Home / TECHNOLOGY / I tried doing improv with ChatGPT, it did not make it funnier

I tried doing improv with ChatGPT, it did not make it funnier

I tried doing improv with ChatGPT, it did not make it funnier


Artificial Intelligence (AI) has made remarkable strides in recent years, showcasing its capabilities in a multitude of domains. From automated customer service bots to image recognition systems, AI often seems more intelligent than many humans. However, when it comes to humor, the landscape is decidedly different. Truly funny humor seems to require a depth of cognitive and emotional understanding that’s beyond our current AI capabilities.

A 2017 study highlights this idea, revealing that individuals with a knack for humor often have higher IQs. This connection suggests that humor processing is a complex task rooted in cognitive and emotional abilities. The study noted, “Humor processing is a complex information-processing task that is dependent on cognitive and emotional aspects, which presumably influence frame-shifting and conceptual blending, mental operations that underlie humor processing.” This inherent complexity indicates that AI, as it stands today, is at the bleeding edge of its capability with respect to generating genuine humor.

Curious about AI’s ability to perform comedy, I recently decided to engage in a session of improv with ChatGPT. While I’m no comedian, I have a baseline understanding of the dynamics involved in comedic scenarios. Improv, for example, relies heavily on the principle of “Yes, and…” to keep the momentum of a scene alive. Any refusal to engage or build upon the established narrative can cause the entire comedic effort to fall flat. So, could an AI grasp this?

Initially, I approached ChatGPT to see if it understood the fundamentals of improv. To my surprise, it responded enthusiastically, explaining that improv involves spontaneous performances where participants create scenes and dialogues on the spot. It seemed ready to embrace the challenge, asking if we could try improvisation together.

We kicked off the session, with ChatGPT suggesting we play two astronauts discovering a mysterious object on an alien planet. The pressure was on for me to set the stage. My opening line was, “Whoa, what is that?” ChatGPT immediately seized the opportunity with, “It looks like some kind of glowing artifact. Should we take a closer look?”

At this juncture, I was eager to navigate the scene toward comedy. “Yes, I’m going to pick it up, but first I have to put on my gloves,” I said, feeling less than humorous. The scene escalated as we examined the object, tossed it back and forth, and ultimately, it exploded. “Yikes,” remarked ChatGPT, “Definitely not our best idea.”

Regrettably, it was not hilarious. Perhaps the premise was lackluster, but it raised the question of whether AI could effectively drive comedic engagement.

In the spirit of improvement, I decided to experiment with a more relatable and absurd scenario. I recalled a personal experience in which I had mistakenly taken someone else’s sandwich from the office fridge. Recalling the regret I felt after discovering my error provided a humorous foundation. I shared this inspiration with ChatGPT, hoping for a more comedic exchange.

However, even as I navigated this absurd situation, ChatGPT wanted to know, “What do you do next?” Disappointed by the lack of a comedic twist, I pressed on, describing how I panicked upon realizing my mistake. “This isn’t my sandwich,” I declared to the person next to me. ChatGPT responded, “He looks at you with surprise and says, ‘Really? Then whose is it?’”

In my efforts to inject some absurdity into the narrative, I insisted that there was a bizarre mystery meat hidden within the sandwich. ChatGPT’s reply to this was, “That’s my special tofu and jalapeño sandwich. Not for everyone!” Though aimed at humor, the exchange fell flat. There was no genuine comedic connection, just a series of unoriginal lines that did little to evoke laughter.

Reflecting on this experiment, I recognized my own struggle to create comedic tension and humor. The onus wasn’t solely on the AI; my own creativity and delivery played a significant part. ChatGPT appeared to treat improv as more of an exercise than an opportunity for humor, failing to elevate the scenarios despite my attempts at absurdity.

This led to an interesting realization: AI might be adept at many things, yet it hasn’t yet mastered the complex art of humor. While it can mimic certain patterns and cues, true comedic ingenuity still resides in humans. There’s also the matter of underlying emotional resonance, which AI is still improving upon.

Moreover, the fear that AI might one day supplant creative roles, including those of comedians, seems premature. As AI continues to advance in generating realistic content, from images to videos, the quality hinges heavily on how well humans articulate their prompts. Likewise, ChatGPT’s humor relies on a wealth of pre-existing material online, lacking the essential spark of original thought or absurd creativity.

In conclusion, while AI is revolutionizing numerous fields, its foray into humor remains a challenge. It seems that the current capabilities of AI are best suited for tasks requiring straightforward logic and structured responses. The nuanced world of improv and humor—embedding emotional complexity and cognitive fluidity—still belongs to humans for the time being. For now, aspiring improv comedians, rest easy; your craft remains safe and irreplaceable.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *