Can AI Perform SEO? Experimenting With OpenAI’s GPT-3

Here, we can observe how the AI presents each question. The answer it selects is indicated by the highlighted green text.

Below, we can observe how the AI is presented with each question and its corresponding answer in the highlighted green text.

BrightEdge provides a table that displays the questions along with their correct answers in one column, and the answer chosen by the AI in the adjacent column:

The results from GPT-3 are encouraging, as the AI achieves a score of 16/20. Even so, it lacks the context to understand where it has gone wrong.

Now, let us inquire about recent updates from Google concerning the launch of Google MUM on May 18, 2021.

What is the primary purpose of Google MUM?

Once again, as you can see, the AI provides a perfect response.

Overall, we have a reliable SEO response engine when it comes to asking very specific questions.

The Limitations

Now, let us consider the limitations of AI-driven SEO.

Here is a non-exhaustive list:

The AI's ability to resolve unknown or poorly documented issues.

In the SEO industry, it is often time-consuming and cumbersome to comprehend the exact nature of problems when they arise, and sometimes no explanation is available. AI cannot find solutions with data that is not part of its analysis.

The capacity to encompass all existing factors.

There are typically numerous restrictions on what has already been implemented or tested on a website or a set of websites. Consequently, it is presently impossible to provide an AI system with all the source codes and project documentation required to deliver an accurate answer that takes all existing information into account.

The ability to assess the true quality of content or an idea based on its originality and its representation of the current situational context.

AI systems presently lack real-time functionality and suffer from a lack of originality and creativity. As a result, the assessment of content quality is currently quite limited. The evaluation of quality content is often performed by human beings. AI systems are primarily effective at identifying spam and poor-quality content.

To summarize, while an AI system may surpass an SEO novice or someone with limited SEO knowledge, it still has a long way to go before it can replace an experienced SEO expert.

In conclusion, the combination of AI and human expertise will always remain the most effective approach, with humans focusing on tasks that provide the highest value-added.

Ultimately, seasoned SEO experts far surpass AI-powered SEO.

Also read: Google States That AI-Generated Content Violates Guidelines