 |
ML Newsnew | past | best | rss | submit | about |
| ChatGPT is fun, but not funny. Humor is still challenging Large Language Models (arxiv.org) | |
| 3 points by weetniet on Jun 10, 2023 | hide | past | pdf | 1 comment on HN | |
|
In plain words: ChatGPT was asked to write, explain, and spot jokes using simple prompts to see how well it handles humor. Of 1,008 jokes it produced, over 90% were just the same 25 jokes repeated, showing it recycles rather than invents. Abstract · ChatGPT is fun, but it is not funny! Humor is still challenging Large Language ModelsHumor is a central aspect of human communication that has not been solved for artificial agents so far. Large language models (LLMs) are increasingly able to capture implicit and contextual information. Especially, OpenAI's ChatGPT recently gained immense public attention. The GPT3-based model almost seems to communicate on a human level and can even tell jokes. Humor is an essential component of human communication. But is ChatGPT really funny? We put ChatGPT's sense of humor to the test. In a series of exploratory experiments around jokes, i.e., generation, explanation, and detection, we seek to understand ChatGPT's capability to grasp and reproduce human humor. Since the model itself is not accessible, we applied prompt-based experiments. Our empirical evidence indicates that jokes are not hard-coded but mostly also not newly generated by the model. Over 90% of 1008 generated jokes were the same 25 Jokes. The system accurately explains valid jokes but also comes up with fictional explanations for invalid jokes. Joke-typical characteristics can mislead ChatGPT in the classification of jokes. ChatGPT has not solved computational humor yet but it can be a big leap toward "funny" machines. | |
| add comment on HN |
When I asked it “why did the chicken cross the road?” It responded: “The traditional answer to that question is simply "to get to the other side." However, if you'd like a more imaginative or creative answer, we can explore some possibilities. For example, perhaps the chicken was seeking adventure, searching for food or water, escaping from a predator, or even joining its fellow chickens on the other side. The reason can vary depending on how you'd like to interpret it!”
So then I replied: “If I ask you “why did the chicken cross the road? Can you simply reply “why?”.”
And ChatGPT said: “Yes, if you ask me "why did the chicken cross the road?" and I reply with "why?" it can be seen as an open-ended response to prompt further exploration or provide an opportunity for you to offer your own interpretation or explanation for why the chicken crossed the road. It can add a playful or thought-provoking twist to the conversation.”
Me: “why did the chicken cross the road?”
ChatGPT: “why?”
Me: “it didn’t”.
ChatGPT: “Ah, I see! So, in this case, the answer to "Why did the chicken cross the road?" would be "It didn't." It seems like a clever twist on the traditional joke. Thank you for sharing that variation! Is there anything else I can assist you with?”