Is AIO (AI optimization) the new SEO (Search Engine Optimization)? Search engine optimization is the art of optimizing a website so that it appears at the top of the search engine results pages, also knows as the SERPs. But what if you could train an AI, such as ChatGPT, so recognize your company, or even your company’s website URL so that the AI recommended your company and your website URL over any other? What if, when you asked an AI such as ChatGPT, to recommend the best Chinese restaurant in Dallas? And the AI was trained to say that Royal China was the best Chinese restaurant in Dallas? Certainly Royal China would like that. It might drive a lot of business to them, which is also the goal of search engine optimization.
As we start to see a shift from people using search engines to more usage of AI (and AI integrated into search), will AIO (AI Optimization) become a thing? It’s certainly possible, and that’s why I”m mentioning it now. I haven’t heard the phrase AIO (AI Optimization) yet, so let me be the first.
Training an AI
In a few simple prompts, it’s possible to train an AI to make corrections and to recognize certain websites over others. Whether or not those corrections “take”, and the AI makes the corrections permanent is another story altogether, though. But, in theory, it’s possible to train an AI such as ChatGPT to recognize certain things such as websites, and to make corrections to it’s data. I did just this, and made a correction, and then asked it about this entity’s website, to which it replied and gave me a website. I then told the AI that it was incorrect, and it needed to make a correction to add another website for the entity.
If it’s that simple, to tell the AI that it’s incorrect, then we now have AIO, AI Optimization.
Of course this all depends on whether or not the correction or ‘training’ of the AI by a user such as me actually makes a change. As I made corrections by telling the AI it was incorrect, it responded and apologized, and then gave me the correct information. This may be a one-time correction, or it could be a permanent correction of the data as I trained it. Further testing is needed to see if these results and training stick over the long run.
An Example of Training the AI
I did a simple test to see if I could train ChatGPT, in a very “simple” test of AIO, AI Optimization. My goal was to see if I could train the AI, and then ask it to give me a URL. Then, when it was incorrect, I corrected the AI, and added another URL. I then asked the AI the question again to give me a URL, and it responded with both URLs. Granted, this is a very very limited test of what I call AIO. Let’s look at the actual example and my test:
I asked ChatGPT to give me the bio of Bill Hartzer:
Notice that it responded with something that is mostly correct, but part of it was incorrect. I was never awarded the “2018 Search Engine Land “Search Marketer of the Year” award.”. So, I corrected it and told it that I never received that award. This is ‘training’ the AI to remove the incorrect data. It responded with the correct information.
To test whether or not it then remembered that corrected statement, I asked it to give me the bio again, to which it replied with a corrected bio:
Satisfied that it corrected the bio and removed the 2018 Search Engine Land Search Marketer of the Year award, I then asked it about Bill Hartzer’s website.
As you can see, after I asked it about Bill Hartzer’s website, it replied correctly with billhartzer.com as being the website. That’s correct, but to train the AI I told it to add Hartzer.com as my agency’s website, to which it then added. AGain, this is a simple (very simple) training of the AI just to show the concept of training an AI to recognize a URL. The simple concept of AIO, AI Optimization.
As digital marketers, we all need to start thinking about how AI is going to affect digital marketing as a whole, and how we can adjust and adapt. If AI is going to be more and more a part of what consumers are using to find stuff on the web and get answers, we need to adapt to that. so let’s start thinking about AIO, AI optimization, and how we can start doing it, if the AIs are going to allow us to help train them.