New Benchmark Tests Show AI Far Inferior to Humans in UI Creativity
A new study introduces the Design Creativity Bench to measure AI creativity in UI design. In tests, GPT-6 Astra led AI models with a score of 65.5, though human designers scored much higher at 92.3. Models lagged far behind humans in originality and creative range but slightly outperformed them in meeting requirements, with the top model achieving a 99.2% compliance rate—higher than humans' 98.0%.
New Benchmark Reveals AI Lags Far Behind Humans in UI Creativity
A new study introduces the Design Creativity Bench, which measures AI creativity in UI design. In the tests, GPT-6 Astra led AI models with a score of 65.5, while human designers scored as high as 92.3.
The models scored far below humans in originality and creative range, but performed slightly better at meeting requirements. The best model achieved a 99.2% compliance rate, surpassing humans at 98.0%.
Design Creativity Bench | arXiv
https://arxiv.org/abs/2610.11539
