AI-written content has recently become a focal point for niche websites. The consequences and potential rewards have brought on a new wave of questions.
Sat Nov 05 Written by: Tyler Bishop
AI-written content has recently become a focal point for niche websites. The consequences and potential rewards have brought on a new wave of questions rife with complexity, according to Google’s often-referenced search engine front-man John Mueller.
In summary, John says that Google largely — but not inherently —considers most automatically generated content (AI-written content) as “webspam.”
🤓 FUN FACT: Google’s first team internally focused on rankings was established nearly 2 decades ago and was actually called the web spam team.
Google and the people that read, search for, and use content all see content — whether it was written by a university professor or a robot — the same way. These questions get to the heart of why and how AI written content can be viewed in complete simplicity…
1.) Is the content providing what someone is looking for?
2.) Is it unique in what it provides vs. alternative content available?
If one is true, but two is not, that means there’s zero opportunities for the website to create that content with AI because it will essentially be scraping the existing content available, which in Google’s eyes is redundant.
If 1 and 2 are true, but the information is extremely simple, Google doesn’t need people to use off-the-shelf AI to make short answers to questions. Google builds leading NLP systems, and they can have their robots write rich snippets if they need to.
The question isn’t whether or not AI can write content. It can. It can write really well with the right programming and application.
Artificial intelligence does its best writing when data and information is available, clearly defined, and need to be aggregated and summarized. AI can do a slightly worse job than a human but can do it exponentially faster in many cases. Thus, the application would be for instances where a human producing the content is time or cost prohibitive.
If the content needs to be better than other content or in any way can be compared to content available written by a human, it will be much worse, without question.
Some examples:
Without the human touch, some types of content are just plain bad and Google has a massive index that is expensive to manage and maintain. They can do without most AI content and it’s why they’ve discouraged it every chance they can publicly.
Imagine a robot trying to explain how to get through a breakup, manage anxiety, or have “the talk” with your teenager.
A robot cannot experience anxiety, have its heart broken, or understand the gravity of human reproduction.
Now, here’s my favorite example because I’ll use an example that goes a little more in-depth.
You search for “how to stop crying baby”.
AI can write an article about “why babies cry” which is a bulleted list that could easily be a rich result.
But, HOW to stop a crying baby is content best provided by someone with expertise and AI cannot provide this without getting it from a human.
Google would certainly not want to rank what an AI would have to get from a medical journal or another piece of content written by a human.
Personally, I don’t have any kids. I have very little experience caring for babies that are crying, but my mother sure does. I’m sure I cried a lot, thus, if we were both writing articles on the topic, Google would want to understand the difference between my fake expertise and her actual expertise.
Google can understand the confluence of words, topics, and information — just look at how they identify adjacent information in Google Trends.
The quality of the content produced about “how to stop a crying baby” is directly determined by the source. In this case, the rankings would go…
1.) my mom
2.) me
3.) AI
My mom went to school to be a journalist as well, so there’s an exponentially higher chance of her blowing away my content in a distinguishable way; however, I stand an even better chance against the AI. I could probably feign experience that an AI would lake the human understanding needed to duplicate.
Distinguishing the expertise in my example above is a far cry from what people think of as SEO today. AI content doesn’t need to be identified. It simply can’t produce this content and will stick out like a sore thumb. Its not good for most queries.
If AI can produce a short answer that is essentially a commodity, Google will use their robots.
Not every piece of content needs to be based on deep human insights or a full range of emotions. News articles, financial reports, weather updates, and promotional social media reminders can all be easily written by content AI without much drawback or decrease in quality.
Mad libs-type articles on stock ticker movements are a great example. The info is available, rapidly changing, and pretty much out-of-date within 24 hours. It is cost and time prohibitive to pay a person(s) to make it on a daily basis.
This is a great application of AI, and this Benzinga above was the number one news article at the top of search results when I wrote this for the topic, “amazon stock price today.”
In short, we’re talking about commodity information: financial news, sports reports, weather updates, data analysis, and the like — stuff readers don’t mind hearing from a computer. The truth is, you don’t need a human to create every piece of content; AI can sort and analyze data infinitely faster than humans and can produce text summarizations that are actually useful for readers.
While video may be harder and more costly for humans to produce, it’s actually much easier and a better application for AI.
In fact, the distinguishing feature of AI content that we discussed above that allows Google to easily parse the AI from human content (quality) doesn’t exist the same way with video.
WHY?
AI is producing video the same way a human would and the resources they’d use are equally available and in most applications for the web don’t require more than that.
This is why we made Flickify to turn articles into videos.
Humans have already produced the human part of the content that matters. Assembling stock footage, editing for different mediums, and transposing captions and narrations can actually be done far faster and better by machines than an amateur video producer.
AI will not be making Blockbuster movies or generating unique video where there are few resources to draw from any time soon — or maybe ever. But, the act of taking existing content and assembling it in video form is a near-perfect application of what I shared AI was best at in my opening paragraph.
“AI can do a slightly worse job than a human but can do it exponentially faster in many cases.”
With video, if the production is done by an amateur, AI can probably apply unknown best practices that even make the content better when produced by a machine. Therefore, AI is faster and better than a human at turning written content into video for most real-world applications where video is needed today on the web.
For that reason, there’s no signature or means of determining if AI made a video vs. a human made it in those cases. Google’s primary concern is mitigated and visitors win out either way.
Seems counterintuitive, but if you work in this space day in and day out, it should make sense.