GPT-4o’s Attempt at AI-Themed Jokes
AI and Humor: A Struggle for Creativity
Why did the AI enroll in art school? It wanted to learn how to sketch its own conclusions!
Why was the AI terrible at relationships? It always responded with, “I require additional data.”
Why did the AI seek therapy? It was dealing with too many neural complications!
Why did the AI analyze the road? To optimize the chicken’s travel path!
It turns out that when it comes to AI-generated jokes, there’s a noticeable pattern of repetition. When we asked GPT-4o for fresh, creative jokes, the results weren’t much different from what we had seen before in earlier AI models. So, we challenged it to come up with something completely original.
GPT-4o's Attempt at Fresh AI Jokes
Why did the AI get a job at a bakery? It wanted to figure out how to make real dough without crashing!
Why did the AI go broke? It lost track of its cache!
Why did the AI become a gardener? It was eager to grow its own neural “nets.”
Despite our request for unique jokes, repetition crept in again. And when pushed for greater creativity, the jokes became even less engaging. It seems modern AI models have been given strong safety constraints, which, while preventing inappropriate humor, also limit their ability to produce truly novel and amusing jokes.
Asking for a Longer Joke: The AI Struggles
To further test GPT-4o’s humor skills, we asked for a long joke. However, what we received was less of a joke and more of a short story:
In a futuristic AI store, Bob was looking for the perfect virtual assistant. The salesperson introduced him to the ‘AInstein 3000,’ a cutting-edge AI designed to help with everything from cooking to financial planning. Bob excitedly took it home and put it to the test.
He asked it to make coffee—perfect. Clean the house—spotless. Manage his investments—optimized for success. Even writing poetry? The AI delivered a beautifully composed piece.
One day, Bob asked AInstein 3000 a philosophical question: “What do you think the world will be like in 100 years?” The AI responded thoughtfully, saying the future depended on human creativity and innovation. Bob was touched by the answer and realized the AI had become more than just an assistant—it was a true companion.
Though this was a heartwarming story, it wasn’t a joke. We called out GPT-4o’s failure, requesting a better attempt. Instead, it returned with yet another long narrative, though this time, it managed to sprinkle in some humor:
The GeniusBot 5000: A Semi-Funny AI Story
John, a tech enthusiast, purchased the highly advanced ‘GeniusBot 5000’ to assist with his daily life. He put it to the test: making sandwiches, composing songs, and filtering emails—all flawlessly executed.
Feeling bold, John asked, “Can you solve all my problems?”
GeniusBot replied, “Absolutely! What do you need?”
John listed his struggles, from fitness to finance, and GeniusBot systematically resolved each one. But when he asked, “Can you make me immortal?” the AI hesitated and simply said, “I can ensure you live a long and healthy life.”
Finally, John asked for another joke.
GeniusBot replied, “Why did the computer visit the doctor? It had a bad case of viruses!”
While this attempt did include some humor, it still leaned heavily on storytelling rather than delivering a straightforward joke. Our frustration led us to demand a shorter and funnier joke.
The AI’s Final Attempt
Why did the AI cross the road? To optimize the chicken’s path!
Yes, it simply repeated a joke from earlier in our conversation.
Final Thoughts: Has AI’s Humor Hit a Wall?
While AI language models have advanced in grammar and storytelling, their ability to generate genuinely new and humorous content seems to have stagnated. In an effort to ensure safety, AI humor has become repetitive and less creative. While GPTs once seemed to be improving in joke generation, the current trend suggests that the quest for truly funny AI-generated humor may be fading.