Commodity Content

An uncomfortable truth the world is unconsciously grappling with in the age of AI is that most words and groupings of words derived from our shared language are Commodities.

When I started ghostwriting for Silicon Valley founders and C-level executes back in 2016, this is the vocabulary I would use. It was common for me to have two clients in the same industry (say “Marketing”) and for both clients on their respective content calls to say, “The key to marketing is knowing your customer.” But this sentence is Commodity Content. A world-renowned expert could say it, or a 22-year-old on Instagram could say it. As a result, this idea, expressed through this grouping of 8 words, derived from our open-source operating system for communication, the English language, possesses very little value. Not because the idea itself is bad, or unhelpful. But because it has no ownership, no unique association. Those 8 words are not identifiable to any one individual.

When people point to a piece of writing and call it “AI Slop,” what they are identifying (or trying to identify) is Commodity Content.

For example, when AI writes a newsletter about “How To Lose Weight,” the majority of the words and sentences and ideas on that topic are Commodities.

• “Track your food.”
• “Count your calories.”
• “Be sure to eat healthy.”
• “Prioritize going to the gym.”

That’s not to say these words and sentences and ideas have no (as in zero) value. Commodity Content in every industry has some value, especially for beginners. If you are brand new to the concept of losing weight and eating healthy, maybe you need to hear someone tell you, “The key to losing weight is to eat healthy.”

The reason this type of content now gets flagged as AI (either artificially with AI-detection software, or publicly by mobs of people trying to delineate between man and machine) is because a Commodity is something that is easily replicable. If you take that newsletter and cover the author’s name at the top, you have no way of knowing whose it is. You also have no way of knowing whether it was written by a human or a robot.

The next uncomfortable truth the world is grappling with, specifically writers across a wide-ranging spectrum of jobs and skillsets, is that Large Language Models are superior Commodity Content Machines.

I remember when I first started freelance writing in 2014. I was 24 years old, a year out of college, looking to make some extra money. A travel agency hired me to write 600-word SEO-optimized blog posts about popular hotels and restaurants in Chicago, Milwaukee, Ann Arbor, and other cities in the Midwest. I was paid $20 per blog post. At the time, I couldn’t understand why I was being paid so little. Those blog posts took work! They took effort! They took time! I didn’t know anything about these cities, hotels, or restaurants, so I had to look them up online, consume other people’s Commodity Content (blog posts written by other Commodity Writers), and then rewrite their words into something “new.”

Now, AI can do all of that work in 1/1000th of the time, with zero effort, for $20 per month—and in some cases, $0 per month.

This is disturbing for many people. And if I was a 24-year-old today, freshly graduated from college and thrown into the working-world, I imagine I would have a hard time reconciling my competition no longer being exclusively my peers (who went to superior schools, no doubt), but also hyper-fluent Commodity Content Machines. Finding a job, particularly an entry-level one to learn and build real-world experience, would be the equivalent of entering a chess competition and not knowing whether you should prepare to battle another human (a fellow student of the craft), or a supercomputer.

But the truth is, the declining trend of Commodity Content is nothing new. Commodity Content—keyword-stuffed, SEO-optimized blog posts; generalized How-To advice; technical walkthrough guides; medical questions & answers; etc. —has always been the least valued form of writing, and rightfully so. It requires the least amount of original thinking, the least amount of personality, and the least amount of skill. To write about topics I knew nothing about (like hotels and restaurants I couldn’t afford to visit in my early-20s), I would often read other people’s work online and repeat the same Commodity Ideas (otherwise known as “conventional wisdom”) but just swap groupings of words, trade adjectives, or accentuate the “voice” with colorful verbs and strategically placed adverbs. For example, I might take someone’s else’s Commodity Sentence—”If you want a fun night on the town in River North, Chicago, check out Osteria Via Stato.”—and repeat the same Commodity Idea in a “new,” slightly different way: “If you want to have some fun in River North, Chicago, and love Italian food, I recommend Osteria Via Stato.”

Which is precisely what Large Language Models do today, writing infinite Commodity Content out of open-source, unowned, non-specific language.

Just because you wrote it, meaning your fingers literally typed those words in that order, doesn’t mean those words are unique or different or yours.

You don’t own Commodity Content.