Suspicion hung on glances cast across the table. Sweat beaded on our brows. The heater blew hot air into the already sweltering room. It was a random hot November day. The quick switch in temperature left no time for the climate control system to adjust, the machinery just kept pumping out heat like it was 40º and not the muggy 82º reality. I sat in a Creative Nonfiction workshop in the early 2020s and the use of AI on college campuses felt taboo—especially in creative writing programs. Every em dash viewed with skepticism, every colon proof of an LLM shortcut. Students prompting their way to an essay instead of doing the real work. It felt like the Salem witch trials or McCarthy-era Washington D.C.: I saw Goody Rivera with ChatGPT.
Okay, a little over the top, but the news at the time didn’t help to calm anyone’s nerves. Every day a report on the evening news, a new think piece in The Atlantic, an existential broadcast on NPR all talking about the disruption and decline of higher education at the digital hands of artificial intelligence. Professors and students, parents and teachers, all scrambling to make sense of what was appropriate, what was legal, and what was the point of it all anyway? The fear—swift; the response—slow. People guessing at how these tools should or shouldn’t be used. People guessing at their continued relevance.
Since then the world has adapted. While there are still many unanswered questions, Generative AI has become ubiquitous, and most people I know use it. New language has been assimilated into the lexicon. Slop, hallucination, deepfake have all taken on expanded meaning and prominence. Corporate America has always been adept at co-opting tech terms: bandwidth and download are joined by prompt whisperer and human-in-the-loop. Technology has a way of changing how we talk, for better or worse. New words, dialects, and cadences get incorporated into everyday speech. It has happened for centuries.
There are many words in use today that are byproducts of technology adoption or evolved from technological origins. “Hello” wasn’t in prominent use until Thomas Edison marketed the word for the telephone. Had Alexander Graham Bell won out, instead we’d be saying “Ahoy!” The telephone also introduced “hang up” which is still in use despite phones no longer having hookswitches. Radio brought along “on the same wavelength” and “tune out.” I remember sitting in my teenage bedroom late at night. The room dark except for the eerie blue light of my monitor as I typed brb and lol into America Online Instant Messenger. It wasn’t long until we were actually speaking those acronyms along with tbh and idk. Social media ushered in an onslaught of new terminology: “going viral,” “getting ghosted,” “blocked.”
We started to organize our thoughts around hashtags. I can be absolutely ridiculous in this regard. I often let the mask of professionalism slip and reveal a bit more of my internet influenced self. I’ve been known to be on calls or in meetings and randomly throw up a four-fingered hashtag while verbally saying hashtag. I started doing it as a joke after a Justin Timberlake sketch on Saturday Night Live pointed out the extreme cringe of the action, but it eventually engrained itself into my natural flow of speech. A random “hashtag could’ve been an email (#couldabeenanemail), hashtag Lorde Rules (#lorderulz), hashtag Justice for Solar Power (#jfsp).”
Aside from generational grumbling and despondent grammar purists wringing their hands over the language drift, these incorporations of tech speak into common vernacular are generally harmless and can enrich how people communicate. While AI is following this pattern I’m seeing something else emerge along with it. In previous cycles, technology has shaped how we talk by introducing words and conventions—when typing on a T9 multi-tap keyboard, it is economical to shorten everything to acronyms and to find other shortcuts. But that feels different from the impact AI is having on content today. We’re tying ourselves up in knots to make sure we come across as authentically human. Scrubbing our prose of any whiff of AI generation so that our writing is accepted by human readers. In tandem, we’re acutely aware of anything that might offend an algorithm’s sensibilities and limit our reach, so we find workarounds to connect to our audience. These forces have the potential to erode and censor—limiting the ways we actually connect and communicate.
