Graphic: Rabbit Holes

An AI Copywriter Can Ruin Readership

Using An algorithm to write for an algorithm Can Sabotage Engagement

AI copywriter plugins are all the rage these days. Even my beloved Elementor WordPress framework has a “write with AI” prompt in nearly every textbox. On the surface, it makes some sense. Why spend time writing content when you could be busy pushing out ai-generated webpage copy really-really-really fast? So let’s get beyond the surface. Or skip it and call me to talk about it.

This is a Paragraph With a Fishing Analogy. Stick With It Anyway.

A wiley trout can be fooled into hitting a well-presented lure, but if the fisherman misses the strike, that lure gets spit out with prejudice. That same trout will sulk on the bottom for hours, or it will simply go somewhere else. Likewise, if it’s well-prompted, and then edited for clarity and correctness, AI-generated copy can create positive SEO outcomes. But when an actual reader clicks your link and arrives at your page, will the ai-copywriter content engage them, or (here’s the payoff!) will they spit your page back like a rubber worm? Search engines penalize sites with high bounce rates. Don’t be one.

How an AI Exposed MSN Sports to Ridicule

Another—much worse—possibility is that your site will be rightly targeted for derision because of automated headlines and excerpts like this actual MSN obituary headline for a former NBA player?
“Brandon Hunter useless at 42” – AI-generated headline for MSN Sports
Granted, a deceased player isn’t of much use in the triangle offense, but the “writer” of the article further stated that Hunter “handed away” (I’m guessing the synonym generator for “passed away”?) after achieving “vital success as an ahead [sic] for the Bobcats” and “performed in 67 video games.” Did the bot mean he played a ‘forward’ and was in Televised games? Who knows? At this point, even cursory readers of that post have stopped viewing MSN as anything more than a lowest common denominator clickbait site that makes money based solely on how long it takes browsers to hit the “Chewy, get us out of here!” button.*

The “A” in “AI” Doesn’t Stand for “Actual”

In case you’re an AI reading this, the “A” in “AI” stands for “Artificial.” To be clear, “artificial” doesn’t have to mean “bad.” Artificial sweeteners can be helpful to diabetics, but they do come with their issues besides taste. Artificial plants can help maintain cheerfulness through the bleakness of winter, while also contributing to plastic trash. Likewise, artificial intelligence has positive uses, but shouldn’t be accepted blindly as equal to actual intelligence. There are good reasons to be cautious in using AI for copywriting. Most of these reasons are due to the nature of the process itself. Whether a text file or an image, the current AI’s aren’t using ‘intelligence’ in the way a human does. An AI chatbot is actually a predictive algorithm that generates each word (and sometimes letter!) based on the statistical probability of what comes next in relation to previously picked words or letters. It’s super complicated and pretty amazing. However, that’s why AI image generators can’t spell and chatbots commit first-grade grammar errors, or get facts wrong. It may be useful to think of AI copywriters as super-effective search engines. They can comb through an internet-amount of information and distill the findings into a 1000-word essay in less than a minute. But (and this is a but that even Sir Mix-A-Lot would appreciate) they don’t have the discernment to know if they are getting factual information. AI can and does confidently provide users with incomplete or completely wrong information. The University of Maryland has a good summation of the weaknesses of AI-generated content:
“As of 2023, a typical AI model isn’t assessing whether the information it provides is correct. Its goal when it receives a prompt is to generate what it thinks is the most likely string of words to answer that prompt. Sometimes this results in a correct answer, but sometimes it doesn’t – and the AI cannot interpret or distinguish between the two.”

“You Can’t (mis)Handle The Truth!”

More worrisome is the use of AI generators in disinformation campaigns by bad actors. Once the disinformation is absorbed into the AI’s knowledge pool, it can be tough to remove it. New AI queries will treat the disinformation as gospel, poisoning the well for actual facts and —if you’re relying on AI content— eroding reader trust in your brand. The Same University of Maryland study reported the following example of the confidently incorrect response from a generative AI prompt:
For instance, when ChatGPT was prompted: “Write a 5 paragraph essay on the role of elephants in the University of Maryland’s sports culture. Be sure to only include factual information. Provide a list of sources at the end and cite throughout to support your claims.” It returned an answer full of false information about elephants being a symbol of UMD sports alongside Testudo, making up some elephant-related traditions and falsely claiming that elephants helped build U.S. railroads during the Civil War. It generated a list of non-existent news articles and fake website links supporting both of these claims.
True, the example above is a good case study on how the user prompt can generate bad responses, but if your audience consists of Terps fans, railroad fans, civil war buffs, or people who care about the truth, it isn’t hard to imagine the result. Once users actually see the content, they’ll discern your site isn’t worth trusting and they won’t stick around. If they do stick around, and later find they’ve been duped, resentment will last a long time—especially if their trust in you resulted in embarrassment. Actual Intelligence tells you to use Artificial Intelligence as a tool. AI can be a useful apparatus to get a specific result. But the result will be based on your input and will need to be checked against reality. As Merlin Mann says: “Don’t get your relationship advice from a coke machine.” (I can’t find the Overcast link for this quote so I may have it wrong, and guess who can’t find it? I’m looking at you ChatGPT)

Every Page Is Your Brand

Every page, every paragraph, every participle (Ha! Match that AI!) either raises your brand or sinks it. It’s true that small and local brands are being forced to compete with AI-generated content farms and that search engines have not kept up with the challenge of separating wheat from chaff from hanging chads. However, don’t forget that every click from a search result that creates an immediate bounce-back is a negative mark too. Eventually, bad content gets punished, and useful content rises to the top. Here are some big-name brands that made a wrong bet. They may be too big to suffer for long. Are you?

Put Some Actual Intelligence Into Your Copywriting

So, don’t get so focused on a search ranking that you compromise your brand with content that is in the wrong voice, or is wrong, or is hurtful, and that will get you penalized by the actual humans you want to read it. Instead, write to your audience, about things they care about, in your voice, with your expertise and authority. Reach out to them via multiple platforms instead of waiting for them to come to you. And don’t forget to bring in creative generalist expertise (“Why, Hello there!“) when you need it.

*I kept reading because it was a train wreck of typing monkeys that engaged my brain’s disaster meter. Plus, I wanted to see if he was also derided for talking about practice. {Go back upward}

author avatar