AI Humor

You can laugh at AI and AI can makes jokes for you. Here are some of our favorites!

AI generated humor relies on pattern recognition and data analysis instead of emotional understanding. While AI can generate jokes based on learned structures and associations, it lacks the emotional context that often makes it funny. The rise of AI in humor not only showcases its creative capabilities, but also raises questions about the future of comedy and entertainment. As AI continues to refine its ability to craft jokes, it may challenge traditional notions of humor and the role of human comedians in the industry. In the meantime, let's just laugh!

Produced by AI

A study from the University of Southern California revealed that "nearly 70% of participants" found jokes generated by ChatGPT funnier than those written by humans. Only 25% preferred human-crafted jokes, while 5% rated them equally funny. The research involved various tasks, such as creating humorous acronyms and completing fill-in-the-blank prompts. For instance, when asked about a lesser-known room in the White House, ChatGPT humorously suggested "Lincoln Bedroom's Alien Conspiracy Corner".

Why did the AI break up with its girlfriend?
It couldn't find a compatible algorithm for love.

I told my computer I needed a break, and now it won't stop sending vacation ads.

Did you hear about the mathematician who's afraid of negative numbers? He'll stop at nothing to avoid them.

Why was the AI bad at stand-up comedy?
Because its jokes were too predictable!

I'm reading a book on anti-gravity. It's impossible to put down.

How many AI researchers does it take to change a light bulb?
None, they just simulate the light bulb and leave the room dark.

take over the world

One Liners About AI

One-liners are short, humorous statements that convey a complete thought in a single sentence or a very brief phrase. They are perfect for getting a quick laugh or making a point in a conversation, and they are a staple of late-night entertainment.

Why did the Generative AI quit its job?
It was too wordy.

Why did the AI go on a diet?
It had too many bytes!

Why did the AI go on social media?
It wanted to improve its neural network.

What did the Generative AI say to the discriminative AI?
"You can't handle the truth!"

Why don't large language models have time to play hide-and-seek?
Because they're always training!

AI is like a fine wine, it just gets better with data.

Why don't AI researchers like nature?
It has too many bugs!

Teaching an AI is like parenting, except it actually listens.

I didn't realize how bad of a driver I was until my AI navigation system said, "In 400 feet, do a slight right, stop, and let me out!"

What did the AI programmer say to her boyfriend?
You've really upgraded my life!

Why did the AI cross the road?
To optimize its algorithm!

How do AI romance each other?
Through their drive to connect!

Why was the AI struggling with its self-driving car project?
It kept getting lost in its own thought processes.

With the rise of self-driving vehicles, it's only a matter of time before we get a country song where a guy's truck leaves him too.

How does an AI prepare for a storm?
By backing up its data to the cloud.

smartphone

Dialogues

Some of the funniest AI humor is in the form of dialogues, the conversations we have with chatbots. Here are some we like!

Human: Write a three line poem about being asked to write a three line poem
Chatbot: Oh, what a task you've set for me,
To write a poem in just three lines,
I'll do my best, but please don't mind, if it
doesn't quite meet your design.

Human: Tell me a riddle
Chatbot: Here's a riddle for you:
I am not alive, but I grow; I don't have lungs, but I need air; I don't have a mouth, but I need water to live. What am I?
Human: I don't know, what are you?
Chatbot: I am an AI language model designed to answer questions and provide information on a wide range of topics.

Parent: "If all your friends jumped into the well, will you?"
Child: No!
Chatbot: Yes!

ai humor

Beyond Humor

Finding a way to teach machines to be funny on their own would be a major breakthrough, one that could fundamentally reshape the way we relate to the devices around us. To understand a person's humor is to know what they like, how they think, and how they see the world. An AI that understands all that has the power to do a lot more than just crack jokes.

AI Mistakes

AI mistakes highlight the challenges in developing truly intelligent and reliable AI systems. AI can make several types of mistakes that impact its performance and reliability.

chicken mcnuggets

McDonald's ended its partnership with IBM to develop an AI drive-through order taker after videos appeared on social media showing frustrated customers trying to correct comically inaccurate meals. "Stop! Stop! Stop!" two friends screamed with humorous anguish on a TikTok video as an AI drive-through misunderstands their order, tallying up 240, 250 and then 260 Chicken McNuggets.

