Unfortunately, the way viruses spread is something that we’ve all become quite aware of after having faced a pandemic that controlled all of our lives for a while. You know the drill. Someone coughs. Little tiny bad guys travel across the room, just as your friend inhales. Those viral particles set up shop in your friend’s body. Their business plan is very good and they start pumping out new models using their converted factories. Pretty soon, your friend also coughs. The cycle goes on.
But have you noticed that the same situation repeats with many other things too? A song will worm its way into the ears of everyone you know. Memes make people laugh and go from cell phone to cell phone. We even say that something like a popular video on YouTube has gone viral.
That also happens with words too, and not just new slang (like bro and 6-7) but even ordinary words. In High School, my best friend instead of calling something a thingamajig when his brain was buffering, he would say puppy-dog, as in “Please hand me that…a…that puppy-dog.” Because of him, I say that too, even though he has long since abandoned the habit.
I’ve since heard other people say that, because I do. They’ve never even met my HS friend. Possibly, long after my death, some kid in Australia will say that, after the term moved on, just like a virus does.
I’ve been thinking a lot about that over the last few years. I’ve kind of been obsessed with this viral nature of language. As a writer, I’ve studied how words and phrases move through the world. As an avid user of AI tools now, I have to also be aware of how that writing is being perceived, the very words that I’m using, and how people are reading my very words through the filer of “is that human?” This is even harder as words and phrases are carried along person to AI to person, just like viruses.
A lot of us are less aware of how this happens in the more sinister corners of the internet. But rest assured, this affects you, even if you aren’t using the Dark Web.
The Dark-Mirror
Have you ever gotten one of those scam Nigerian prince emails? The first thing that you may have noticed was the improper grammar or spelling. You’d laugh at the stupidity of the scammers. But, did you know that was more of a feature than a bug in the message?
See, the scammers knew that if you are the type of person that would instantly notice that and scoff, then you wouldn’t fall for the scam anyway. But maybe you would read the email as someone desperate for help, with English as their second language, and miraculously they were reaching out to you for help.[1] The scam was meant to trick the uninitiated and those whose hearts were inclined to intercede when someone needed their help.
The arms-race between the vast public (mostly consisting of Westerners, who much of the rest of the world has deemed to have much more in dollars than sense) and the shadowy villains who want to scam them has gone up a level in recent days. Now, the scammers are using better technology tools in their efforts.
Today, I got a text. It said in summary that their investors had noticed me from my work with The Marketing Cowboy, then quoted something I’d said on LinkedIn, and said they might be interested in talking with me about how I might help with their project.[2]
I responded, “OK, tell me about your project.” I was pretty sure that this was a scam, but I was game to play along.
They responded with an even longer message about the project, responding quite clearly to what I’d asked. At the end was something like, “My bosses would love to discuss it with you. Do you have another messaging tool, like WhatsApp?”[3]
First of all, I don’t have a WhatsApp. I don’t want to have that. Finally, they somehow (and no, my cell isn’t listed on LinkedIn) found me on my cell, so just keep talking to me on my cell.
I responded, “No. We can keep messaging here. I don’t need to talk on WhatsApp. There are too many scammers who use that.” A subtle dig, and I knew that I probably wasn’t talking to a human. I just hope that I made someone laugh. I couldn’t help myself.
They didn’t respond.
I also don’t care. It was a scam. I can’t stop it from happening. The scammers don’t call me personally from call centers in Kolkata any more than Sam Altman is asking for my advice about Chat GPT. I’m a nobody in the AI circle of life.
But that’s ok. I’m fine with that. What I’m not fine with is the increasing amounts of sophistication that they are using to try and separate me from my money.[4]
They are using legitimate AI Engagement tools under false pretenses. They claim to represent real businesses and those running the tools don’t know any better. They are also not really motivated to find out, when that investigation could result in the loss of a paying customer, no matter how bad they may be.
If they move you to WhatsApp, then the AI Engagement tool has no ability to monitor what happens after that.
The other option is to use in-house AI. That is expensive, but will at some point become the norm. At that point, my guess is that they will cease trying to get you to move to WhatsApp. That just involves another layer that is out of the scammer’s control, and in some way opens them up to getting caught.
Why am I bringing this up in a conversation about language? Good question. The increasing use of details and good grammar will lead to more distrust of these details, and well-formed sentences. When I see increasing levels of personal data and precise language, my distrust grows in proportion.
This is the leading edge, not the AI posts on LinkedIn. I don’t know about you, but I’m already starting to ignore those detailed LinkedIn posts. If you aren’t yet, you will. I’ll bet on it.
AI Slop
A lot of hay is being made right now about “AI Slop.”
“It’s over-polished” someone says.
“I ignore all em dashes.”
You might hear someone say, “It just doesn’t resonate with me.”
That last one, I’ve actually heard. It made me laugh out loud (a real LOL). Nothing perks up my ears more than when I see the word “resonate.”
Google has a tool call Google Ngrams. It tracks the usage of a word in printed literature over time. It is hard to use in this instance, because it doesn’t track things as recent as AI is, but I’ve also used other tools in addition for this too.
