A recent study comparing jokes by humans to those told by ChatGPT suggests that people might need to step up their humor game. Published on July 3 in the journal PLOS ONE, the research aimed to see if artificial intelligence could outwit humans in humor, according to Drew Gorenz, a doctoral candidate in social psychology at USC Dornsife College of Letters, Arts, and Sciences.
“Since ChatGPT can’t feel emotions itself but tells novel jokes better than the average human, these studies show that you don’t need to feel the emotions of appreciating a good joke to tell a really good one yourself,” Gorenz explained.
In one part of the study, participants rated the funniness of novel jokes generated by regular people and by ChatGPT, without knowing the source. The results showed that nearly 70% of participants found ChatGPT’s jokes funnier than those crafted by humans. Around 25% rated human-created jokes as funnier, and only about 5% found jokes from both sources equally funny. These results were consistent across various demographics.
To see how ChatGPT would fare against professional humor writers, the researchers conducted another study where ChatGPT was tasked with creating headlines in the satirical style of The Onion. A separate group of 200 participants rated the funniness of original Onion headlines and ChatGPT-generated ones, finding them equally amusing.
AI Jokes: The First Study
In the first study, both ChatGPT and 105 participants completed three humor tasks, generating over 945 jokes. ChatGPT produced 20 humorous answers for each prompt, resulting in 180 jokes. Participants then rated the funniness of these jokes.
**Investigating ChatGPT and Humor: The Second Study**
For the second study, ChatGPT generated new headlines based on The Onion’s style. A group of 200 participants rated the funniness of both the original and AI-generated headlines, again without knowing the authorship.
Gorenz’s idea for the study was influenced by the ongoing debate in the entertainment industry over the use of large language models (LLMs) like ChatGPT. This issue has been highlighted by recent Hollywood strikes, with writers and actors concerned about the potential threat LLMs pose to their professions and creativity. Gorenz, an amateur stand-up comedian, and his co-author, Norbert Schwarz, an expert on consumer judgment and a co-director at the USC Dornsife Mind and Society Center, noted that while LLMs can enhance everyday communications with humor, they also pose a serious employment threat to professional comedy writers.
Whether AI will get the last laugh remains to be seen.
