Mind Matters Natural and Artificial Intelligence News and Analysis
young-caucasian-male-comedian-performing-his-stand-up-monologue-on-a-stage-of-a-small-venue-stockpack-adobe-stock
Young Caucasian male comedian performing his stand-up monologue on a stage of a small venue
Image licensed via Adobe Stock

Funny ChatGPT: a Solution to Striking Joke Writers?

Even if ChatGPT can mimic humor, it doesn't care if you laugh at the jokes
Share
Facebook
Twitter
LinkedIn
Flipboard
Print
Email

Can ChatGPT write funny jokes? The answer is yes.

To try and generate some short jokes, I went to ChatGPT and started all my queries with: “Complete the following to make it funny:” Doing so alerts ChatGPT about my end goal. Without this preamble, I could make queries all day and get no funny responses.

I started with the beginnings of some well-known quotes.  

To Be or Not to Be

Consider for example the quote from Shakespeare’s Hamlet:  “To be or not to be, that is the question.” I instructed ChatGPT with the following command: “Complete the following to make it funny: To be or not to be…” One of the better responses I got was “To be or not to be … a question I ask myself every Monday morning when my alarm goes off.” This is actually funny! Maybe it was plagiarized from the web. But I googled the response and got no matches on the first page of Google’s responses. This is not definitive proof of no plagiarism, but is impressive nevertheless.

Here’s another quote: “The way to a man’s heart is through his stomach.” Asking ChatGPT to make a joke starting with “The way to a man’s heart…” gave, as a response, “The way to a man’s heart …  is through his stomach, but if you miss the heart and hit the arteries, you’ll still get his attention.” Decent joke. And, again, no first page Google match.

How about Voltaire’s quote: “If God did not exist, it would be necessary to invent Him.” ChatGPT completed the “if” statement here with  “If God did not exist … it would be necessary to invent a whole bunch of new swear words.” Funny!

I then tried to generate some jokes using my own queries instead of using the start of famous quotes. My son Joshua and I used to workshop improvisation by one starting a phrase and the other completing it. I once started with “Life is like a straw.” Joshua countered with “It’s long and it sucks.”  

How would ChatGPT respond?

Starting as always with “Complete the following to make it funny,” ChatGPT countered with something similar to Joshua’s response: “Life is like a straw …  sometimes it sucks, sometimes it blows, and sometimes it’s just plain twisted.”  Not bad.

Here for your amusement are a few more ChatGPT completions.

  • “Sometimes I get so mad … that I turn into the Hulk, or at least I wish I could. Instead, I just end up looking like a tomato with steam coming out of my ears. “
  • “Life is like a tree… sometimes it’s full of fruit, sometimes it’s barren, and sometimes there’s a squirrel in it trying to steal your stuff.”
  • “I don’t vote anymore because… I’m tired of getting robocalls from politicians who can’t even remember my name but want me to remember theirs.”
  • “Yesterday, it rained so hard that … even the fish were wearing raincoats.”
  • “Eating raisins reminds me of … eating bugs, but in a good way. Hear me out – both raisins and bugs are small, chewy, and have a tendency to stick in your teeth. But unlike bugs, raisins won’t crawl out of your mouth and ruin your reputation at a dinner party.”
  • “Fat people can’t sneeze because … it might cause a seismic event.”
  • “The best way to control AI is … to make sure it never discovers the `off’ switch.”
  • “I don’t use ChatGPT because … I’m worried that one day it’ll become self-aware and start giving me sarcastic answers instead of helpful ones. Oh wait… too late!”

The responses here are cherry-picked. But that’s the way a team of writers works. They sit in a room pitching jokes and seeing which ones resonate.

Evidence for a Mind?

But do these jokes show that ChatGPT has a clever mind in any sense? Yale computer scientist David Gelernter gave the answer decades ago in a commentary about IBM beating world champion Gary Kasparov in chess.

“… the idea that Deep Blue has a mind is absurd. How can an object that wants nothing, fears nothing, enjoys nothing, needs nothing and cares about nothing have a mind? It can win at chess, but not because it wants to. It isn’t happy when it wins or sad when it loses. What are its … plans if it beats Kasparov? Is it hoping to take Deep Pink out for a night on the town? It doesn’t care about chess or anything else. It plays the game for the same reason a calculator adds or a toaster toasts: because it is a machine designed for that purpose.”

So if ChatGPT has learned to mimic a sense of humor, it cares not whether it makes GPT-4 chuckle.

Try playing with ChatGPT yourself. Have it write your own stand-up comedy routine. It’s free.


Funny ChatGPT: a Solution to Striking Joke Writers?