Basically, though, the word resonate used to be almost completely restricted to its use as a musical term, as in, “The tuning-fork resonates” In writing, people would seldom use it in any other way. There was no law people were following, of course. They just didn’t use it in a metaphorical sense.
But a few years ago, some clever writer thought that it would be a great idea to use it as something applied to a person (i.e. “That resonates with me.”). Much like Mrs. O’Leary’s cow, it would make sweeping changes in the landscape. It caught on.
Mind you, people were not using it a ton, but it was in common usage. You probably saw it. That’s why when you read it from AI you thought nothing of it.
See AI loves that word, because it is versatile. It lacks both positive and negative judgement.[5] When I say “That resonates with me,” you aren’t sure from that sentence alone whether I like the thing, or I hate it. It just means that the stimulus is reactive. It’s a word that a junior-higher would love. It requires a whole other sentence or phrase just to determine the goodness or badness of a thing. It pads your word-count.
“Bud Light really resonates with me. When I see that label, everything in me says get away, and fast.” As a phrase, it might taste great, but it also is much more filling. It really fills up the page. It is a pretty thought, making a big impression on you, but it doesn’t actually tell the reader much. But no matter it’s origins, people like the word.
So, when LLM’s found it, they latched on. It has become a huge hallmark of AI writing, but no one noticed. Don’t use it for AI detection, either. You will fail at that task, because of what happened next. That is a great case of the virality of language.
As people began reading AI and copy-pasting, then reading what others had copy-paste posted, we began to incorporate that into our everyday writing. Now I see it all the time. Congratulations. Now you will too.
And boy is it everywhere. Now, resonates is used far more metaphorically than in musical use. If you said, “The tuning-fork resonates,” someone might say, “Yeah. It resonates with me too.” Every time you see it, know that this word is an instance of how AI actually virally affected the words humanswe use.
Em dashes and chiasms
The case that you often hear about the em dash[6] is related, although because we humans decided to universally loathe it so much, you won’t see it make the jump from AI to humans. It did jump from humans to AI, though.
Don’t believe me? I’ll prove it. Your word processor will often change the double-dash into an em dash automatically. See—I just did it! The reason that your Microsoft Word does that, is because you like it. You used that a lot. LLM’s didn’t create that on their own. You used to do that. You don’t anymore. Now, like a bad light beer, it resonates[7] deep in your cerebellum.
But the reason that AI likes it is basically the same reason that it likes resonate. It is both versatile and devoid of judgement. It can function as a comma. It can set apart independent clauses and lacks the weirdness that we’ve decided semi-colons carry. It can also function like a parenthesis, but seems to be less…well…parenthetical.
The em dash is like grammatical duct-tape. It does almost anything. AI loves it. We hate it. For us, it’s the COVID mask of the grammar world. It stops the spread cold, and using it makes all witnesses snicker.
Another known format of AI writing is the “It’s not X, but Y.” Usually, you will see it in some LinkedIn post trying to be way more clever than it really is.
“You might think that your struggling sales is the economy, but it’s actually your sales staff training.”
You thought that it was someone being clever. Nope. Most of it is AI now.
This one is harder to pin down. People have always tried to cleverly point out the genius behind their unique take on things. It justifies their writing. You might think that I’m doing the same here. Maybe so, but to be fair, I’m not offering any unique take. I’m just following the research.[8] And in case you are wondering, I used no AI at all in writing this.
The Greeks used to call this structure a chiasm, or at least half of one. An actual chiasm was a bit more clever and complex. This is really nothing new. It is people using AI to try and do what they used to have to do real intellectual labor to achieve.
I guess my point in all of this is that much of what we consider as AI slop is nothing new at all. It’s just language doing what language has always done, transmit person-to-person and evolve all along the way. Your grandchildren will be skibidy-ing because AI told them to skibildy. Their AI will skibidy because people skibidied a lot.[9] Meanwhile, we will talk about how anyone using that word is just copy-pasting AI.
Maybe we are not that much different than the LLM’s, anyway. I hope that thought resonates with you.
[1] I actually knew a kid in a church youth group years ago, who asked me for guidance. He was “about to come into a lot of money.” He told me that he’d been contacted by a Nigerian prince…and you know the drill. I tried to let him down easy, but I also wanted him to avoid falling for what I knew was obvious.
[2] Strikes #1 and 2— overly long and TED talk-type formal language is a giveaway, and I don’t know why scammer AI always loves the word “project.”
[3] Strike 3—You’re out. A scam will always try to get you to another less secure app like Signal, Telegram, or WhatsApp (the most commonly suggested).
[4] If you are a scammer, please be aware that I don’t have much of that anyway.
[5] According to my research, this is one of the keys to AI writing. It will come up later.
[6] If you don’t know, it is the double dash, that your computer word processor puts into a long dash automatically
[7] See what I did there?
[8] Well, and also like a good boy, what Claude tells me to do.
[9] If you are very confused, ask a 13 year old.
