AI Beats Average Humans on Creativity Tests, Study Shows

Artificial intelligence has crossed a new threshold: it can now beat the average person at certain creativity tasks.

Researchers at the University of Montreal tested more than 100,000 people against leading AI systems including ChatGPT, Claude, and Gemini. The results mark a turning point in how we think about machine creativity.

Professor Karim Jerbi led the massive study, published in Scientific Reports. His team used the Divergent Association Task, a psychological test that measures creative thinking by asking participants to list ten words as unrelated as possible. Words like "galaxy, fork, freedom, algae, harmonica, quantum, nostalgia, velvet, hurricane, photosynthesis" score highly.

GPT-4 and other advanced models scored above the average human on these creativity measures. The AI systems showed genuine ability to generate original ideas and unexpected connections.

But here's where it gets interesting. When researchers looked at the top half of creative humans, every single person outperformed the AI. Among the most creative 10 percent of people, the gap grew even wider.

The team then tested whether AI could handle richer creative work like writing haiku, crafting movie plots, and composing short stories. The pattern held: AI beat average performance, but the most talented human creators consistently produced stronger, more original work.

The Bright Side

The research reveals something reassuring about human creativity. While AI has become impressively capable at generating ideas, peak human imagination remains unmatched. The most creative minds among us possess something AI can't replicate.

The study also found that AI creativity can be adjusted. Changing technical settings called "temperature" makes AI responses more adventurous or conventional. Even the way you write prompts matters: asking AI to think about word origins leads to more unexpected associations.

Renowned AI researcher Yoshua Bengio participated in the study. The team designed their tests to be fair comparisons, using identical evaluation methods for humans and machines.

This research doesn't diminish either human or artificial creativity. Instead, it shows us where each excels and suggests we're entering an era where both can complement each other in creative work.