Have you heard the buzz, or have you been hiding under a rock?
Everyone in SEO is talking about GPT-3, and for good reason. We’ve made some elements of AI-generated content part of our everyday lives (autocomplete in your Gmail emails, for example), but how does GPT-3 content differ?
Created by OpenAI, GPT-3 content generators have polarized marketers and content writers.
Some people in the industry worry about being replaced by this new tech, and others relish the opportunities the new tool brings.
So, who is right?
In this article, we’ll explain what you need to know about GPT-3 for SEO and how it’s going to affect your content writing.
What is GPT-3?
GPT-3 (short for Generative Pre-Trained Transformer 3) is a deep-learning language model developed by OpenAI. It has been trained using internet data to develop text that actually sounds like it was written by a human, based on only a small amount of input from the user.
GPT-3 learned to spot patterns in human language by consuming a huge amount of information from the internet. From there, it uses natural language generation to create output.
Source: Towards Data Science
So you type in a sentence or two, the tool will analyze the language you’ve used to continue the prompt, and predict what content will be most useful.
GPT-3 content generators have been used to create everything from poetry to news reports. This has even extended advertising copy and computer code.
What Makes GPT-3 Different From Other Versions?
As the name suggests, GPT-3 is the 3rd model of its kind. So what makes it so different from its predecessors?
Before GPT-3, Microsoft had the most expansive trained language model around with Turing NLG. Turing NLG had only 10 billion parameters, while GPT-3 has 175 billion. It is this massive increase in training that enables the tool to create text that mirrors human language so closely.
Source: GPT Crush
While we’ve had access to AI tools for a while, GPT-3 is the first to work consistently across different tasks. Some of the best AI writing tools can create programming code that actually works, mock up whole websites in the wireframing tool Figma, and even create memes and comic strips, like the example above.
Some platforms have even begun to be used in whichever platform you’re working on, thanks to extensions. For example, Jasper lets you auto-populate content while you’re writing a message in Gmail or chatting in Slack. This is a stark change from the past, when previous AI tools (such as the autofill in your Gmail account) were restricted to their native platform.
Minimal Adjustments Needed
Previous versions of GPT (and other AI tools) have needed a lot of adjustments from human writers and editors to make them sound, well, human. GPT-3’s vast database of language learning can spit out an article that only needs minimal adjustments.
In fact, a New York Times article in 2022 judged that GPT-3 could create prose that matched the originality and fluency of a human. It’s important to remember that while GPT-3 can emulate human prose, specialized content will likely need a human touch.
How GPT-3 Ties Into SEO
This is all pretty interesting, but what does it have to do with SEO?
Can GPT-3 content help you boost your Google ranking? Or is it going to harm your position?
Just because a piece of content hits all the necessary keywords doesn’t mean it’s headed for page 1. Especially since the release of Google’s Helpful Content Update, there are other considerations when you’re looking to create search-worthy content.
In the image below, you can see how AI tools can be directed to account for specific SEO-based directions in real time. The key is to remember that machine-created content isn’t perfect. You’ll still need to adjust the content based on your SEO needs once it’s in review.
When you’re creating GPT-3 SEO content, you also need to focus on creating helpful content. Google’s algorithms prioritize original, helpful content, as this update was developed primarily to fight the rise in keyword-stuffed AI-created content that wasn’t helpful to readers at all.
This means that in order to keep those rankings and get in front of the right audiences, you can’t let GPT-3 run off and create content on its own. You’ll need checks and balances.
These checks and balances are real writers and editors that make sure what GPT-3 is producing is helpful to your audience.
Here are some tips to help you create helpful content:
- Write for an intended audience that will find your content helpful
- Provide valuable takeaways to readers
- Create content with first-hand expertise
- Be sure your site has a primary objective or purpose
- Be knowledgeable on Google’s updates and guidance for product reviews
- Refresh old content
What Does GPT-3 Mean for the Future of Content Writing?
Now for the reason you clicked on this article in the first place: What does GPT-3 content writing mean for my job or my clients?
Relax — it’s good news. Let’s start with the things GPT-3 content generators can really help with.
