Have you heard the one about the robot who walks into a bar? No? Well, maybe that’s because artificial intelligence hasn’t quite nailed stand-up comedy yet.
While AI can beat us at chess, drive cars, and even compose music, making us laugh seems to be its final frontier. I asked ChatGPT to write us a few jokes to show you examples:
Me: Write me a funny joke.
ChatGPT: Why don’t scientists trust atoms anymore?… Because they make up everything!Me: Write me a funny joke but with dark humour.
ChatGPT: Why don’t graveyards ever get crowded?… Because people are dying to get in!Me: Write me a funny joke but it’s absurd humour.
ChatGPT: Why did the scarecrow become a successful neurosurgeon?… Because he was outstanding in his field but needed a change of scenery!
What do you think? Standing ovation? Edinburgh Fringe award for best newcomer? I’m not sure I’m convinced.
My Edinburgh Fringe musical comedy, Artificially Intelligent, was inspired by the time I asked ChatGPT to write me a show based on my life, and it gave me some hilariously terrible results. I perform some of its materials in the show, and I even make ChatGPT write a song for an audience member live on the spot, and it’s consistently hilarious in its terribleness.
It generated songs telling an audience member who is an optometrist to poke out the eyes of her co-worker who she is always arguing with, and even told an audience member who is a teacher to fail the students that he didn’t like so he can have a more chill time at work. However, this wasn’t independent thinking; we had to feed ChatGPT as much information as possible to get any results from it. When it comes to actually getting AI to craft jokes on its own, it can’t.
The problem with AI and comedy is it lacks any of the human nuance that makes something funny, like your perspective on life, shaped by your personal experiences. AI might be able to create a joke structure, but it does not have the ability to pull from personal experience. It lacks a point of view, situational awareness, context, and cultural awareness, which are all necessary to create good comedy.
Irony is the perfect example of a sophisticated form of human communication that AI struggles with due to its reliance on context, tone, and cultural knowledge. Unlike humans, AI lacks the ability to infer meaning beyond explicit statements. Irony involves saying one thing but meaning another, relying on shared context and nuanced delivery to land the joke, but AI only processes text and can’t pick up the audio and visual cues that signal irony to humans like the hint of sarcasm in a voice or a deadpan delivery. AI has to signal that it is, in fact, making a joke.
AI has made impressive strides in areas like natural language processing and machine learning. These technologies allow AI to understand and generate text, which includes jokes. AI models, such as OpenAI’s GPT-3 and its successors, have demonstrated the ability to generate puns, one-liners, and even short comedic skits. However, these jokes often rely on simple wordplay or common joke structures, and they aren’t original.
Despite all of these limitations, will AI ever be funny? As of now, absolutely not, but in the future theoretically yes. When technology advances to the point where robots are walking around with free will and independent thinking, experiencing life and potentially even emotions, they would theoretically begin to have enough perspective to create comedy, or an activity that voices their experience of being an AI. However, this would create its own culture and would only be relatable to other AIs.
The truth is that audiences are really enjoying the AI-generated material in my show, even if the joke is that it’s hilariously bad. But – technically – does that make it funny?
Comments