There is skepticism around AI generated content and its actual value. The response is to subtract the elements that are considered AI tells. The em dash and the colon are just two small examples, but they are prominent ones. As generative AI rose, so did the suspicion of its use. If I were in the em dash business, I would’ve gone bankrupt after the first wave of articles rolled out enumerating the telltale signs that something was written by AI. The villains: em dashes and colons, repetitive paragraph structures, formal transitions like moreover and furthermore, and hyphen heavy compound descriptors. In many ways it reads like the guidance many of us were given in high school on how to write strong five paragraph essays, then are later taught to reject to have more lively and original writing. Writing with a voice. Why, because AI was trained on all of the tools and tricks we’ve used for generations. AI relies on em dashes as spackle to smooth over disjointed thoughts. Stitching together clauses without having to do the work of creating a voice. In human writing, that disruption may be the voice.
To be honest, I love a good em dash. For better or worse, I’ve employed them often in my writing for as long as I can remember. It is a great way for me to express how my thoughts—or my character’s thoughts—interrupt themselves. It is a way to have a breathless excitement, thoughts crashing into each other. A way to provide context the same way I would as an aside to a friend during a conversation. Removing them means commas or parenthetical statements, and while I can and do employ those methods, sometimes an em dash is the right tool for the job. But there have been so many moments in revision that I have removed them for fear that they’ll raise the specter of AI generation. I’ve tried to move away from the rule of threes and many other tricks that provide easy frameworks but become energizing when used in interesting ways. Of course, there are many other elements of syntax and style that indicate something is human generated. But some of those things require closer reading or reading out loud to hear the variation in pacing and rhythm, and it is easier to point the AI finger by relying on those signals.
It is an odd predicament. The pressure to produce content is massive. No matter whether you are in a corporate role, a creator, or an academic, the need to feed hungry audiences and stay relevant is ever present. The best way to meet this need is with the support of AI tools, so long as the writer can prove human hands and human authenticity. I can see a future where content is certified human. I am not making a judgement on that future, in fact if anything, I think the human premium is essential to the future of art and literature. I imagine myself walking through a bookstore and picking up a novel with a big gold star that reads “100% Certified Human Grade” like how we label organic beef and free-range chickens.
But none of that matters if the work can’t reach an audience. In a similar way to how writing has to pass the human sniff test, content must also pass the algorithm. We’ve seen similar adjustments in the past. Search engine optimization turned web copy into keyword dense tools to attract visitors and rank on the first search engine results page. I remember the days of keyword packing and back linking to send the right signals only to have Google update the algorithm to reward time on site or load speed or some other lever to pull. If you went back and looked at corporate blogs and mapped them by Google update, you’d be shocked at how much influence those updates had on how we engage an audience.
SEO has given way to algospeak. Where SEO was about feeding the algorithm what it wanted to hear to get noticed, algospeak is about obfuscating what it dislikes to avoid being silenced or punished with a ban. A way to slip subject matter past the algorithm to reach an audience. Content creators use words like “unalive,” “corn,” “accountant,” and “mascara” to slip moderated language past the algorithm filters. In many ways it reminds me of how films used similar tactics to undermine the Hays Code of the 1930s to 50s. Subtle turns of phrase, loaded glances, and visual shorthand allowed filmmakers to tell real human stories without triggering the studio alarms. The messages were there for those looking for them, but invisible or explainable to the overseers who controlled what films were allowed to be seen. We’re doing the exact same thing today. We’re trading rich vocabulary for coded slang just to convince an algorithm our content is safe. It inspires ingenuity but it also elicits a “what the actual duck are we doing here?” (Thank you autocorrect.)
Technology and communication are forever linked. Folks started to write differently when the printing press rolled around so this is nothing new. And often, these changes come because communication gets easier. The audience broadens. But there are aspects to this we should keep an eye on. Do we disregard AI generated content? Do we tier it? Are there purity tests? And how much are we willing to cede to pass the test?
I think back to that class. That hot workshop, all of us staring at each other, sharing our work and reading our essays. We had many conversations that quarter and subsequent quarters on AI and how it would impact the work. But the overwhelming fact was that we were all writers. Our drive was to write, to create, to articulate our observations and our experiences of the world. The question is whether AI becomes an amplifier, a collaborator to push our craft to new limits, or becomes a liability, a way to avoid the deeper, more thoughtful practice required to produce the work we hunger for. Or, maybe the proof of human doesn’t rest in the absence of AI but in the presence of genuine, embodied attention.
Leave a Reply