Should You Use AI to Write Blog Posts?
If you asked me what question we’ve been getting most often from current and prospective clients lately, I wouldn’t hesitate before telling you that it is AI-related.
Sometimes, it’s skeptical: “You don’t use AI to write content, do you?”
Other times, it’s hopeful: “Can’t you just use AI to cut down on the time spent writing and reduce your price?”
The answer to both questions is no.
This is something we’ve wrestled with as a team for the last couple of years, since AI really became a thing. And although we continue to keep an open mind as AI evolves, we don’t see GPT being able to write better than an experienced, creative, and literate human brain.
To be clear, we are absolutely not anti-AI. Throughout the company, our team uses ChatGPT for lots of tasks, and entertain each other with AI-generated images like this one:
But beyond the simple, clear, and low-stakes (and fun!) queries like our request for Elizabeth noble cat portraits, we believe that AI reveals its deepest, darkest secret.
AI Is Just a Stupid Machine
I’m not denying that AI has changed the world, or at least the internet, as we know it.
It has opened up possibilities that were only available to us in our wildest dreams before (Elizabethan noble cats!), and in five or 10 years, I will likely take a different stance on it.
But for now, AI is great at computing but terrible at thinking. That means if you give it a direct, unambiguous query, it will give you a (usually) great answer. But AI has no real-world experience, no sense of logic, and no ability to reason, which is why you see hilarious TikToks of botched AI photos, or this horrifying (and inaccurate) depiction of shoe anatomy:
Computing is a hard skill, while thinking and reasoning are soft skills. Similarly, there are hard skills and soft skills to writing. The hard skills would include things like efficiency, spelling, grammar, word count, and paragraph structure. These are simple, concrete, and contained rules that AI can usually follow with little trouble.
But how does it do with soft skills, like accuracy?
When you type a command into an AI tool, it scans the internet for information that matches your query, quickly spots patterns that it assumes correlate with truth (in other words, it assumes that the most frequently occurring information must indicate the best answer), and then synthesizes its research into a cohesive essay. It’s easy to use, gives you fast results, and produces a piece of writing that is, objectively, fine.
That doesn’t mean it is great. And that definitely doesn’t mean that it is smart.
There is a reason that before you even type in your query, you see this warning:
In addition to accuracy, other soft skills that are essential for high-quality, authoritative writing include creativity, intuition, voice, critical thinking, context, truthfulness, making logical connections or comparisons across dissimilar concepts, uniqueness, and wordplay. AI, at least as it stands now, is horrible at all of this.
So, yes, AI writing is fast, easy, and cheap. But it is not (always) accurate, never unique, and certainly not authoritative.
Google Ranks Authoritative Content
Here’s where the discussion becomes extra relevant to SEO, and why we don’t envision AI being appropriate for this kind of content creation for the foreseeable future.
If AI writing is generated from content that is already found on the internet, there is no way that AI — in its current iteration — can produce anything novel or move the conversation forward.
And if the content you’re publishing is just a recycled copy of what already exists, why would Google consider your site to be contributing anything worth showing to its users?
As an aside, Google was founded by two Ph.D. students, so much of the logic of the Google algorithm is rooted in the customs of academia. Given this, it’s not hard to imagine the weight that Google would give a primary source (in this case, something written by an expert with personal experience with the subject) vs. a secondary source (in this case, AI, which learns from what others have written about their experience and expertise).
By comparison to AI, the professional (human) writers that we work with at Growth Machine are subject matter experts who know how to reiterate the important established information about a topic while also adding a unique perspective based on their (offline) personal experience and expertise. This is the kind of work that moves the discussion forward and helps further educate the user — which is exactly what Google seeks to do.
So yes, human writers are more expensive, but they are better able to write the kind of content that Google wants to highlight in the highest positions of the SERPs. What’s more, human writers compose more engaging content with unique information and are less prone to glaring errors.
As for efficiency, well, it typically takes an editor as much or more time to edit and revise an AI-generated piece to get it to Google’s standard than it does for them to review a human-generated one.
So the advantages of writing AI may not be as vast as they seem, and what’s more, using AI to write your website’s content may be undermining your ranking efforts.
Google Has Given Us a Clue About Their Position on AI Writing
As much as Google believes in and is investing in AI, I also believe that it is putting the pieces in place to be able to penalize AI-written content in search.
Despite many claims, there is no reliable way for a machine to detect AI-written content yet. Sure, there are tons of “AI Detectors” out there that claim to be able to flag AI-composed writing, but there is no secret code left behind by AI tools that these detectors are able to see.
They’re actually looking for certain markers of AI content, as a proxy for being able to actually flag AI writing. Incidentally, we’ve found that many of these markers align with SEO best practices for writing, such as simpler sentence structure, shorter paragraphs, and junior high or high school reading levels.
In fact, our VP of Editorial spent some time composing articles on the fly directly into a number of popular AI Detector tools, following some of our SEO best practices for writing. All but one of the detectors said that her manually written content was >50% likely to be AI-generated.
So, that brings me back to Google. I believe Google wishes it had the means to accurately distinguish between AI-generated and human-generated content, and suppress the AI-generated content. But, like AI Detectors, it has to rely on a proxy system for flagging content that is likely to be written by AI. That’s why I believe that the March 2024 core update focused on spammy and low-quality content was Google’s first step toward penalizing AI-generated content, along with other unuseful content, in the rankings.
Note some of the wording in their announcement:
Granted, this isn’t just aimed at AI-generated content, since Google is constantly trying to weed out low-quality, unhelpful content. But with every core update, Google signals and reinforces their priorities and expectations, and I don’t think that this is a coincidence.
AI Is Great! (Just Not For Everything)
Don’t mistake this for an anti-AI article. We love ChatGPT and use it regularly — but not for writing.
Here are some examples of what we do use it for:
- Research, ideas, and inspiration
- Outlines and prompts for internal content (never for client work!)
- Parts of job descriptions
- Spelling and grammar checks
- Punching up or refining email communication
- Short-form content, like some of our social media posts
As you can see, AI is a great writing assistant, if not a great writer.
I also asked ChatGPT what it thinks (just kidding, because it doesn’t think — that was a test!) the benefits of AI for content are, and it told me: efficiency, scalability, consistency of voice, and content enhancement. I would agree with most of those.
So, it really comes down to knowing when and how to use AI writing tools to your advantage. There are a lot of benefits to AI, especially from an efficiency standpoint. And it might be tempting to save money upfront by using a robot writer versus investing in a premium agency like Growth Machine.
But, ultimately, we know how to write content that ranks, and AI doesn’t, so which would you prefer?
about the author
Nora Schlesinger is the CEO of growth machine. She has been producing keyword-optimized content for the internet for over a decade and has very strong opinions on AI. You can read more about her here.