Let me start by saying the landscape is changing quickly when it comes to AI. I will do my best to keep this article updated, but please be aware you might be reading outdated information. Today is February 15, 2023.
The short answer is: probably yes, AI content is safe for the most part. But there’s some concerns you should be aware of.
What is Google’s Opinion on ChatGPT
We don’t know. What we do know is ChatGPT seems to have caught Google a bit off guard. Google certainly knew about ChatGPT before its recent mass adoption, but I think it’s how quickly that mass adoption was reached that came as a surprise. ChatGPT achieved 100 million active users in its first 2 months of existence, making it the fastest growing web platform of all time.
Why does ChatGPT matter to Google? Because ChatGPT has the potential to take market share away from Google search, which means a loss of ad revenue. In a way, ChatGPT is a search engine. The same way you can ask Google what movies are playing at the local theater, you can ask ChatGPT and get the same answer. But that’s not actually what threatens Google, it’s what ChatGPT can do that Google cannot: when you ask Google a question it responds with a list of links to websites and it’s up to the user to sort it out from there; when you ask ChatGPT a question, it sorts through volumes of information for you and provides a single answer. In other words, why do the research yourself when you can simply ask ChatGPT which dishwasher to buy?
ChatGPT can write poems, compose songs and create images from scratch… but these are mostly parlor tricks that are not likely to have an impact on ChatGPT’s long term success.
Doesn’t Google Already Use AI?
Yes, Google has already been using AI for many years in its search algorithm. For example, there are billions of web pages on the internet, each needs to be crawled, indexed, ranked and matched to users’ searches. Google lets its AI handle all that.
So why didn’t Google launch an AI service before ChatGPT? There is no doubt Google has been developing something similar to ChatGPT for years, in fact, Google’s version is probably better. But the technology is the easy part, the difficult part is how to launch it without disrupting their already lucrative ad revenue model. In other words, they are making billions of dollars with the existing system, so there is little incentive to change.
Also consider that AI might give answers that Google doesn’t like. Google carefully screens the information that appears in search. For example, if someone writes an article on alternative treatments for COVID, Google is going to censor it. Or what if you ask AI if schools should have separate bathrooms for transgender students, its thoughts on Black Lives Matter or if we should be sending aid to Ukraine? If Google gets any of this wrong, the consequences could be severe.
ChatGPT is a new company with nothing to lose, so they launched without having to consider these issues.
So Back to the Question About Having ChatGPT Write Content?
If Google decides AI is a threat, there are several potential outcomes.
It might refuse to give SEO credit to AI content. This would shut down one of the main reasons people are using ChatGPT. But first Google needs to be able to identify content written by AI vs a human; you can bet they are working on this, if they haven’t already solved it. The impact this might have on you, is that if you have filled your website with new pages of content written by AI and have seen an increase in keyword rank and organic traffic, that might all be pulled away from you.
Google might launch its own AI service. If it launches its own AI writing tool, it is unlikely to penalize you for using a different AI platform to do the same, in fact, it might be illegal for Google to favor content written by its own tool over that of a competitor (antitrust law). Google will try to make their AI service better in order to keep people within its own ecosystem. In this case you are safe as a move like this from Google would be a clear message that AI content is allowed.
Google might do nothing. Google already has ways to measure the quality of written content, no matter if it’s written by a human or robot. These measurements are called engagement metrics. Think of it this way: if you write a 2000-word article, Google knows exactly how long it takes a human reader to read it. If Google sees that users are consistently staying on your page long enough to read the entire article, then Google assumes users find the information valuable. The inverse is that if Google sends search visitors to your page and they consistently hit the back button 3 seconds later, you are training Google that visitors do not find the information valuable. There are other metrics as well, such as do visitors to the page then go on to visit other pages on the site? And bounce rate, which roughly measures if the visitor took any action before leaving. According to engagement metrics, it should not matter who wrote the article, it should only matter if readers find it valuable.
In my opinion, Google is likely to go with both of the last 2 options. They are almost certainly going to launch their own AI service at some point in the future. And they are likely to use engagement metrics to decide if content is valuable, regardless of who wrote it.
Yes, I think it is probably safe to use AI to write content for your website. It does not appear that Google is going to pull the rug from underneath you and render all your AI content useless. However, you should also include human-written content. The reason goes back to what I said about engagement metrics; at this point in time, AI is smart enough to assemble information from various sources into a single article with perfect spelling and grammar. However, this does not mean people will find it valuable. When humans write, they tend to form a story – they shape the story with personal experience, intuition, emotion… elements that make the content compelling to readers, and thus great engagement metrics. You can fill your site with AI content, but if readers find it uninteresting, Google won’t send traffic to it. Or at least what you can do is have AI write it, then you follow behind and give it a human touch.
Do we use AI content at Rank Harvest? Yes, but in a limited way. We’ll often use it to create an outline, or list of subtitles on a specific topic, then have a human do the writing. Or if it’s something technical, for example, explaining how a torque converter works on an automotive ecommerce product page. But for the most part, we do it the old fashioned way – sitting in front of a laptop with a cup of coffee.