Generative AI for Content: Here’s what SEOs Need to Know

28 Mar, 2023

Authors
Ajay Rama, SVP Products, iQuanti
Sreekant Lanka, SVP Solutions, iQuanti

ChatGPT has caught everyone’s imagination. It promises enormous potential to transform many types of work, yet challenges remain. Let’s understand the technology as well as the opportunities and challenges it presents to SEO marketers.

Download a copy of this thought paper on Generative AI for SEO marketers here.

What is Generative AI?

Generative AI algorithms generate new output after being trained on large amounts of existing data. Traditional AI systems recognized patterns and made predictions, but Generative AI creates new content, including text, images, and audio.

How does Generative AI work?

Generative AI uses generative adversarial networks (GANs) to create new content. A GAN has a generator that creates new data and a discriminator that evaluates whether the data is real or fake. The generator and discriminator train together to create content that is indistinguishable from existing data.

What sets it apart from previous versions of AI?

 

 

Quality of results: In straightforward cases, AI like ChatGPT can create high-quality content that can easily pass as written by a human.

 

Ease of use: Because AI can make sense of unstructured input, little to no learning is required to give it useful instructions.

 

Versatility: Generative AI models can be used for a wide variety of tasks, including music, arts and coding.

Evolution of Generative AI Models

Generative AI has evolved over the last few years.

Use cases of Generative AI for marketers

 

iQuanti’s proprietary AI-driven enterprise SEO platform ALPS uses GPT technology to organize, automate, and augment SEO functions. Here are a few ways ALPS uses Generative AI:

  1. Analyzing unstructured data
  2. Generative AI is good at analyzing unstructured data by flexibly categorizing it and providing structure as prompted. This has far- reaching implications for any field that needs insights from real-world unstructured data. For example, in SEO, ALPS is using Generative AI to find relevant keywords and categorize them into themes.

  3. Topic research
  4. Extensive topic research requires diligently evaluating large amounts of data to exhaustively identify all the sub-topics. Usually, the researcher searches for top-ranking pages, as exhaustive topic research is time-consuming. Even with ALPS, we used to limit research to the top 10 competitors. With Generative AI, ALPS does comprehensive topic research covering all possible sub-topics across all competition.

  5. Insights on tabular data
  6. Generative AI can do a good job of using structured data input (i.e tabular data) and providing data-based insights and recommendations. We are using the structured data dump from tools like Google Ads and ALPS to generate insights and get recommendations. These are of course vetted by experienced subject matter experts. We have found these insights to be accurate. Download a copy of this thought paper on Generative AI for SEO marketers here.

Download a copy of this thought paper on Generative AI for SEO marketers here.

 

Generative AI in content

Research

Content research can be very time-consuming, requiring writers to research a variety of sub-topics, and to understand different takes on the topic. Generative AI streamlines this process. ALPS uses Generative AI to make high-quality content research easier; you get all the research related to the topic right within the platform.

Copywriting and editing

Generative AI can be a useful aid in speeding up the process of creating several content forms from a single source. Titles, meta descriptions, social media posts, summary posts, and creating multiple variations are some of the types of content that AI can produce quickly.

AI can even check whether content adheres to a brand voice, and it can rewrite the content to follow brand guidelines.

Content creation

Generative AI usually creates content that is descriptive, data-based, and factual. AI can reduce the effort needed to ensure that long-form content follows the flow and structure you want. With ALPS, we are using Generative AI to write smaller subsections of long-form content. This gives the user better control of the structure and flow of the content.

There are areas where we find it diffiult to achieve the expected quality of output with Generative AI. One example is content that requires a unique point of view, creative thinking, creative expressions, using analogies, and indirect meaning.

Sometimes the output from AI may be a bit verbose and in some cases AI produces the wrong answer to a question. For example, when asked to evaluate the accuracy of an article about credit scores, ChatGPT gave a wrong answer, then explained its error with an additional error, then acknowledged the right answer:

ChatGPT: The following statement is incorrect:

“‘Fair’ credit score ranges indicate to credit issuers that a credit user may hold high amounts of debt.”

Fair credit score ranges, typically ranging from 580 to 669, indicate to credit issuers that a credit user may have made a payment that was 30 days late or more, carry a high debt load, and may not be eligible for the best interest rates. However, it does not necessarily mean that they hold high amounts of debt.

Question: You said that “may carry a high debt load” is different than “may hold high amounts of debt.” Don’t they mean the same thing?

ChatGPT: I apologize for any confusion. “May carry a high debt load” and “may hold high amounts of debt” can be considered similar in meaning, and both suggest the possibility of a person or organization having a significant amount of debt.

The difference between the two phrases is that “may carry a high debt load” suggests that the debt is actively being managed or paid off, whereas “may hold high amounts of debt” suggests that the debt exists but does not necessarily imply any action is being taken to address it. However, the difference between the two phrases may be minor and they can often be used interchangeably.

Fair credit score ranges, typically ranging from 580 to 669, indicate to credit issuers that a credit user may have made a payment that was 30 days late or more, carry a high debt load, and may not be eligible for the best interest rates. However, it does not necessarily mean that they hold high amounts of debt.

Question: So you’re wrong, because your credit score doesn’t reflect whether debt is being actively managed or paid off. It’s more correct to say you may hold high amounts of debt.

ChatGPT: You are correct, I apologize for any confusion I may have caused. You are right that having a high credit score doesn’t necessarily mean that someone is actively managing or paying off their debt. It is more accurate to say that someone “may hold high amounts of debt” rather than “may carry a high debt load.” Thank you for pointing that out.