Create Short Form Copy at Scale
You can use GPT-3 content generators for tasks like alt text, footers, and meta tags that are important for SEO, but not overly complex to create. This will free up time for the SEO pros on your team to focus on more complex tasks.
Draft Long Form Copy
When it comes to longer pieces of content, the sheer volume of data GPT-3 has consumed makes it a useful tool for drafting. If you have to produce a large quantity of copy in a short space of time, then it can help you with ideas and outlines. For example, you can use AI to build out an outline, which you can then turn into a more finalized piece of content — with a human touch added, of course.
With the demand for content continuing to grow, the need for more output without additional spending will be the balancing act for many marketers.
Have Writers Work More as Editors
While we don’t think using GPT-3 content is going to make copywriters obsolete, we do think your role might change. With the speed that these tools can create basic drafts, it’s likely there will be less sitting down and looking at a blank page, and more editing and refining. Humanized AI-generated content should become an integral part of your content marketing efforts to increase efficiency while focusing on quality.
Leveling Up the Copy
GPT-3 has some impressive language skills, but it’s only learning from what’s already out there on the internet.
Real human writers can take its output as a starting point, and inject creativity, empathy, and knowledge of their audience. They can also make even standard copy unique to the industry they’re writing for.
These are areas that GPT-3 is a bit lacking in at the moment. Does anything else fall into this category?
Disadvantages of GPT-3 Content
This brings us to the not-so-good parts of using GPT-3 in marketing. What do you need to watch out for when using the tool? Where does it fall short of human writers?
Copy Could be Too Similar to Competitors
Even with a whole internet of content, when everyone is using the same tool, there’s potential for duplication. GPT-3 marketing content across an industry could end up all looking similar. A large set of similar content has implications for SEO and customer trust. For example, if you’re writing on a very general topic like traffic lights, there are only so many ways you’ll be able to say something about them. AI-created content on this type of topic may become repetitive as more people use GPT-3 to write about it.
Copy Could Have Incorrect Language
While GPT-3 is incredibly impressive, it isn’t 100% foolproof. It does make mistakes, repeat itself, get into circular arguments, pull fictitious stats, and in some examples, has put inappropriate content in its output. Putting a piece of GPT-3 content out into the world without a thorough check is asking for trouble. The image below shows how easily a GPT-3 created stat can seem legit until a source check reveals it’s not real.
May Not Work for Every Industry
If you’re working in a particularly technical industry, or one which requires copy packed with empathy, then GPT-3 probably won’t meet your needs. While the range of uses for the technology is extensive, relying on it completely is ill-advised.
According to Google’s Search Engine Advocate, using AI-created content is against Google’s Webmaster guidelines, and seen as spam. However, Google doesn’t currently claim to be able to detect the content. Rather, its webspam team will remove any content they find to have been automatically generated.
GPT-3 can be used by anyone, including businesses. According to OpenAI’s terms, you are considered the creator of any work you produce with the tool, so you own the rights to it. However, its pricing model may put it out of reach for smaller businesses looking to create large amounts of content.
GPT-3 is currently the most advanced AI language processing system in the world. GPT-4 is currently in development, being trained on an even larger number of parameters.
Because it was trained on the internet, output from GPT-3 content generators holds many of the biases found online. It can often repeat itself, or produce similar content to what already exists on the internet, and it still needs an editor to ensure reliability and accuracy.
So is the future bright for GPT-3 marketing?
While we’re excited about the giant leap GPT-3 has made in AI content creation, any business needs a strong content strategy, and skilled writers to execute it.
To create content that provides your target audience with real value and ranks high in search results, you’ll need to add a human touch.
When used wisely, AI tools are a great way to supplement a strong core content team. I’m a firm believer in this, with a variety of my own offerings.
Ubersuggest’s AI paragraph rewriter helps improve content or repurpose it for a different use. My Headline Generator makes it easy to create attention-grabbing headlines for articles and social posts. The Listicle Generator takes content creation one step further, by helping you create an article people want to click on.
Why not try them out for yourself?