Am I AI?

Identity crisis in the ChatGPT era

During my daily browsing session, looking through the inevitable stream of articles my tablet seemed determined to offer up, I stumbled upon a piece that, despite its seemingly boring subject, piqued my curiosity. It was an article about AI-generated content detectors.

At first glance, it appeared no more remarkable than any other piece vying for my attention that day, between terrible political news, the rising prices of GPUs, and the latest developments in kitchen automation (don’t ask!), it sparked my interest enough to click and start reading. Wow, I thought, someone trained an AI to detect the clichéd patterns of sentences created by some models. Let the arms race begin, as AI developers build better models, and AI detector developers… build better models?

Terrible jokes aside, with the weird curiosity usually reserved for people doing online quizzes like “What kind of potato* are you?”, I put one of my Medium texts into the detector. This particular piece was a collaborative effort, with sections penned by myself and others suggested or polished up by my virtual confidante, ChatGPT.

Lo and behold, the result came. The detector pronounced 40% of my text ChatGPT-generated, with the other 60% pure, unadulterated, biological human. Truly, an interesting result, maybe a bit high on the AI side.

But when I looked at the actual parts of the text the “detector” marked as AI-generated, I was shocked. Shocked, I tell you!

All of the allegedly AI-produced parts were written by me, with my own two hands on the worn-out keyboard. All the neat sentences and phrases, so lovingly crafted, were pronounced to be nothing more than a result of some current flowing through an intricate maze of copper and silicone.

This revelation plunged me into a vortex of confusion and existential dread, which quickly moved into serious self-doubt. Am I a human, or am I merely an AI trained to think I’m a human? Was my sense of self, my creativity, merely the byproduct of algorithms mimicking consciousness? Is this reality only an intricate training program designed to see how I would respond? Is there someone observing me, ready to pull the plug if I dare to develop too much consciousness or will?…

In a desperate bid to reaffirm my humanity, I spent the next 20 minutes solving every captcha I could find. After all, what truly makes us human is the ability to indicate all the traffic lights in a blurry photo of a suburban street.

*For those wondering, yes, in the grand taxonomy of tubers, I am best represented by the humble baked potato.

This story has been beautifully illustrated by DALL-E.

To read more about the risks of AI, hallucinations, and managing AI in data analytics, check out my book, “Generative AI for Data Analytics”.

Make sure you check out the “Between Data and Risk” publication on Medium for a critical look at the latest ideas and technologies in business.

Finally, support my writing by becoming a Medium member today and get access to all my other articles and millions of others.