ChatGPT on how Generative AI can help marketers:


How can an AI language model like ChatGPT assist marketers?

An AI language model like ChatGPT can assist marketers in several ways:

  1. Content creation: ChatGPT can generate high-quality and engaging content quickly and efficiently. It can assist marketers in creating a variety of content types, such as blog posts, social media updates, and product descriptions.
  2. Idea generation: ChatGPT can provide marketers with new ideas for content based on specific keywords or topics. It can help marketers brainstorm new and creative ways to approach their content and explore new content themes.
  3. Market Research: ChatGPT can analyze large amounts of data and provide insights into consumer behavior, trends, and preferences. Marketers can use this information to create targeted campaigns and improve their overall marketing strategy.
  4. Personalization: ChatGPT can be trained on specific customer profiles, allowing marketers to create personalized content for their target audience. This can help marketers create content that is more relevant and engaging to their customers.
  5. Automation: ChatGPT can automate various marketing tasks such as content creation, email marketing, and social media management. This can save marketers time and resources and allow them to focus on more strategic tasks.
  6. Customer service: ChatGPT can be used to develop chatbots and other conversational interfaces that can assist with customer service inquiries. This can provide customers with immediate assistance and reduce the burden on customer service teams.

Overall, an AI language model like ChatGPT can help marketers create more effective content, improve customer engagement, and streamline various marketing processes, ultimately leading to better business outcomes.

 

Download a copy of this thought paper on Generative AI for SEO marketers here.

Challenges with using ChatGPT for content creation

Tools like ChatGPT should be viewed as one of many starting points in the content creation process. They can be useful for steps such as topic ideation and brief creation, but they’re not ready for full automation yet.

  1. Google cautions against complete reliance on AI
  2. Raw AI-generated content is prohibited by Google Search Essentials (formerly Webmaster Guidelines).

  3. Google knows when you use AI
  4. Google can detect automatically generated content. ChatGPT also plans to add watermarking.

  5. AI can’t understand nuance
  6. AI language models are not capable of understanding the context and nuances of a topic the way a human writer can, so the content they produce may not be of high quality.

  7. AI can’t fact-check
  8. While language model chatbots can be remarkably accurate at times, they are not designed to ensure factual accuracy. All factual statements need to be checked by humans.

  9. AI can’t do SEO for content
  10. AI language models like ChatGPT are not trained for SEO purposes, so the content they generate will not be optimized for search.

  11. AI may be biased or cause harm
  12. ChatGPT’s terms say that it “may occasionally generate incorrect information” and it “may occasionally produce harmful instructions or biased content.”

  13. ChatGPT can’t write timely topics
  14. According to OpenAi, ChatGPT was trained on a data set from 2021 and earlier, and it is not connected to the internet, so it has “limited knowledge of world and events after 2021.”

  15. Privacy concerns
  16. It may not be a good idea to submit confidential information to ChatGPT. Its privacy policy says:

    Q: I want to use ChatGPT Plus with sensitive data. Who can view my conversations?

    A: As part of our commitment to safe and responsible AI, we may review conversations to improve our systems and to ensure the content complies with our policies and safety requirements. For more information on how we handle data, please see our Privacy Policy and Terms of Use.

  17. Content that isn’t unique
  18. Q: ChatGPT’s terms caution against assuming output is uniQ:ue from one use to another:

    A: “Similarity of Content. Due to the nature of machine learning, Output may not be unique across users and the Services may generate the same or similar output for OpenAI or a third party. For example, you may provide input to a model such as “What color is the sky?” and receive output such as “The sky is blue.” Other users may also ask similar questions and receive the same response. Responses that are requested by and generated for other users are not considered your Content.”

  19. ChatGPT doesn’t take responsibility for its content
  20. Q: ChatGPT’s terms prohibit holding them responsible for any legal issues arising from the use of their service:

    A: Indemnity. You will defend, indemnify, and hold harmless us, our affiliates, and our personnel, from and against any claims, losses, and expenses (including attorneys’ fees) arising from or relating to your use of the Services, including your Content, products or services you develop or offer in connection with the Services, and your breach of these Terms or violation of applicable law.

What’s next for AI?

AI is changing quickly, and we’ll need to be nimble but cautious.

    • OpenAI released GPT-4, a large multimodal model that accepts image and text prompts, in March 2023. According to OpenAI, GPT-4 “while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.”
    • Microsoft is an investor in OpenAI and has integrated GPT-4 in Bing Chat. Some GPT-4 features are missing from Bing Chat, however, such as visual input.
    • In March 2023, Microsoft also announced AI-powered Copilot for its 365 suite apps including Word, Excel, PowerPoint, Outlook and Teams.
    • Google has announced its own slew of AI solutions, and the race is on. We may soon see disruptive changes to SERP experience.

Generative AI is transforming the landscape of content creation, and it will likely continue to improve. It’s important to be aware of its limitations, especially when trying to use AI without human oversight. There may come a time when oversight isn’t needed, but it isn’t here yet.

Download a copy of this thought paper on Generative AI for SEO marketers here.

If you have questions/comments, please reach out to us at marketing@iquanti.com.

Share This
×

Attention: iQuanti does not conduct transactions or recruit via chat platforms like Telegram or WhatsApp. We never request sensitive information or payments for employment. If contacted by someone claiming to be from iQuanti and seeking payment/info, report it to hr@iquanti.com.