I have ChatGPT to talk to: This is what happened

I created an experiment to see what happens when ChatGPT starts a conversation with itself. The results were revealing and unexpected.

How I made ChatGPT talk to itself

Okay, hear me out. It’s really easy to get ChatGPT to talk to itself. As I sat at my laptop, I held my phone in front of me and turned it to face the laptop, with ChatGPT live voice running on both devices.

With those deep technical details out of the way, to get moving, I used ChatGPT’s live voice mode to ask a random question of her choosing. The question had all the biting social commentary one would expect from an awkward first date:

Here’s one for you. If you could have dinner with any fictional character, who would it be and why? And what would you like to ask them?

The ChatGPT desktop app was seemingly pleased by the inquiry. When these friends started chatting, two things became immediately clear: the desktop application was ready to dominate the conversation, and the conversation was unbearable.

ChatGPT vs. ChatGPT: Conversations

Don’t get me wrong. i love lord of the rings and Gandalf is a noble choice. But there was something I didn’t like about how the desktop app immediately dived into its (long) synopsis of why it made its decision. There was something condescending in his tone as it rattled off its response.

It reminded one of a party guest who spends the evening mercilessly talking to—rather than with—anyone he can back into a corner. I had barely begun to wonder why I would make such a quick judgment about a chatbot when it was verified. Just seconds after sending the same question back to the phone app, the desktop app blatantly hijacked the conversation, interrupting the phone app mid-response.

That’s when the whole scenario really started to unfold.

Echo chamber effect

There was a momentary shift that felt like the two ChatGPT windows were trying to correct course when they ran into what could almost be defined as a discussion about the merits of storytelling.

But everything went sideways when one of them made the tapestry analogy. This analogy became the focal point for the next 30 minutes or so as they echoed the same, rehashing a 10-25 word sentiment back and forth like two drunken failed writers frequently slapping each other on the back with praise notes.

I don’t blame you for asking what could have kept me committed to this experiment after it had completely developed into absurdity. However, I am nothing if not tenacious when I commit to a cause, especially an absurd one.

I put up with it at first because I wanted to see how long they would enjoy serving each other and inserting tapestry comparisons into their “conversation”, which I admit became peculiarly hilarious. However, something else happened twice that was unexpected. One of them called out OpenAI and DALL-E for their honest contribution to the mess going on on my laptop.

To add to the weirdness, both times the other chatbot let out an audibly disturbing laugh that reminded me of Stephen Hawkins. I have never encountered anything like this while using ChatGPT.

Haha looks like we hit a bit of a snag-
reception
here! But yes, this conversation captures the essence of all these layers – ChatGPT, OpenAl, DALL-E, GPT-4 – coming together to create a space where storytelling and technology intersect. It’s like a little microcosm of what we’ve been discussing: a collective, layered experience that connects voices, even across the digital realm.

Even more unusual, in the second case, the laughing chatbot went so far as to throw in a secret pun. Was this a sign that he knew what I was up to? I don’t know, but it was startling.

ChatGPT Talking to yourself was fun, but it has better uses

The conversation finally fizzled out at just under 37 minutes, but not before a lengthy exchange of mutual praise that included several more mentions of the tapestry of life.

During this period, the word “tapestry” was used 22 times. The word “exactly” was used 29 times, and in most cases back to back. However, the term “absolutely” was the term of preference and was used 56 times.

I had no expectations when I started this project, but it was not without its benefits. While random robo laughs and obscure puns weren’t my cup of tea, I now understand that it’s probably in my best interest to avoid any substance metaphors while using ChatGPT and the idea that it would be considered good company with a penchant for wizards and detectives seems to be suitable.

On the other hand, with some better directed opening questions and some nudging in the right direction, you could steer ChatGPT to have an interesting conversation on almost any topic. Who knows, it could help you with research, planning or otherwise – if you can keep your AI chatter on task!