Another day, another revolutionary change in the world of search engine optimization. I kid, of course; this does feel like the start of something actually new, but let’s just say artificial intelligence (AI) tools for SEO are off to a slow start relative to all the hype about its current capabilities. And why, then, you shouldn’t take anything at face value you read, hear, or watch on YouTube about when an AI-powered economy will form or what it will look like.
Yes, the rollout of AI for mass consumer consumption, including AI upgrades to essential SEO tools, has been well-funded, the technology is impressive, and AI will completely change SEO. Likely not the functions currently required to optimize a website for search engine visibility in the near-future landscape, but in how they’re performed and the quality and quantity of the output – eventually.
We should, however, keep three things in mind when talking about artificial general intelligence (AGI) and AI in SEO:
- Artificial intelligence is a product like almost everything else.
- As long as Google’s in charge of how websites rank in their search results, and they continue to dominate the search engine market (currently 91.37%), what they say on SEO goes. What they say with every update is what they’ve been saying since the beginning – the way to future-proof SEO success is by publishing useful/helpful content, designing an inviting UX, and taking care of your website’s back-end technical workings.
- Revolutions take time. Despite how things sometimes appear, nothing happens overnight.
So, let me explain why you should manage your expectations of AI’s impact on SEO.
Artificial Intelligence is a Business, Like Everything Else
The marketing behind AI products overall is undeniably successful. Sure, science fiction has ruminated over the implications of artificial intelligence with respect to human vulnerability since before the personal computer, so the hype was already built in.
But AI developers and VC investors have used the myth of AI and married it with grand promises about its potential use cases, and now it’s all the business world is talking about. With good reason. If AI developers deliver on these promises of automating various business functions, business owners are looking at a massive rise in productivity and profitability.
Unfortunately, an unbiased, honest assessment of where the technology is at and realistic predictions of its future potential are nearly impossible to find. Since those leading, working for, or investing in AI businesses are essentially our only sources of information at this point, it’s hard to know if their claims are sales-driven or not.
For AI to develop any further, it needs unspeakably large amounts of capital investment, and all the CEOs of leading AI firms have been on a world tour touting its benefits to drum up the financial support needed to fund the vast amounts of “compute,” i.e., energy, needed before AI can become a potentially viable choice over human cognitive labour.
There’s also the possibility, we must acknowledge, that the accelerated timelines and possibilities of AGI being touted are not that far off. There’s the recent and viral example of Tyler Perry putting a planned $800M studio expansion on hold after seeing OpenAI’s text-to-video model Sora in action.
For our purposes, however, let’s focus on AI and its potential for automating SEO.
AI vs Search Engines
Are we/have we been using AI for SEO? Yes, absolutely. It can cut down the time needed to complete tasks to varying degrees depending on the task. AI’s ability to analyze vast amounts of data is undeniable, and because of this, it can make certain SEO tasks possible that were previously a lot more cumbersome.
As for causing radical alterations to search engine results pages (SERPs) that upend how SEO is performed, eliminating the need for a link-building expert, for example, automating it altogether, or circumventing the use of search engines in the first place?
In the near term, no. But what either scares people in all industries, or excites them depending on their side of the paycheque, are the exponential growth curves in performance that might reasonably be expected. The unpredictability around if/when they will happen and the level of reconfiguration and resulting adaptations to each business sector should scare everyone, at least a little. Will you even have a viable profession or business in 10 years, or will you have to eventually take that leap into something else entirely?
SEO is as good an industry to look at as any, considering the most popular use of AI is ChatGPT and other large language models (LLMs.) Their very nature of generating human-like responses to user queries is a double-whammy for SEOs if claims about their potential hold true.
Potential of Large Language Models to Replace Search Engines or Fully Automate SEO Content Creation
First things first, if you can get the information you need from an LLM, there is no need to use a search engine, which could mean the term ‘SEO’ goes the way of the term ‘rotary-dial phone,’ and the concept of optimizing your website for maximum exposure migrates to where the majority of your target audience gets their information.
And on the other hand, if LLMs can produce effective marketing communication at scale as promised, you can automate SEO copywriting, which, from the looks of things now, would be the last hill for LLMs to conquer in a campaign to automate a business’s SEO/marketing functions.
It would be foolish to make a bold prediction either way, but here are a few reasons we feel, based on our experience and the information available, neither scenario seems likely to happen anytime soon.
Despite the Complaints, Search Engines are Still a More Efficient Way to Get Reliable Information
In a recent conversation with Lex Fridman, Sam Altman, OpenAI CEO (the company behind ChatGPT), stated his belief that AI can provide a “much better way to help people find and act on and synthesize information” than Google or another ‘Google copy,’ and that he feels ChatGPT “is that for some use cases.” He also said that he’s not interested in creating a better search engine.
He prefaced these opinions by raising common complaints about sponsored links and ads dominating Google Search engine results pages (SERPs) and intimated his opinion that showing searchers “ten blue links” doesn’t provide them with the information they are searching for.
While there are valid arguments that Google Search has to do a MUCH better job at providing links to pages that provide reliable answers and detailed information a user is looking for, asserting that ChatGPT is a better source of information in “some use cases,” is subjectively untrue and, in fact, ChatGPT is currently a worse, and a lot more dangerous, source of information. Here’s why.
AI Hallucinations & Lack of Reading Comprehension
Hallucinating is basically when an artificial intelligence just makes shit up. This can be anything from ‘facts’ that sound realistic to fabricating imaginary sources it cites for its responses. AI has also failed mathematical tests, bringing its data analysis abilities into question.
