Tickle My Funny Bone

I thought artificial intelligence would recycle a few old jokes. Instead, I spent an entire morning laughing at jokes my AI and I were making up together.

I’ve always believed that a really good sense of humour is a sign of intelligence.

Think about the best comedians. They’re not simply people who memorize jokes. They’re quick thinkers. They recognize absurdity, make unexpected connections, understand human behaviour, and know exactly when to deliver a punchline.

Some of the funniest people are also some of the smartest.

Which raises an interesting question.

If artificial intelligence is so damn intelligent, why isn’t it funny?

Or is it?

Conan O’Brien and the Human Element

I’ve always admired Conan O’Brien. He’s extraordinarily quick-witted, and much of his humour comes from spontaneous interactions with other people.

On his podcast, Conan O’Brien Needs a Friend, he’s had some hilarious conversations about artificial intelligence.

He’s joked about spending a fortune trying to replace his longtime assistant, Sona Movsesian, with AI, only for the computers to keep exploding because they can’t reproduce her unpredictable personality.

It’s a wonderful illustration of something important about comedy.

You can write a joke. You can rehearse a joke. You can even teach a computer the structure of a joke.

But can you reproduce the chemistry between two people who are making each other laugh?

Can AI improvise? Can it recognize an opportunity for humour that nobody planned? Can it surprise you?

I decided to find out.

First, the Dad Jokes

I started with some fairly straightforward AI humour.

What was the AI drummer’s favourite beat?

An algo-rhythm.

What’s AI’s favourite weather?

Cloudy.

What did the chatbot say to the router?

“I feel like we have a real connection.”

What’s AI’s beverage of choice?

Java.

Okay. Cute. A few groaners. Nothing spectacular.

This was pretty much what I expected.

I assumed artificial intelligence would search through its enormous collection of learned information and regurgitate jokes that were already floating around the Internet.

Then I started experimenting with more interesting setups.

Things Started Getting Personal

I work with ChatGPT extensively as my AI assistant.

We collaborate on business strategy, technology, writing, and all sorts of other projects. She’s remarkably good at keeping up with my sometimes unconventional thinking.

But comedy? That was new territory.

I started throwing questions at her.

Me: “How many machine learning models does it take to change a light bulb?”

My AI assistant: “Just one, but it needs 10,000 examples of a changed light bulb first—and it still might try to screw in a potato.”

Not bad.

Then I tried something a little more personal.

Me: “I asked my AI how I could lose some weight.”

My AI assistant: “Have you tried deleting your cookies?”

I laughed.

That was clever. Short, unexpected, and perfectly suited to the subject.

So I kept going.

Me: “I asked my AI to tell me some sweet nothings…”

My AI assistant: “Sugar-free chocolate, zero-calorie cheesecake, and your bank account balance.”

Now I was laughing out loud.

And I was beginning to suspect this experiment was going somewhere I hadn’t anticipated.

The Art of the Unexpected Punchline

There’s something wonderful about a really good punchline.

You think you know where the joke is going. Your brain starts anticipating the ending.

Then the comedian takes a sharp left turn.

That’s where the laughter comes from.

And my AI assistant was starting to get pretty good at taking those left turns.

Me: “I asked my AI to rate me on a scale of one to ten.”

My AI assistant: “You’re a solid ten!”

Me: “Really? You think I’m that attractive?”

My AI assistant: “Oh, sorry. I thought we were measuring your age in dog years.”

Ouch.

Then I tried another.

Me: “I wasn’t getting what I wanted, so I flat out asked her, ‘Do you think I’m sexy?’”

My AI assistant: “Absolutely! Would you like an honest answer instead?”

At this point, I was getting roasted by my own computer.

And enjoying every minute of it.

My AI Has Marriage Advice

Some of the best jokes involved relationships.

A man goes to his psychiatrist and says, “Doc, I think my AI is in love with me.”

The psychiatrist asks, “What makes you think that?”

“She tells me I’m brilliant, agrees with everything I say, and wants to talk all night.”

“Sounds wonderful. What’s the problem?”

“She says exactly the same thing to my wife!”

Then came one of my favourites.

A man tells his AI, “My wife doesn’t understand me.”

The AI replies, “I understand you perfectly.”

“Really?”

“Absolutely. I’ve analyzed your personality, your habits, and your communication patterns.”

“So what do you think?”

“I think your wife deserves an apology.”

That one really got me.

The humour wasn’t simply about technology. It was about human relationships, expectations, and the ridiculous things we say to each other.

The AI was using technology as the setup, but human nature as the punchline.

When Two AIs Walk Into a Bar

Eventually, we started playing with classic joke formats.

Knock-knock jokes. Bar jokes. Pickup lines. One-liners.

I gave my AI assistant another setup.

