“Researchers started by asking three AI chatbots—OpenAI’s ChatGPT and GPT-4 as well as Copy.Ai, which is built on GPT-3—to come up with as many uses for a rope, a box, a pencil, and a candle as possible within just 30 seconds.”
– Rhiannon Williams
MIT Technology Review reports on a study demonstrating how AI systems generate abstract concepts that often show greater creativity than humans. The researchers blindly assessed creative concepts from AI systems and 256 people.
The average scores for concepts generated by the AI models were judged to be higher than those from humans, but the highest scores were for human-generated ideas.
The results leave unresolved the question of whether AI systems are, on average, more creative than humans or more adept at doing tests for creativity. The researchers say the study usefully adds to understanding relationships in how humans and AI models process original ideas.
The researchers are from universities in Finland and Norway.
AI just beat a human test for creativity. What does that even mean? | MIT TECHNOLOGY REVIEW | September 14, 2023 | by Rhiannon Williams
SEE ALSO: The complete study published in Scientific Reports