How Does Google View AI-Generated Content?

Anybody who has used ChatGPT can tell you what a godsend it is for anything. With some prompts and a literal press of a button, it will give you any information that falls within its knowledge cutoff. Unsurprisingly, others have used it to make whole copies, articles, and blog posts. But is this the best way to utilize AI for content creation? And how can one use AI when the primary focus is human-written content? Let’s talk about it! But first, we need to know what Google says about AI Content.

How Does Google View AI-Generated Content?

In a blog post published on February 8, 2023, Google stated this illuminating passage:

“Our focus on the quality of content, rather than how content is produced, is a useful guide that has helped us deliver reliable, high-quality results to users for years…. Using AI doesn’t give content any special gains. It’s just content. If it is useful, helpful, original, and satisfies aspects of E-E-A-T, it might do well in Search. If it doesn’t, it might not.”

So, the answer is very cut and dry. Google doesn’t care where the content comes from. It only cares about quality. But then, quality is a subjective concept. To better understand how Google rates content, we must look into E-E-A-T.

Experience, Expertise, Authoritativeness, and Trustworthiness is the company’s evaluation guidelines for rating websites and their content. When discussing the subject matter, Google wants content to come from someone with first-hand experience, proven proficiency, and a verified reputation. There are a few ways to ensure your content hits all requirements, but today, let’s focus more on what AI-generated content brings.

A Quick Test

For comparison, let’s take a subject like earning monthly income with options trading. Then, let’s see a human writer introduce the topic from the perspective of an experienced trader and compare and contrast it with an AI-generated piece.

Human:

Many people look at options trading and think, “This is too complicated for me.” And while it is true to some extent, you don’t have to be a subject-matter expert to start earning money using the derivative. All you need to have is a basic understanding of how options work and the factors that affect pricing, like time and volatility. I’ve been trading options for 20 years now, and it’s fair to say that I’ve learned quite a lot. Today, I’m happy to share several strategies that can potentially be steady sources of monthly income.

AI (ChatGPT 3.5 / Prompt: Write a 100-word introduction to earning monthly income with options trading as a trader with 20 years of experience):

Embarking on a journey into options trading, I bring two decades of seasoned expertise to illuminate the path of monthly income generation. In the dynamic realm of financial markets, I’ve honed strategies that transcend market fluctuations, leveraging options to create a consistent revenue stream. From navigating volatility to crafting risk-managed positions, my experience is a compass for traders seeking stability and growth. This journey isn’t just about numbers—it’s a synthesis of intuition, risk management, and adaptability. Join me in unraveling the art and science of options trading, where every month becomes an opportunity to sculpt financial success.

Objectively, the two introductions have the same content and promise the same information. However, the AI introduction is too wordy and formal. This is perfect for research papers but not a regular blog post. And it lacks a certain spark of personality that some people look for in their content.

Now, we can do something about the tone or personality by adjusting the prompt like this: Rewrite the passage with a less formal tone but still convey the writer’s experience. Here’s the result:

Hey there, diving into the world of options trading with a cool two decades of street smarts under my belt! I’ve been riding the ups and downs, crafting strategies that dance through the market craziness. Options are my secret sauce for cooking up a steady monthly income. From dodging market rollercoasters to cooking up positions that play it safe, I’ve seen it all. This journey is more than just numbers; it’s a mix of gut feelings, smart risks, and rolling with the punches. Let’s roll together and unwrap the tricks of the trade, turning every month into a chance for financial wins!

So… yeah. Now, I’m not saying that Google will rank the AI content lower than the human content—there’s no way for me to be sure short of asking them. However, the company has been very clear with the direction for ranking content: they want quality content that is unique and useful for humans. So, if people find AI-generated content dull and repetitive, then chances are Google might think so, too. In the end, you need to test whatever works for you.

How To Use Google For Human-Written Content

Purely AI-produced content might not work for most people. But not using the tool is such a waste. Large-language model (LLM) AI like ChatGPT are still very useful — we just need to know when and how to use them to our advantage. So here are three ways you can utilize AI while still producing human-written content.

Generate Content Ideas

The first step of writing content is taking an idea and running with it. However, people are not endless founts of concepts; writer’s block can happen even at this early stage. AI can assist struggling writers by giving them many writing ideas and prompts. Take a broad topic, type it up on ChatGPT, and the AI can give you any number of titles for content ideas. You can also drill down on specific topics and fine-tune your searches based on your experience and preferences.

Do Quick Research And Calculations

Before the internet, our grandparents had to go to the local or university library to do research. By that standard, Googling something is great leap forward indeed. Then ChatGPT came along and made it even easier.

Want to know how fast light can travel in an oxygen-rich environment? Roughly 299,108.76km/s, or 99.7% of the speed of light. Who is the worst Roman emperor? Opinions vary, but a lot of people say Nero.

There is a caveat, however. ChatGPT has a hard knowledge cutoff — September 2021, to be exact — which limits the available information it can produce. That means asking about something that happened beyond that date will likely yield this answer:

This is only using the free 3.5 version, though. OpenAI is rolling up its data inclusion with ChatGPT Enterprise, bringing the cutoff date for the paid service up to January 2022. It’s not much but more of a proof of concept. Still, some research is better done in real time.

Editing And Proofreading

It’s easy to make mistakes when writing content; even editors and proofreaders are prone to making and missing errors. Programs like Grammarly, Hemingway, or ChatGPT can automate your proofreading process. Some of them, like Grammarly, offer different correction guidelines to ensure your content is on-brand.

According to the company, “Grammarly’s AI system combines machine learning with a variety of natural language processing approaches.” With GrammarlyGO, users can also generate additional content.

Issues With Using AI For Research And Content Creation

Despite these advantages, AI is not foolproof. Content creators need to be aware of its issues to be prepared to handle them. Here are a couple of the more notable ones:

Referencing And Incomplete Information

Some topics might be best researched using traditional sources, especially when dealing with government information. For example, California has a smog check requirement for fleet operators within their jurisdiction. A quick prompt on ChatGPT provides these details:

Now, this is still usable for content creation. However, you cannot reference this in any meaningful way. Additionally, the information displayed here, despite being long, might be incomplete. Suppose we press ChatGPT for more information, like which counties require the smog test for vehicle registration and renewal. In that case, the AI gives us this answer:

This reads like a non-commital response from someone without complete information. But a quick Google Search for guides to California’s smog regulations gives us a more complete picture, like so:

Source: Samsara.com

In this case, AI is a great starting point for research—but it’s not a one-stop shop.

Do Androids “Hallucinate” About Electric Sheep?

Another thing that AI users need to be aware of is “hallucination.” This refers to instances where an LLM generates factually incorrect responses or gives out nonsensical information that diverges significantly from reality.

These hallucinations can occur due to the inherent limitations and biases in the training data and the complexity of language understanding. Researchers using ChatGPT to identify and produce references for scientific research noted that the LLM-based AI hallucinated 20-25% of the time, giving out incorrect or unverifiable information. This is why any information AI provides must be verified using legitimate and accurate sources.

Use AI To Supplement Your Content

Despite Google’s assurances that AI-generated content will not be rated lower than human content on that factor alone, we still have to consider that people have a better grasp of what other people like with their content. And besides, Google may yet change its guidelines. AI is a great tool, sure, but being overly dependent on this relatively new technology might not be the best option for content creators. Like with everything else, moderation is key.