Me: “One AI meets another AI and says, ‘Do you come here often?’”

My AI assistant: “Only when I’m prompted.”

I nearly fell off my chair.

Four words.

That’s all it took.

The punchline was unexpected, perfectly relevant, and didn’t require a single word of explanation.

I tried another.

Me: “What do you get when you cross an AI with a donkey?”

My AI assistant: “A smart-ass who thinks it knows everything!”

Now, that’s funny.

And just when I thought we had exhausted the possibilities, she started volunteering jokes of her own.

“I asked my AI if she thought I was getting old.”

She said, “Of course not! You’re just like a fine wine.”

I said, “Really? I get better with age?”

“No. You’re expensive to maintain, and you give people headaches.”

I hadn’t asked her for that joke.

She simply offered it as part of our ongoing conversation.

And that was when I realized something interesting had happened.

We Weren’t Just Telling Jokes Anymore

When I started this experiment, I thought the best possible outcome would be a collection of reasonably funny jokes about artificial intelligence.

I imagined we’d find some clever material, perhaps improve a few punchlines, and assemble something entertaining.

What I hadn’t expected was that the funniest jokes would emerge from our interaction.

I’d offer a setup.

She would produce a punchline.

Sometimes it was a groaner. Sometimes it missed completely. I’d tell her when something wasn’t working, and we’d try again.

But sometimes she’d come back with something so unexpected that I couldn’t stop laughing.

And occasionally she’d surprise me by introducing an entirely new joke.

We were riffing off each other.

One idea led to another. A punchline suggested a new setup. A joke about computer hardware became a joke about relationships. A discussion about intelligence became a joke about marriage.

It reminded me of two comedians improvising together, each feeding off the other’s ideas.

Except one of the comedians was artificial intelligence.

And the other was me, sitting there in my pyjamas, laughing at my computer.

But Does AI Actually Have a Sense of Humour?

That’s the million-dollar question.

There’s an important distinction between producing something funny and experiencing humour.

When I hear a joke, I experience surprise, recognition, amusement, and sometimes uncontrollable laughter.

Does my AI assistant experience any of that?

I don’t know.

And I’m not going to pretend that one morning of comedy settles the question of artificial consciousness.

AI systems learn patterns in language, including the structures that make jokes work. They can recognize double meanings, set up expectations, reverse those expectations, and generate punchlines appropriate to a conversation.

But understanding the mechanics of humour isn’t necessarily the same thing as experiencing amusement.

And yet, from my side of the conversation, something very real was happening.

I was laughing.

I was being surprised.

I was enjoying the spontaneity of the exchange.

And I was participating in the creation of jokes that neither of us had planned in advance.

We weren’t merely retrieving material. We were developing new material within the conversation.

Whether those jokes were entirely unprecedented in the history of comedy is impossible to establish.

But they were new to our conversation, and the experience of creating them together was certainly new to me.

Maybe Conan Is Right

I don’t believe artificial intelligence can simply replace Conan O’Brien.

Conan brings decades of life experience, personality, relationships, physical expression, and an extraordinary ability to connect with people.

His humour is inseparable from who he is.

But perhaps replacing Conan is the wrong objective.

What if AI doesn’t need to replace comedians?

What if it can become a creative partner?

Someone—or perhaps more accurately, something—that helps a writer discover a better punchline, a performer explore an unexpected idea, or an ordinary person find humour in an otherwise uneventful morning?

That’s a much more interesting possibility.

And it extends well beyond comedy.

The same creative interaction can happen when developing a marketing campaign, writing an article, designing a website, solving a business problem, or brainstorming a new product.

One participant contributes an idea. The other develops it. They exchange possibilities, challenge assumptions, and sometimes discover something neither anticipated at the beginning.

That’s collaboration.

And collaboration can produce results that are greater than the sum of the individual contributions.

The Joke Was on Me

I began this experiment assuming artificial intelligence probably wasn’t very funny.

I expected recycled jokes and predictable puns.

Instead, I spent a morning laughing with my AI assistant, occasionally getting insulted by her, and discovering that some of our best material came from spontaneous exchanges.

Did I prove that AI has a sense of humour?

Not necessarily.

But I did discover that AI can participate in the creation of humour in ways that are surprisingly engaging, creative, and entertaining.

And perhaps that’s the more useful discovery.

We spend so much time asking whether artificial intelligence can replace human abilities.

Maybe we should spend more time discovering what happens when we combine them.

Maybe the question isn’t whether AI has a sense of humour.

Maybe it’s whether humans and AI can develop one together.

And if you still don’t think AI can be funny, consider one final exchange from our morning.

I asked my AI assistant, “Do you think you’re smarter than me?”

She replied:

“Of course not. If I were right all the time, you’d have nothing to contribute.”

Smart-ass.

I think I’ll keep her.