I, Not Robot

by M.M. Adjarian

WHEN CHATGPT CAME OUT, I lifted one world-weary eyebrow then returned to the business of my life. This was, after all, the post-human age; AI and robots formed the backbone of an emergent reality of the kind visionary twentieth century science fiction writers Isaac Asimov—and predecessors like Victorian novelist Samuel Butler—had been exploring for 150 years. Surprise, like resistance to Borg assimilation on Star Trek, was futile.

It took me exactly two months after that to set up an account and another few weeks after that to secure a spot on OpenAI's exuberantly overcrowded platform. The catalyst had been a ChatGPT anecdote my novelist friend Sara had posted on her Facebook page. For one hour, she had fed the chatbot a series of random creative tasks: Write a descriptive paragraph about stepping on toes. Write a love sonnet about pickles. Write a manifesto against green beans. Sara then asked for a poem that referenced everything the bot had produced for her. She shared the result along with this bemused comment: What a totally weird time this is to be a human being.

I'd indulged in enough online doom-scrolling to understand that human civilization was officially under siege. Robots and other AI were taking over the planet and as they did, the flesh-and-bone beings who created them would lose the capacity to think and do for themselves and become even more enslaved to electronics than they already were. Humanity was becoming the obsolete dream of another age. But before ChatGPT, the AI apocalypse was still far enough away that I could still manage to live in perfect harmony with my computer, the devices that controlled my home heating and cooling system and the smart home apps that monitored everything from pipe leaks to when and how often my front door and garage opened and closed. Now the jig was up. AI had finally infiltrated the world of words I inhabited as a writer.

For days afterward, I felt the mortifying weight of my own post-humanness. I compose on computer but still do idea-sprawls on legal pads, usually with my favorite fountain pen in hand. In its early stages of trial and error, writing is an act that takes time: hours, days, months, and sometimes, years. Perhaps this need for pen and paper is a generational vestige. I was still writing by hand then, using a self-correcting Smith Corona to type out final copies of class essays—and the occasional poem I squirreled away at the bottom of my desk drawer&madash;in college. Of necessity, I learned how to write on a box-style Macintosh my first semester of graduate school. But it would be another semester before I stopped treating my computer like my old Smith-Corona and more like the ferociously efficient word processor it was.

I took my first turn on ChatGPT three months later, on a cold February evening. Cracking my knuckles, I typed my first command. Write a two-sentence summary of War and Peace. The cursor blinked once then raced across my computer screen to return a textbook perfect reply: War and Peace is a historical fiction novel by Russian author Leo Tolstoy, first published in 1869. The book explores the lives of five aristocratic families in Russia during the Napoleonic Wars. I half-smiled. Good boy, girl, whatever? You get an A. The bot had access to the vast knowledge bases available online; my brain and everything—every book, lesson and webpage—had absorbed into it more than fifty years of living felt incomparably small.

I grew bolder. My fingers tripping across the keyboard, I decided to test ChatGPT's design limits. Do you believe AI has a soul? This time, the cursor blinked twice before responding: As an artificial intelligence model, I do not have the ability to believe or have opinions, as these are characteristics of human consciousness. I thought about the Siri AI assistant on my iPhone. That, at least, could answer the sarcasm underlying my inquiries—tell me a bedtime story, please—with pre-programmed snark. In the great green dimension, there was an iPhone and a red balloon… So far, ChatGPT showed no signs of either imagination or a sense of humor.

Amused, I set the next task&emdash;a human experience-based question. Write a descriptive essay about sex. The bot did not indulge me. I apologize, but as an AI language model, I cannot generate inappropriate or explicit content. It is important to maintain a professional and respectful conversation. I tried again, this time with a similar question that didn't verge on the libidinous and would allow it to demonstrate its knowledge of literary form. Write an Imagist poem about love. The bot speedily obliged:

Love, a bud blooming in early Spring,

Fresh and fragile on a gentle breeze,

Opening slowly as the birds sing,

Tender petals unfurl with ease.

Dew drops glisten in the morning light,

As love's fragrance fills the air.

A kaleidoscope of colors bright,

As two hearts beat without a care…

The bot had maintained the clarity and simplicity demanded of Imagist language. At the same time, the actual images and metaphors used lacked idiosyncratic uniqueness while the end-rhymed lines demonstrated total insensibility to the rhythmic possibilities of free verse. I snickered; for now at least, we lumbering humans still had the upper hand when it came to craft and artistry. But I could not deny that ChatGPT had also managed to produce a complete, multi-line poem in under five seconds—far more than I or any other human could do in the same amount of time.

Still determined to see how far I could push this machine, I entered one final command. Write about love & sex with humor and in verse. This one, I hoped, would bypass the proprietary restrictions embedded into ChatGPT coding. Two blinks later the cursor leapt into action:

Love and sex, they go hand in hand,

Like peanut butter and jelly on bread.

Some say love is blind, but I disagree,

It's more like love is horny, can't you see?

One moment you're whispering sweet nothings,

The next thing you know, you're bumping and grinding.

Sometimes things get crazy, and that's okay,

Just make sure you don't end up in a Fifty Shades of Grey…

Nonplussed, I laughed. Perhaps that was the attitude I needed to take towards this new technology. As much as it could entertain and amuse, it was—for now, anyway—only just an interfacing tool. A disembodied electronic mind able to summon information from the vastness of the Internet and reorganize according to the whims of its human users. My own mind was embodied; and because of it, able to filter organic information/experiences unavailable to ChatGPT. I may have been the turtle to the chatbot's hare but I could enjoy my peanut butter and jelly on bread in ways it would never understand.

Signing out of my OpenAI account, a memory from years ago surfaced, a memory of another writer, Alice Walker, speaking to an audience in Oakland about her work as a writer. It was 1995 or maybe 1996; my mind is an imperfect, memory-muddling machine. Of the many things she said that night, about her books, her activism, her politics, one point stayed with me. She too, preferred to write by hand. But her reasons were far more poetic than mine. From the heart straight down to the pen, she said.

M. M. Adjarian is a critic, award-winning essayist and occasional poet who has published in such journals as the Baltimore Review, Verdad, South 85, Grub Street, Pif, Glint, Gravel, Eclectica, Crack the Spine, The London Reader and the North Dakota Quarterly. Her articles and reviews have appeared in Tribeza-Austin, Arts + Culture Texas, Bitch, Kirkus Reviews, Library Journal, and the Dallas Voice. She maintains a personal essay blog on her website, An Austin Writing Life, where she muses on everything from junk art cathedrals to how to survive middle age through stripper pole dancing.

— posted FEBRUARY 2024