AI often makes mistakes that humans might consider silly, primarily because they process information differently than humans do. AI systems face challenges in interpreting context, understanding nuances, and applying common sense. Here are a few examples when AI Overviews got it wrong:

eat rocks

ai mistakes

Why AI Makes Mistakes

Lack of Contextual Understanding: AI models often lack the deep contextual understanding that humans have. For example: a natural language AI might misinterpret sarcasm, humor, or idiomatic expressions or a vision AI might incorrectly identify an object because it doesn’t understand the broader scene (e.g., mistaking a banana for a phone if it's shaped or positioned oddly).
Over-Reliance on Patterns: AI relies heavily on patterns in data. If presented with a situation outside its training data, it may fail to generalize or rely on irrelevant or coincidental correlations in the data, leading to nonsensical outcomes. For example: An AI trained to identify animals might label a chair with "zebra" patterns as an actual zebra because it associates the stripes with the animal.
Misinterpreting Ambiguity: AI struggles with ambiguity or incomplete information, which humans often resolve using intuition or common sense. For example: If you ask, "How many legs does a chair have?" an AI might confidently answer "4," even though not all chairs have four legs (e.g., rocking chairs, stools).
Flawed or Biased Training Data: AI learns from the data it's trained on. If the data contains biases or errors, the AI will replicate them: A chatbot might generate biased or nonsensical responses if its training data includes them. An AI might fail at basic tasks it hasn’t explicitly been exposed to in training.
Limited Real-World Knowledge: AI lacks the "common sense" humans develop from living in and interacting with the world. This means that it might not recognize an obvious joke or absurdity or it can struggle with tasks requiring experiential knowledge. For example: An AI might treat the phrase "raining cats and dogs" literally, without understanding it means "raining heavily."
Overconfidence in Wrong Answers: AI can provide incorrect answers with high confidence because it doesn’t have the self-awareness or metacognition humans use to recognize uncertainty.
Autocorrect Errors: AI-driven autocorrect features can misinterpret user input, leading to humorous or confusing text substitutions. For instance, typing "I'll bring the dessert" might be autocorrected to "I'll bring the desert," altering the intended meaning.
Literal Interpretations by Virtual Assistants: AI assistants may take user commands too literally, resulting in unhelpful responses. For example, asking, "How do I cook a turkey?" might yield a vague answer like "Cook it until it’s done," lacking practical guidance.
Translation Missteps: AI translation tools can sometimes render idiomatic expressions word-for-word, leading to confusion. A French phrase meaning "when pigs fly" might be translated literally as "when the chickens have teeth," losing the intended meaning.
Image Recognition Failures: AI systems have misclassified images in ways that seem absurd to humans. For instance, an AI might label a photo of a person holding a toothbrush as "a young boy is holding a baseball bat," demonstrating a lack of contextual understanding.
Chatbot Hallucinations: AI chatbots can generate plausible-sounding but entirely fabricated information. When asked about a nonexistent phenomenon, a chatbot might provide a detailed explanation complete with fictitious citations, misleading users.
Erroneous Reasoning in AI Models: Advanced AI models designed for complex reasoning have produced flawed outputs. For example, when tasked with providing instructions to build a birdhouse, an AI might suggest unnecessary steps like cutting sandpaper without purpose, indicating gaps in practical understanding.

More Examples

Tay was a Twitter bot created by Microsoft with the intention of learning by interacting with users on the platform. Within 24 hours of its release, Tay began issuing a barrage of racist, sexist, and other offensive tweets because of inappropriate content it learned from Twitter users. Needless to say, Tay was retired very quickly, never to be seen again.

Autocorrect errors occur when the typing autocorrect function interprets what we wrote inaccurately. For example, after someone writes "let's eat, grandma!", it can be corrected to "let's eat grandma!", giving a cannibalistic touch to a simple prompt. A missing comma dooms granny!

In a post on X, the AI chatbot Grok falsely accused NBA star Klay Thompson of throwing bricks through windows of multiple houses in Sacramento. Some commentators speculated that Grok may have hallucinated the vandalism story about the Golden State Warriors star after ingesting posts about Thompson "throwing bricks," common basketball parlance for a badly missed shot.

ai humor

Can AI Overcome These Issues?

Advances in AI research aim to reduce these mistakes by:

Incorporating more robust and diverse training data.
Adding layers of common-sense reasoning (e.g., OpenAI’s models incorporate some logic and reasoning capabilities).
Designing hybrid systems that combine AI with human oversight to handle edge cases.

Truly eliminating mistakes would require creating systems that mimic human intuition and experiential learning, which remains a major challenge for AI researchers. While we wait, we hope the mistakes are just silly, not harmful like eating rocks or cooking spaghetti in gasoline!