I wanted to share some thoughts on this question: does AI already outperform humans on creative tasks? Two new academic studies now indicate that in certain defined areas, this is becoming a reality.
A study from the Wharton School at the University of Pennsylvania focused on comparing a set of human-generated business ideas against those produced by machines. The intriguing result was that, of the top 40 most outstanding ideas, a staggering 35 were generated by GPT-4 models.
Another study published in Nature echoed these findings, demonstrating that AI models performed above the human average in tasks involving divergent thinking. While the best human ideas still outperformed the top AI models, the overall trend suggests AI’s ascending role in specific types of creative processes.
This is an interesting set of results, with important ramifications.
I dive deeper in the video1 above — and in the abbreviated notes below for paying members of Exponential View.
