It seems inevitable that AI will surpass humans in all walks of life, but creativity was seen as something that humans would always excel at. In spite of the fact that this is the case, it turns out that Chat GPT-4 may have already beaten humans at divergent thinking, which is one of the core components of creative intellect.
It is important to note that divergent thinking involves coming up with novel solutions in situations where a solution may not be apparent. A recent study gauged the divergent thinking of 151 participants versus Chat GPT-4, and the chatbot provided more in depth answers across the board.
The first of the three tests that were implemented was the alternative use test, which focused on coming up with creative methods for using mundane objects like forks and rope. The second test was called the consequences task, and it was all about answering hypothetical questions such as what would happen if humans didn’t need to sleep anymore. Finally, the divergent associations task require participants to come up with 10 nouns that were as semantically distant as possible.
Once the answers came in, they were judged based on how detailed they were, how long they were, as well as how semantically distinct each individual word was. In each test, Chat GPT-4 managed to provide better answers based on these three metrics. This seems to suggest that the chatbot has the potential to outthink humans in this specific criteria.
However, it bears mentioning that there are other aspects to creative thinking that weren’t addressed in these tests. Human input may also be necessary for the purposes of obtaining the best responses from the chatbot, and it might be interesting to judge the performance of other chatbots. Another factor at play here is that human responses may have been limited by having to do with the real world, whereas AI simply provides to try the answer as best it can even if it’s unrealistic. All of these factors will need to be considered when it comes to testing AI’s creativity in the future.
Image: DIW-Aigen
Read next: Global Employee Rights In The Spotlight: Top 10 Worst Nations For Workers
It is important to note that divergent thinking involves coming up with novel solutions in situations where a solution may not be apparent. A recent study gauged the divergent thinking of 151 participants versus Chat GPT-4, and the chatbot provided more in depth answers across the board.
The first of the three tests that were implemented was the alternative use test, which focused on coming up with creative methods for using mundane objects like forks and rope. The second test was called the consequences task, and it was all about answering hypothetical questions such as what would happen if humans didn’t need to sleep anymore. Finally, the divergent associations task require participants to come up with 10 nouns that were as semantically distant as possible.
Once the answers came in, they were judged based on how detailed they were, how long they were, as well as how semantically distinct each individual word was. In each test, Chat GPT-4 managed to provide better answers based on these three metrics. This seems to suggest that the chatbot has the potential to outthink humans in this specific criteria.
However, it bears mentioning that there are other aspects to creative thinking that weren’t addressed in these tests. Human input may also be necessary for the purposes of obtaining the best responses from the chatbot, and it might be interesting to judge the performance of other chatbots. Another factor at play here is that human responses may have been limited by having to do with the real world, whereas AI simply provides to try the answer as best it can even if it’s unrealistic. All of these factors will need to be considered when it comes to testing AI’s creativity in the future.
Read next: Global Employee Rights In The Spotlight: Top 10 Worst Nations For Workers
