On November 30, research lab OpenAI released ChatGPT, a chatbot that runs on top of the company’s large-language artificial intelligence (AI) system, GPT-3. While the press has been writing about AI changing business and everyday life for years, for many people ChatGPT represented the first time that the AI-powered future seemed to have arrived at their doorstep.
What is the difference between GPT-3 and ChatGPT?
GPT-3 (short for “Generative Pre-training Transformer 3”) is the third-generation language model from OpenAI that uses deep learning to generate language that can sound remarkably human. As soon as it was released in 2020, digital marketers and people in the SEO industry (who were lucky enough to get access) started demonstrating remarkable examples of GPT-3 generating text that many would consider “good enough” for a lot of everyday uses. And offshoots of GPT-3, including OpenAI’s DALL·E, gave us a lot of laughs as we created AI-generated images in 2021 – frequently spectacular, sometimes disturbing, and often hilarious.
While GPT-3 was already impressive, OpenAI may have finally found its killer app when it released ChatGPT to the public. It’s a variant of GPT-3 that’s optimized for chat applications – you ask it a question or give it a task, and within seconds it gives you what you asked for. Ask it a follow-up question, and it remembers the context up to that point, so you’re not starting over with each question. Services such as Apple’s Siri and Amazon’s Alexa have been decent at answering users’ questions for a while now, but overnight it felt like the whole industry had taken a once-in-a-generation leap forward.
How can ChatGPT be used for SEO?
When GPT-3 and other large-language models were first released, swarms of people in SEO and content marketing said, “I bet you can generate a lot of content very quickly with this.” And now, there’s no shortage of copywriting tools that promise to unleash the power of AI to help you generate thousands of words in mere minutes.
Although Google was slow to directly answer the question of whether or not AI-generated content met their guidelines, John Mueller did eventually address the matter in 2022. According to Mueller, like all automatically generated content, AI-written content is against Google’s Webmaster Guidelines. Later, however, Google tweaked some language to put the emphasis on “spammy” auto-generated content. Just last month, the company changed its quality rater guidelines to add “Experience” to its “Expertise, Authority and Trust” criteria. Clearly, Google is still figuring this all out.
The next obvious question is whether Google can detect AI-generated content. While we doubt Google or any tool can perfectly detect it, Google’s frequent “helpful content” updates and spam updates in late 2022 suggest that the company may already be dialing in that capability.
While we know some search marketers who have had great success with AI-generated content, we also have heard stories of sites that were doing great until they suddenly weren’t. So, proceed at your own risk. This may be a cat-and-mouse game that plays out over at least the next few years.
So, let’s leave “write a million words a day!” aside, and talk about other ways that SEOs can make use of ChatGPT right now. This article isn’t meant to be an exhaustive list; the following are just a few examples of how we’re already using (or testing) ChatGPT here at Hennessey Digital. These ideas may give you a good starting point if you’re still becoming familiar with ChatGPT and its possibilities for SEO:
Translation text with terrific results
Google Translate has been good for a long time, so much so that we think nothing of visiting a website that’s in another language and clicking one button in Chrome to be able to read the page. The understanding has always been that it’s a “good enough” translation for most everyday purposes, but try to translate a large passage with the tool, and the output will be crude compared to what a native speaker can write.
Things are changing quickly, though. A recent comparison of ChatGPT and Google Translate shows that ChatGPT can produce translations that are almost perfect, picking up on language subtleties that Google Translate rarely can. The bar for translation quality just got raised significantly, and it wasn’t Google that raised it. At Hennessey Digital we’re already playing with potential uses for our clients.
Generate schema markup for your content
Generating schema is not terribly difficult, especially if you use a WordPress plugin, but now you can simply go to ChatGPT and ask it to generate schema for you. We’ve tested this a bunch at Hennessey Digital already, and we’ve so far found it to be very reliable for this repetitive task.
Write (and learn about!) regular expressions
For those who can easily write regular expressions (regex), they’re powerful search operators when using tools like Google Search Console and Screaming Frog. But even computer science majors can sometimes get lost in the sea of backslashes and brackets. Enter ChatGPT, which will not only turn your plain-English input into regex, but it will also explain what it did so that you learn along the way.
Generate meeting summaries
Okay, this isn’t strictly an SEO use for ChatGPT. I have already tried uploading a Zoom meeting transcript to ChatGPT to generate a written summary, and the results have been impressive. “I’ll send you the meeting transcript” is sometimes helpful, but when it is a two-hour meeting and you want to make sure you didn’t miss the highlights, ChatGPT can come in handy. With a bit of extra work you can even do it starting with just the audio file of a meeting. Note that you may run into word limits if the transcript is too long. If that happens, just break it up into several sections, and submit them to ChatGPT one at a time.
These are some really helpful uses for ChatGPT, but…
How will ChatGPT actually affect SEO?
We already have many examples of how ChatGPT can change how SEOs work, but what could the technology mean for the industry overall? The biggest change may come from what it means to search the internet in a couple of years.
Right now “search” means a user typing in a search term, and the search engine presents a list of links along with ads. But what if there simply is no more list of links? What if searching “car accident lawyer” means the application returns one or two results of nearby providers? No Page 1 of results to work hard to get onto? No 3-pack of map results to fight for?
But what if there simply is no more list of links?
These questions aren’t entirely new – people have been fretting about what voice search could mean for search marketers for a while now. But voice search, as useful as it’s become for certain uses, still seems rather niche, and it hasn’t made an appreciable dent in how most people search for things online every day.
This does feel different, however. Just as most of us said, “Wow, I like this better” the first time we used Google search, there has been a lot of talk about how people can now sense how things will change in the next several years. Maybe Siri and Alexa will get a lot smarter (competition tends to do that). Maybe we’ll use Google search a lot less, and just trust an app to give us what we need, without the need to jump from page to page in a long list of links, hoping to find that perfect macaroni and cheese recipe. Maybe the app will understand each of us better after thousands of conversations over time.
If that happens, search marketing may become even more of a zero-sum game. Getting to the top of the Google SERPs is one thing, but if the “search engine” (that term may even become obsolete one day) talks to its user and recommends just one thing, marketers will have to fight like hell to become that one recommendation. And if there are no links for users to click on, then generating traffic from organic search may become less important than it has been for the past two decades.
How will ChatGPT affect the search engines themselves?
Without a doubt, the search engine leaders are spending a lot of time on ChatGPT right now. Google’s management declared a “code red” not long after ChatGPT was released, and many believe that there’s a huge debate happening inside the company about whether Google should disrupt its own business with its own version of the technology, or if it should protect its existing model for as long as it can.
The first place we will probably see changes is in how users interface with the search engines. It’s been little more than a month since ChatGPT was released to the public, and there are already reports that Microsoft is working on a version of Bing that will use ChatGPT to provide more conversational answers to users’ queries.
Search engine upstarts You.com and Neeva have also entered the fray, introducing “AI search assistants” over the past several weeks. I’ve tried You.com’s YouChat, and it’s not hard to imagine the majority of search queries being answered this way in a couple of years, if not sooner:
Whether these enhancements truly make search more useful for users remains to be seen, but this all seems a lot closer to reality than it did just two months ago. Who knows… we may finally get the elegant question-and-answer experience that Ask Jeeves promised back in the 1990s.
What else might happen to the search engine business? Let’s give ChatGPT the final word: