Read the latest tech headlines and you’d think we’re living through the rise of the robots.
In a world where AI tools have recently passed the bar exam and the U.S. Medical Licensing Exam, it’s easy to imagine they can do just about anything – or will be able to before you know it.
Artificial intelligence is exciting. And there are a lot of things it already does well, even in the world of marketing.
But it still has plenty of blind spots. In fact, we’re only at the very beginning of the AI era. And if you try to use it for things it’s just not any good at yet, you could find it digging you into a hole faster than you can dig your way out. That’s not the kind of AI-driven automation you want.
Unfortunately, we’re already seeing it every day in search engine optimization.
And there’s one area where it’s having the worst effect on businesses like yours.
That’s AI content writing.
AI Content Isn’t Good for Search Engine Optimization – And It Won’t Be for a Long Time
The hottest area of AI research today is Natural Language Processing (NLP).
As the name suggests, NLP is all about giving AI the power to parse huge amounts of text so it makes connections and responds to questions in a human-like way. Long before ChatGPT burst onto the scene, we saw this in the form of AI chatbots that interact with website visitors.
And before Google released its new AI, Bard, it already used large language models to make connections between seemingly unrelated search queries. That’s why, regardless of how you phrase a query, Google can now connect you with dozens of variants that mean the same thing.
But when it comes to search engine optimization, ChatGPT and the rest are a little bit overblown.
“But ChatGPT can write like Shakespeare,” someone might say. “Can’t it write a simple blog post?”
It’s true that AI using large language models can generate new content based on the corpus of writers like Shakespeare. But that’s a long way from “writing like Shakespeare.” Because, no matter how great AI is today, one thing is for sure: It does not actually read language.
Today’s artificial intelligence can only mimic language. It has no understanding of what words mean, only a mathematical probability of which words appear next to other words. Knowing these patterns across trillions of words means it “sounds like” Shakespeare, Hemingway, or whoever.
But an AI can’t create Shakespeare. It doesn’t know if a statement is true, or even if it’s likely to be.
Its effect, it’s predictive text – just like what’s on your phone.
Students are hitting the limits of ChatGPT and the rest already. Since it doesn’t know anything, it can put together paragraph after paragraph of totally false information – the kind that may get a kid an F, a college student expelled for cheating, and a business owner taken to court.
Impressive as it is, artificial intelligence is only as good as the data it’s been trained on. And what data is being used? In many cases, it’s the internet at large, or specific websites with a lot of plain text content.
If you wouldn’t trust a random page on the internet to be factual, you should be leery about letting AI handle your content. It doesn’t know the difference between the truth and a lie told confidently.
To do better, it would need:
- To have a robust sense of whether any given statement is more likely to be true or false
- To understand what words mean (“denote”) and what they mean to people (“connote”)
In other words, before AI content can get really good, we’ll need artificial general intelligence – a true “thinking machine.” By most predictions, it’s still a long way off. And even if the ethics of AI aren’t your cup of tea, there’s a practical reason to shun AI-generated text for search engine optimization: Google.
Google Can (Probably) Already Tell When Content Is Written by AI
The arrival of ChatGPT sent shockwaves through education as teachers and professors alike sounded the alarm. For weeks, everyone thought it would become impossible to detect AI-written content. Grading would become impossible, exams would disappear, and the world would never be the same again.
Weeks before ChatGPT released its own detector for AI-generated text, an independent developer already had one. Today, not only is it easier to find out whether a piece of text is likely to have been written by AI, but businesses are writing angry clauses into their contracts to forbid the practice.
We know a few marketing agencies have already experimented with AI content. By and large, what they found is clients don’t want it – and neither do those clients’ customers. In all too many cases, you’ll end up with boilerplate content that’s almost fact-free (or, if it does include facts, is just plain wrong.)
And Google is on the case.
As of February 2023, Google has confirmed that AI-generated content isn’t against its guidelines. But that may not last long. After all, no other company has more AI resources. Not only does the Big G monitor virtually all online content, but it can also certainly scan to see if something was likely written artificially.
Over the last ten years, Google has made it clear again and again that it’s all about curating the best results to help search engine users meet their needs. Overwhelmingly, people are already saying the standard of content available through AI today isn’t doing that.
If you don’t think Google will act against AI, we have one word for you: Panda.
2011’s Panda Update Demonstrates Google Is Willing to Blow Up the Internet
In 2011, Google unleashed an update that changed the face of the internet overnight. It reduced the value of links from low-quality “link farms” to near zero, impacting millions of websites that held high rankings for valuable and competitive search queries.
The content on link farms was created by a combination of underpaid freelance writers (many of them writing in languages other than their native language) and “text spinners,” the forerunners of today’s content writing AI. While neither of these practices was against Google’s guidelines per se, the result was dismal content the company considered to be undermining its core mission.
At the time, virtually nobody believed Google would do it. In the years since we’ve seen Google can and will act to shape the entire online world. Using AI content might be building your castles on sinking sand.
There Are Some Things Humans Are Just Better At – Content Marketing Is One of Them
Google says “Content is king” – and compelling, informative content that helps your visitors reach their goals is best written by humans. An AI like ChatGPT can be great for ideation, but once that part is done, real search engine optimization experts should take over.
Contact us for content marketing with the human touch.