Experiments with ChatGPT - Is Cre(AI)tivity logical?

Hello dear readers!

Long time no see - clearly.

Well, as always, I'm here to tell you about my recent escapades with anything and everything that fascinated and piqued me. I have a whole lot of things to share, but let me start with the elephant in the room - CHAT GPT

Chat GPT is apparently the saviour of students when it comes to completing their assignments, providing ideas and direction for their projects and basically is that good samaritan hated by teachers and loved by all.

I was fascinated by its capabilities and had reviewed articles that sung its praises to the heavens. I decided to challenge GPT by asking from it the one thing I was confident it could not provide - Creativity.

To start off, I wanted to understand how it worked, so I gave a very simple prompt.

excuse my grammar - but hey it was smart enough to understand me

The A.I model came up with 4 of the most popular short story prompts - and it made sense to me. These were the tried and tested models that conveniently gave a highly positive character arc for the protagonist while making their surroundings and life situations the antagonist. 

I then proceeded with this prompt:

"Write a story with one of the above prompts in 155 words"

The story was.... disappointing, even for one who went with zero expectations. Perhaps for avid readers atleast. On the contrary, it could have easily passed off as a story found in a 11 year old's bed time stories (if they still follow that lovely habit).

As ChatGPT is nothing but a trainable A.I model, I prompted it to tweak the story.

getting better definitely

I liked how GPT decided to add a quirk in Carla's changed character, indicating that she had taken away something apart from positive growth, from her hospital experience. Again, GPT did not think of giving conversational examples of "dark humour". It's interpretation of my prompt was to take up the most optimum route for "including elements of dark humour".

Moving on!

why does Carla have to end up a "positively" changed person so quickly??

The story's conclusion was neither climactic nor surprising. It was quite conventional. By now, I was curious to understand GPT's mindset. Hence, I asked it a very normal question anyone would think of asking an A.I bot (:P)

Carla is the poster child for positivity

Here, I threw a curveball at the bot, confident it could shift the story in an entirely different direction. However, I was given a surprising response.

who created these principles?!

The bot was following the basic principles of simple story writing. It did not occur to it to give Carla layers in her character or take the readers on a rollercoaster journey via a gruelling character arc or heck to even provide context as to Carla's background, her financial situation or her family. It gave a very feel-good, comforting story.

It's ok GPT - we all make mistakes.

And now for the icing on the cake, the best question I came up with!

Thalaiva!

I asked subsequent questions with various other personalities and understood its capabilities better. With consistent data pertaining to a topic or personality, it could come up with believable models. The insane speed with which it could come up with a story that respected fundamental rules of storytelling (who wrote those rules?!- likely inferred from the past data it has access to) while also restricting itself to the constraints is commendable.

I am happy to note that it does not have a shred of uniqueness or creativity to it - in the sense of coming up with something that is not already here (atleast for now). A.I can turn out to be a wonderful sounding board for your ideas, a great tool for forecasting future possibilities provided you give it clear directions, but it can never, ever replace a writer.

Guess one of my dream jobs is safe!

For now.