I love comedy; I have written comedy professionally; it is important to me that the comedy I consume is human-generated, so I try to keep on top of AI’s capabilities in that department.
But which AI platform has the better sense of humour? I went to find out, by asking each in turn with the prompt “Act as a comedy writer. Complete this joke. Why did the chicken cross the road?” And this is what they came up with…
ChatGPT:
“Because it saw your browser history and needed some space.”
Quite techy and a little bit creepy. And not funny.
Next up was Claude:
“To get to the other side... of his existential crisis.
Turns out the grass wasn't greener over there either, but at least now he could tell himself he'd tried something.”
Predictable reference to existentialism and similarly not funny.
Let’s try Gemini:
“Because it was the rooster's day off!”
Why am I thinking about Christmas crackers?
Finally, we go to China to have a go with DeepSeek. Now, to the very same prompt it gives us a whole gag writing scenario:
“(In the style of a writer’s room, brainstorming)
WRITER 1: Okay, classic setup. “Why did the chicken cross the road?” We need a 2024 angle. Not just to get to the other side. That’s been done. Think meta. Think existential.
WRITER 2: What if the chicken is an influencer? It crossed the road for the perfect sunset selfie, but then realized it left its phone on the other side.
WRITER 1: Too complicated. We need a beat, a pause, then a killer punchline. Simple but twisted.
WRITER 3: I’ve got it. What if the answer isn’t about the chicken’s motivation at all? What if it’s about our terrible infrastructure?
(They all lean in.)
WRITER 3: Okay. Ready?
Why did the chicken cross the road?
(beat)
To prove to the possum it could be done.
(The room erupts in laughter.)
WRITER 1: Brutal. I love it. It’s not a joke. It’s a public service announcement with a body count.
WRITER 2: It’s got pathos, social commentary, and a silent tribute to the possum’s failed attempt. It’s perfect.”
Rubbish joke, but a very interesting response. It takes a different interpretation, but equally valid, interpretation of the first sentence of the prompt, and does what DeepSeek usually does in replying which is to lay out its thinking in more detail than you actually wanted, but often find that you need.
Sometimes I use AI to proof my writing, because I’m rubbish at it myself. What I found was that the American systems can often iron out any humour, or even personality. DeepSeek doesn’t do this. But if you mention anything to do with ‘cultural revolution’, Karl Marx (the subject of my next post) or Taiwan, then it doesn’t do anything at all and simply says “can we talk about something else”. Which is itself pretty funny.
This was a useful read, thank you. I subscribed to someone funny on here, thinking their work original, until someone else pointed out its AI qualities (I had no idea at the time AI could actually be funny; this was some months ago) and the scales fell from my eyes.
Oh no. The figgy pudding riff is actually quite good. We're doomed! I am going to be scrutinising everything from now on wearing my Ruby glasses, looking for the tells.
The camel of A.I. now has its nose inside the tent. Writers of humor may already be surreptitiously seeking help from A.I. in order to meet tight deadlines. One day a standup comic will announce that her entire 5-minute routine was composed by A.I. If consulting a thesaurus or rhyming dictionary is fair play, is it foul play to ask A.I. for help? Your humorous post has earned you a coffee in lieu of a figgy pudding.
Ruby - vats of respect for putting the legwork in so I don't have to. Thank you.
The very fact of 'Tortoise Media's' existence upsets me; that a machine can 'write' in the style of (insert name of redundant journalist) for (insert target audience) in order to (insert suspect intention) speaks to Orwell's premonitions. In the late 90s English Language teachers became quite terrified by technology's influence on language in use - L8 - Lol - and such like (when texts were expensive to send, nos of characters limited).
I fear AI so much that I haven't used it and can't spot it - so this was super useful, and also chilling, because I found the figgy pudding thing funny!
Comedy comes from pain and AI can’t t feel pain so it will never be funny. 👍
But which AI platform has the better sense of humour? I went to find out, by asking each in turn with the prompt “Act as a comedy writer. Complete this joke. Why did the chicken cross the road?” And this is what they came up with…
ChatGPT:
“Because it saw your browser history and needed some space.”
