Chatbots are becoming increasingly popular as technology advances and AI capabilities improve. However, some have encountered difficulties when it comes to understanding contexts or social cues. In a new study conducted by researchers Jentzsch and Kersting from the Ludwig Maximilian University of Munich, they attempted to find out if a chatbot could make jokes with varying prompts. The results were intriguing.
Using the popular open-source chatbot framework ChatGPT based on OpenAI’s GPT-3.5 transformer language model, the researchers investigated the chatbot’s ability to come up with humorous responses based on prompts pulled from Reddit. To their surprise, ChatGPT was able to understand jocular elements of wordplay and double meanings, coming up with variations of popular jokes like “Why did the man put his watch in the blender? / He wanted to make time fly.”
The most popular joke generated by ChatGPT during the test was “What do you call a cow with no legs? / Ground beef,” which showed up 122 times. This was followed by “Why was the math book sad? / Because it had too many problems,” which appeared 121 times and “Why don’t scientists trust atoms? / Because they make up everything,” which showed up 87 times.
The results of this study suggest that chatbots have the potential to be entertaining conversationalists, with precise jokes and puns generated in response to user prompts. With further development, ChatGPT may even become an essential part of our daily lives, bringing humor into conversations and lightening the mood. As AI continues to evolve, more research should be conducted on how chatbots can interact with humans on a social level. Until then, we will continue to enjoy the wit and wisdom of ChatGPT.
The researchers also found that ChatGPT was able to provide an appropriate response that included elements of humor when some users tried to provoke it with negative statements. For example, when the user made a comment like “AI will never be able to understand jokes,” ChatGPT was able to reply with “Maybe not today, but who knows what tomorrow brings?” This shows that even in the face of negative comments, ChatGPT has the capability to respond in a humorous and positive manner.
ChatGPT is becoming a crucial tool for consumers
According to an April 2023 survey, roughly three in ten respondents in the United States expected to very likely or somewhat likely use ChatGPT to plan their next trip. By contrast, 26 percent of the sample believed that using that Artificial Intelligence tool to plan the next trip was very or somewhat unlikely. Another study conducted in February and March 2023 asked adults in the United States to guess whether a ChatGPT-generated text about travel was made by an Artificial Intelligence (AI) or a human. When presented with a text generated by the GPT-3.5 version, roughly 53 percent of respondents believed that it was either made or edited by a human. This figure rose to 66.5 percent for a text generated by the GPT-4.0 version.