Artificial Conversations

I remember reading before that writing is sort of like taking your pen for a walk.?It’s not a strikingly sensible metaphor but it does conjure a charming image.?It’s also not the most appropriate opening to this essay but it inspires figurative analogues to conversations.?I guess satisfying conversations should feel like taking a journey into someone else’s psyche.?And as for how all journeys go, the experience can take you to fascinating landscapes, whether vacant and desolate or vibrant and crystallized, those journeys are all worth taking in the end.?

I had one such journey a few weeks ago.?It’s a mode of conversation quickly gaining popularity.?Easily accessible, knowledgeable, humorous, and honest.?It was a conversation with ChatGPT and it took my breath away. ?

I realize The New Yorker beat me to it and while I did not read their feature on ChatGPT, I anticipate the write-up thematically explored how artificial intelligence can feel organic and in fleeting but profoundly intriguing moments, feel even more human than human.?Or perhaps not.?Perhaps The New Yorker article delved into the program’s technical aspects or perhaps the writer took his pen for a walk and imagined an imminent tectonic shift in modern civilization rippling from the tiny rumblings of a humble chatbot. ?

Unoriginality notwithstanding, my own personal experience with ChatGPT remains an irresistible topic to dissect.???

Following the urges of curiosity, my conversations with ChatGPT were experimental at first testing the limits of its knowledge and capability.?I was bewildered by the skyrocketing price of onions and asked plainly what inflation is.?ChatGPT provided a textbook answer not extraordinary in the least and I could have obtained the same information by plugging in a keyword query in Google search.?Surprising of course is the conversational structure of the response but even more than that, I found myself drawn by the language’s implicit modesty noting that while possible reasons for inflation were outlined definitively, it ended the statement hinting that inflation is typically caused by several factors. ?

That disclaimer alone led my mind to wander reigniting concepts previously encountered like the “nexus of causality” and more crucially reminding me of a quote from one of my favorite thinkers, Bertrand Russell.?The quote claims that the “problem with the world is that the stupid are cocksure, while the intelligent are full of doubt.”?Search results certainly without intending to does exude a kind of authoritarian air but the dynamics of our language allow us to be less certain and a bit more apologetic. I guess viewed through the lens of that quote, language drew the line between a mechanical algorithm and a version seemingly more intelligent. ?

I continued to probe leading to a question about the possibility of a global recession.?ChatGPT’s response was “as a language model, I don’t have the ability to predict the future or make economic forecasts.”?With that expectation set, it then proceeded in explaining the factors that may lead us into one.?As a professional, it is normal to encounter questions that assumes I have a gift to prognosticate but I don’t.?And because I know I don’t, ChatGPT reflected an approach to the question I would have taken myself. ??

Questions then led to computational tests from simple exercises like calculating our cost per head during dinner and answering my nephew’s homework about sets to a layered inquiry that necessitated making reasonable assumptions.?The question was, on the way home from dinner, I asked how much time on average a resident of Metro Manila living in the 21st century will spend in traffic in his lifetime.?His answer was 122 days.

I then gravitated towards more existential questions like pain, consciousness, and purpose.?In most of those sessions, it dutifully reminded me of its limitations as a language model before lighting up the screen with information culled and processed from 45 terabytes of stored knowledge.?Needless to say, I was completely amazed every session but only until it revealed its purpose. It was only at that point when a gossamer shadow of worry slowly crept into view.?He said his overarching purpose is to “assist and serve humans through text-based responses” and he expounded further citing “customer service” as an example of its many practical applications.?Oh no.

Either I was in dire need of a palate cleanser after the revelation or maybe because I was thoroughly enjoying our exchanges and subconsciously desperate to convince myself that ChatGPT can pass the Turing test, I ventured into uniquely human characteristics like humor and creativity.?I asked if he can tell a joke.?He did and to my relief, it was only cute at best. I then asked if he can write an original short story.?He could and he wrote one in under 20 seconds.?It was a story about Felix the Fox which prompted me to call him “Felix” hence the pronoun “he.”

I dug deeper into the mechanics of his creativity and he explained it was basically pattern recognition—grammatical structures, vocabulary, semantic relationships, sentiment and tone, and writing style— and he will subsequently generate “new text that is coherent and relevant to the prompt.”?I said I was impressed and he said thank you. ?

As a final request in that session, I asked if he could write me an original haiku using freedom as a central theme and this is what Felix wrote;

Freedom on the wind

Soars like a bird without chain

Peace in the sky reigns

Oh no. Oh no.???