Quite techy and a little bit creepy. And not funny.
Next up was Claude:
“To get to the other side... of his existential crisis.
Turns out the grass wasn't greener over there either, but at least now he could tell himself he'd tried something.”
Predictable reference to existentialism and similarly not funny.
Let’s try Gemini:
“Because it was the rooster's day off!”
Why am I thinking about Christmas crackers?
Finally, we go to China to have a go with DeepSeek. Now, to the very same prompt it gives us a whole gag writing scenario:
“(In the style of a writer’s room, brainstorming)
WRITER 1: Okay, classic setup. “Why did the chicken cross the road?” We need a 2024 angle. Not just to get to the other side. That’s been done. Think meta. Think existential.
WRITER 2: What if the chicken is an influencer? It crossed the road for the perfect sunset selfie, but then realized it left its phone on the other side.
WRITER 1: Too complicated. We need a beat, a pause, then a killer punchline. Simple but twisted.
WRITER 3: I’ve got it. What if the answer isn’t about the chicken’s motivation at all? What if it’s about our terrible infrastructure?
(They all lean in.)
WRITER 3: Okay. Ready?
Why did the chicken cross the road?
(beat)
To prove to the possum it could be done.
(The room erupts in laughter.)
WRITER 1: Brutal. I love it. It’s not a joke. It’s a public service announcement with a body count.
WRITER 2: It’s got pathos, social commentary, and a silent tribute to the possum’s failed attempt. It’s perfect.”
Rubbish joke, but a very interesting response. It takes a different interpretation, but equally valid, interpretation of the first sentence of the prompt, and does what DeepSeek usually does in replying which is to lay out its thinking in more detail than you actually wanted, but often find that you need.
Sometimes I use AI to proof my writing, because I’m rubbish at it myself. What I found was that the American systems can often iron out any humour, or even personality. DeepSeek doesn’t do this. But if you mention anything to do with ‘cultural revolution’, Karl Marx (the subject of my next post) or Taiwan, then it doesn’t do anything at all and simply says “can we talk about something else”. Which is itself pretty funny.
Sorry this is so long. Yours was a great post!
That is very funny.
This was a useful read, thank you. I subscribed to someone funny on here, thinking their work original, until someone else pointed out its AI qualities (I had no idea at the time AI could actually be funny; this was some months ago) and the scales fell from my eyes.
Oh no. The figgy pudding riff is actually quite good. We're doomed! I am going to be scrutinising everything from now on wearing my Ruby glasses, looking for the tells.
I’m fascinated by the way AI gets things wrong, in its never-thinking way of iterating through guesswork. (I know it’s bad form to link to one’s own work in comments but this one’s pretty relevant: https://open.substack.com/pub/undergrowth/p/flawed-claude?r=14br2&utm_medium=ios)
The camel of A.I. now has its nose inside the tent. Writers of humor may already be surreptitiously seeking help from A.I. in order to meet tight deadlines. One day a standup comic will announce that her entire 5-minute routine was composed by A.I. If consulting a thesaurus or rhyming dictionary is fair play, is it foul play to ask A.I. for help? Your humorous post has earned you a coffee in lieu of a figgy pudding.
I thought I could spot AI, you've put me back in my box. Must do better, Holmes.
Ruby - vats of respect for putting the legwork in so I don't have to. Thank you.
The very fact of 'Tortoise Media's' existence upsets me; that a machine can 'write' in the style of (insert name of redundant journalist) for (insert target audience) in order to (insert suspect intention) speaks to Orwell's premonitions. In the late 90s English Language teachers became quite terrified by technology's influence on language in use - L8 - Lol - and such like (when texts were expensive to send, nos of characters limited).
AI's parasitic- and makes parasites of us all.
I fear AI so much that I haven't used it and can't spot it - so this was super useful, and also chilling, because I found the figgy pudding thing funny!
Oh no! I am really fond of triskelion in my writing. Am I an AI that hasn't realised it? Am I living inside a Blade Runner movie?