AI chatbots like ChatGPT and Google Gemini lack humor and produce bland, overly politically correct jokes according to a study by Google’s DeepMind. In the study, 20 comedians tested the chatbots and found that they struggled with creativity and self-censorship, often avoiding dark humor and offensive content. The chatbots were also noted for being too politically correct and for not being able to write jokes about people outside of the Western, white, heterosexual, male mainstream.
The inability of these chatbots to tell jokes is a significant issue for companies like Google and Amazon who want their chatbots to be engaging enough for users to spend time with and potentially purchase premium versions. Humor has become a key component in the AI arms race, with companies like Elon Musk’s Grok and Amazon-backed Anthropic working on making their chatbots more conversational and humorous.
Improving humor in chatbots is a priority for companies in the generative AI market, with initiatives like OpenAI’s GPT-4o being developed to create chatbots that can understand and respond to humor. As more companies invest in this technology, the ability to create engaging, humorous chatbots will continue to be a focus of development in the AI industry.
Article Source
https://www.businessinsider.com/deepmind-researchers-ai-chatbots-bad-humor-not-funny-jokes-comedians-2024-6?amp