChatGPT Images 2.0) added the "WHY ARE YOU LIKE THIS" sign of its own accord and it did - the prompt Scott used was:

Create an image of a horse riding an astronaut, where the astronaut is riding a pelican that is riding a bicycle. It looks very chaotic but they all just manage to balance on top of each other

Posted 25th April 2026 at 4:44 pm