Recent research from the University of Southern California (USC) has found that OpenAI's ChatGPT 3.5 can generate humor that surpasses that of the average human and even matches the quality of professional comedy writers. The study aimed to evaluate how well the AI could perform comedic tasks requiring creativity and timing, which are typically challenging for humans. With participants preferring ChatGPT's humorous submissions 69.5% of the time, it's clear that AI is making significant strides in creative fields traditionally dominated by humans.
The study consisted of two parts. In the first part, participants were tested on their ability to create humorous phrases, fill in the blanks comically, and roast in a fictional scenario. ChatGPT was given the same tasks, and its responses were then rated by another group. The AI's performance was impressive, with a majority favoring its humor over that created by humans. Notably, ChatGPT excelled in the roast joke task, where its humor was rated higher despite the task's aggressive nature.
In the second part of the study, researchers shifted focus to satirical news headlines. ChatGPT was tasked with producing satirical headlines similar to those published by The Onion. Once again, participants rated the AI-generated content on par with that from professional writers. While nearly half preferred The Onion's original headlines, a significant 36.9% found ChatGPT's versions funnier, showing the AI's potential in producing high-quality satirical writing. These findings highlight substantial implications for the comedy industry, where AI could pose a real threat to employment opportunities for professional writers.