A few months ago, a software engineer asked me what I was doing. I was writing. “Why? You know,” he said, “that AI can just do that for you.” His comment did not surprise me. He was the prototypical image of the software engineer type: pragmatic, exacting, skeptical. 

“Have you ever read anything AI has generated?” I asked. 

The piece you are currently reading was entirely created by a human. I’m not a betting woman, but I’m willing to bet that you can tell, from these first few sentences, that what you’re reading has not been AI-generated. Humans speak and write with accents. So does AI. You know when you’re reading a newsletter article by ChatGPT. The text is formulaic. The words logically fit together, but they don’t say much. No bold claims or assertions are made. Generative AI’s work is precisely on par with its design. We are currently getting the exact kinds of outputs we should expect from models that function as predictive text generators. 

The goal of natural language processing (NLP) is to enable computers to understand human languages, and to make these machines accessible to those of us who aren't expert programmers and thus are not proficient in machine-specific language. NLP models use digital neural networks, which copy the human brain’s biological network by linking nodes to construct communication networks, to recognize patterns, classify information, and make predictions. The neural networks that support NLP models like ChatGPT are trained on gigantic textual datasets and are capable of identifying relationships between words and phrases. 

Artificial intelligence systems like ChatGPT are extremely capable writers. If ChatGPT and I were given the same prompt, the large language model would produce an objectively, technically better output than I could. I’ve never had any formal writing training. ChatGPT approaches textual tasks systematically, so its writing would have all of the organization, structure, and proper form which my writing lacks. Also, ChatGPT would be finished writing in seconds, while I would take hours. 

I have one thing which artificial intelligence, at least as it currently exists, does not have: the personality which imbues my writing by the simple fact of my humanity. I have a unique voice. There is something distinctly human about the tone of my work which AI cannot emulate (yet). 

The craft of writing is itself a human impulse. Machines write because we have programmed them to do so. 

Me-generated writing may be rambling and full of technical errors, but my words don’t carry that shadow of the uncanny that plagues AI-generated content. I’ve used ChatGPT, Claude, and their contemporaries to create all sorts of professional copy and plan to continue to do so. I’m immensely grateful that these tools allow me to automate rote, technical, uninspiring sorts of writing tasks. 

Automation removes obstacles. Without obstacles, we have nothing to overcome. I choose to write the things that are important to me, and the things that are difficult to say, by hand and by heart. Writing is awful. It’s hard to stay focused and harder still to try to translate my emotions into words and then hand slivers of my psyche over to the public for judgment. Still, I keep doing it.I write because it’s compulsory for me. To be human is to create. To be human is to toil. I have to write to live. 

Generative AI shows promise. We’ve collectively identified many successful AI use cases. There’s a use case that’s been overlooked. Perhaps the most useful aspect of AI is that it can serve as a digital Rorschach test that reveals to each of us the most important elements of our humanity. Find that thing which you don’t wish to automate, whether that be writing or something wholly different. Pursue that thing even when, and especially when, doing so is difficult. 

Many are of the opinion that AI poses existential risks. I agree, but it’s not the Terminator-style vision of the future that frightens me. I’m not afraid of murderous AI. I’m afraid that humans will decide to ignore the complicated aspects of our humanity in favor of ease and automation. I’m afraid that we’ll stop writing.

Elizabeth Tate

Elizabeth Tate is a writer, speaker, and consultant living in Washington, D.C. Much of her work explores the influences of modern technology on the human psyche and the lived experiences of Generation Z. 

Previous
Previous

Pricing LLMS: The True Cost of Compute

Next
Next

Spatial Intelligence is Helping AI See, Learn, and Do Better