This is partly due to AI models, and LLMs especially, having sucked up almost all of the world’s digital data in their training. (And almost always gathered in nefarious ways – by all the players.) Which means their training data includes a lot of inaccurate online information and biases added to the biases inherent in their designs, reflecting those of their designers in a system that, despite its name, is not what many experts consider intelligent. (As an aside, don’t be surprised if it comes to light that these companies, especially Meta, use private, personal communications to train their models.)
Another issue is that of AI’s inability to understand information fed to it directly. As an example (we could have chosen many, btw), after ChatGPT-4 identified the right website for information on a Canadian taxpayer’s obligations for filing taxes as a non-resident, it came up with the following assertion after scanning the page (look at point 3):
This is inaccurate. The CRA only allows non-resident tax returns to be filed by mail.
An LLM doesn’t know what it is ‘saying.’ Its outputs contain and are based on data, a.k.a. words it has no understading of beyond a mathematical value. LLMs are built to become better at predicting what the next word in a sentence should be.
In a recent poll by, ironically, an AI-based solutions provider named Tidio, 86% of respondents had personally experienced AI hallucinations, with 46% experiencing them frequently. So, using an LLM as your sole answer provider is not a good idea.
The Power of Multiple Sources
Despite the many complaints about Google Search being a lot more frustrating to use, it still provides value simply by showing you multiple sources to navigate and allowing you to decide for yourself which ones contain enough information to answer your question.
This isn’t the most efficient way to get an answer, which is why Google constantly upgrades its algorithm to better rank helpful, organic landing pages that convert and continue generating ad revenue at the top of their SERPs.
In the meantime, Google is also testing the addition of AI overviews to their SERPs, similar to Perplexity AI’s search engine, which provides an AI-produced summary of the issues to consider in a search query based on the search results. Time will tell if these AI-enhanced search results outperform the ChatGPT/Microsoft collab of Bing’s Copilot and how courts and users feel about AI content that presents a verbatim ‘copy/paste’ of pre-existing online content. But note, they still provide links for you to check their work.
Using AI for SEO Content Production
SEO keyword research has already been indirectly using AI, and AI enhancements to tools like Ahrefs and SEMRush can provide even deeper keyword insights on competitor strategies and potential keyword and content opportunities. AI can also streamline SEO functions by automating SEO tasks like website audits and creating local citations.
But because a constant flow of valuable knowledge contained in blog posts and guest articles is essentially what draws organic traffic to a website, many business owners, including ones that provide SEO services, and, of course, people who create content for a living, are absorbed with the question of whether or not large language models (LLMs) like ChatGPT can fully automate that process. The answer is no, for now.
Aside from the issue of the accuracy of AI outputs, there is also the question of value. We’re not talking about how well ChatGPT-4 can write, yet. ‘Value’ here refers to the helpfulness of the information provided.
You Can Rely on ChatGPT 4 and other LLM Artificial Intelligence for Short Bursts of Useful Content
Because LLMs don’t have experience in what you ask them to write about, they can really only help content creators draft outlines and write out passages of information that are easily verifiable and readily available online. By nature, this information is superficial and low-value.
For more complex topics, like finance, law, health, etc., LLMs are still too far off from accurately writing on these topics and need to be heavily fact-checked. They also have an annoying habit of constantly repeating themselves in the copy they write.
So, while an LLM like Chat GPT-4 can help speed up the writing process, it can’t help with research or produce accurate and genuinely helpful content without time-killing oversight and revisions.
Your SEO Content Must be Original
Not only do a lot of currently high-ranking articles, landing pages, and blog posts not provide enough knowledge to answer a user’s questions, but now, with website owners pushing out AI-produced content that goes beyond content written with the help of SEO plugins, we now find several blog posts on the same topic hosted on competitive websites that are similar enough to be considered identical because they use the same AI tools.
It’s become an epidemic to the point that Google’s March 2024 Core Update was dedicated to reducing “low-quality, unoriginal content in search results by 40%.” This update has led to massive fluctuations in website rankings, and future updates will undoubtedly get better at removing AI-generated spam content.
Content Written by an LLM like ChatGPT and SEO Value
So now, when you take into account the discussion thus far about the inability of current LLMs to determine fact from fiction, the true value a passage of AI-generated text has, and the instances of AI hallucinations requiring painful fact-checking, and add these pain points to the fact that AI content still has to be edited to sound human and additions of experience-based knowledge have to be made to satisfy Google’s EEAT guidelines, SEOs and business owners have to assess the value of using AI to create SEO content.
In other words, is it worth it, or is it more trouble than it’s worth to use ChatGPT-4 to write content for you? The answer is not yet, and maybe not ever – depending on your niche, of course. The fact that all possible data has already been used to train AIs and LLMs means all that’s left, essentially, to continue their training, is “synthetic” (AI-generated) data, which can lead to the problem of more garbage going in and, therefore, coming out. ChatGPT-4 and other LLMs, however, work well enough as a writing prompt, for article ideation, and blog post outlines for now.
AI developers are undoubtedly refining their algorithms and rules to help their LLMs and AIs provide better responses, but remember, Google has been at this for a while, training its AI and Search algorithm to better distinguish between accurate, helpful content and inaccurate, unhelpful spam for decades, and we see the issues involved.
All we can do for now is wait for ChatGPT-5 and take it from there.
Collaborate with leading SEO experts in Toronto to produce content that ranks well in search engines and connects with your customers. Reach out to us today!
