ChatGPT is an artificial intelligence chatbot that can have human-like conversations, translate, and provide users with information. 

It can also write good content. Like, really good. 

And now that Google has openly stated, “Appropriate use of AI or automation is not against our guidelines,” it begs the question: Should we use it to produce healthcare content?

Before deciding if and how to use ChatGPT for healthcare writing and SEO, there are several issues to understand.

Let’s explore:

What is ChatGPT? 

Chat Generative Pre-Training Transformer, or ChatGPT, is a natural language processing tool that allows users to mimic human conversations with an AI chatbot.

It can recognize, summarize, translate, predict, and generate large volumes of content based on what it’s learned from massive amounts of public data.

Here is a brief analogy to help visualize what ChatGPT is and how it works and learns:

Imagine a child is given an encyclopedia of the world's information. Once they’ve read it, that child can be trained to answer simple questions. As they better understand the information, they can answer more complex questions more intelligently. Over time, they continue to learn from the types of questions they’re asked and learn which information is likely the best answer.

In Chat GPT's case, their ‘encyclopedia of learning’ is a large part of the internet (data up until 2021). Their 'training' depends on understanding natural or human language, so it ‘understands’ the questions or prompts people ask. From there, it can anticipate the type of answer those questions are looking for.

In summary, ChatGPT learns on the go, refines answers as it goes, and stores knowledge for later use.

At the core of this technology are a few geeky innovations:

• NLP: Natural Language Processing (NLP) helps machines understand language by breaking apart written or spoken words into smaller components—sort of like your English teacher does with Shakespeare!

• ML: Machine Learning (ML) allows machines to train themselves in a similar way in which a human learns—by trial and error—to improve its accuracy and ability to solve problems.

Its ability to generate content quickly and easily is one of the reasons why you’ve been hearing about it so much lately—especially regarding SEO content.

How is it Different Than Other Chatbots or AI-Generated Content Tools?

It’s important to note that AI-generated content tools are not new. 

Tools like Hubspot Chatbot Builder, Salesforce Einstein, Jasper, ContentBot, and WP-Chatbot have been helping businesses automate content production, repetitive customer support, marketing, and even sales tasks for years.

But here’s what is new…

• It uses a gigantic dataset.
  ChatGPT is trained on an absurd amount of data (chatbots typically use a much narrower dataset), meaning it can carry out several more language-related tasks than other chat tools.

• It’s conversational and can answer follow-up questions.
  Unlike other chatbots, ChatGPT remembers the thread of your dialogue, using previous questions and answers to inform its next response.

• It admits mistakes.
  ChatGPT is willing to admit wrong answers when a user challenges a response. However, this requires the user to spot the incorrect answer or misinterpreted question and push back in their next query.

• It challenges incorrect premises.
  ChatGPT can admit its own mistakes and challenge incorrect user queries.

• It rejects inappropriate requests.
  ChatGPT is trained to reject queries that are racist, sexist, homophobic, transphobic, or otherwise discriminatory or hateful.

Google Says Appropriate Use of AI-Generated Content Isn’t Against Its Guidelines IF

Google’s guidance has been consistent for years regarding automatically generated content. “Using automation—including AI—to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies.”

However, they aim to “reward original, high-quality content that demonstrates qualities of what we call E-E-A-T: expertise, experience, authoritativeness, and trustworthiness.”

Meaning if businesses leverage AI-generated content to create original, high-quality, people-first content demonstrating E-E-A-T qualities, then they’re more likely to be rewarded by Google’s search algorithms with a higher page rank.

In a recent update from February 8th, Google reiterated its guidance on AI-generated content, “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.”

So, is it possible to use ChatGPT to produce high-quality healthcare content that Google will reward?

ChatGPT Limitations and Concerns for Healthcare Content Creators

Can ChatGPT be trusted to produce high-quality healthcare content? 

No (at least not yet), and here’s why:

1. Its answers and information can be wrong or incomplete.
A skilled editor should always review and confirm the accuracy of the information. 

Here is an example of inaccurate information I came across while experimenting with the tool:

Spider veins and varicose veins are not the same thing. 

2. It lacks empathy.
ChatGPT is trained to be neutral and polite. It does not write emotive content. Expressing empathy and emotion humanizes your brand and enhances the patient experience.

3. It does not understand your target audience.
AI-generated content does not inherently understand what matters to your target audience and the language that resonates with them.

4. Its knowledge is limited to 2021.
Because it is crawling data from 2021 and earlier, it could get things wrong. It is vital for your content team to verify all AI-generated responses. This ensures the information provided—and the information you share with your patients—is up-to-date and accurate.

5. It lacks experience.
Google has always had strict content guidelines for the healthcare industry. Most recently, they updated those guidelines to include a new quality consideration, experience. As a result, all healthcare content must demonstrate

a. Experience
This includes first-hand knowledge or expertise relating to the topic. Healthcare content becomes more trustworthy when created by, or reviewed by, someone with a wealth of experience, skill, and expertise.

b. Expertise
All content must be from a credible source (e.g., an established healthcare professional or service provider with the proper qualifications and credentials).

c. Authoritativeness
Your content, and the websites that link back to your content (e.g., backlinks), must prove your business is an authority in your industry or specialty.

d. Trust
Perhaps the most important of the four qualifications, your website must have signals that indicate trustworthiness (e.g., secure domain (SSL), trust seals, patient reviews, timely content, etc.)

In next week’s blog post, we dive deeper into the newest member, Experience. But the fact remains that, as of now, ChatGPT can’t provide these attributes through auto-generated content. 

6. It lacks specificity and clarity.
For ChatGPT to function properly, you must be careful, specific, and extremely clear with the information or prompts you give it, which can be time-consuming. Otherwise, you run the risk of having a garbage in, garbage out situation.

5 Ways You Can Use ChatGPT for SEO and Content Marketing

While you may not want to risk getting slapped with a Google penalty for using AI-generated content, you can leverage ChatGPT’s extensive dataset to help you write better, more informed content: 

1. Keyword research and analysis
Traditional keyword research is time-consuming. ChatGPT can fast-track this process for you. Just type in a few of your target keywords and ask it to give you a list of related ones. ChatGPT will effortlessly provide a list based on its understanding of search results. Here are a few examples of terms it returned when I asked for a list of long-tail keywords for “pediatric dentist”

a. Best pediatric dentist near me

b. Children's dental care services

c. Experienced pediatric dental specialist

d. Pediatric dentist for kids with special needs

e. Anxiety-free pediatric dental procedures

2. Content strategy development
Wild, but true. You can ask ChatGPT to outline a content marketing strategy for a specific topic, and it will return a high-level set of tactics. I couldn’t resist trying it, so I asked it to give me a content marketing strategy for healthcare marketing. It returned eight steps with instructions. Though imperfect, they would help anyone fast track the process.

3. Better understanding of search intent
Unlike Google, which takes search queries and simply locates links to the best available answer, ChatGPT tries to improve your query by providing additional information you may not have considered.

4. Title generator
You can tell ChatGPT you’re writing a blog about a specific topic and ask it to generate a title for it. I tried it and got interesting results:

a. Prompt: Create a few title options for a blog about using ChatGPT for healthcare websites
Response:
"ChatGPT for Healthcare Websites: Innovative Tool or Dangerous Gamble?"
"Navigating the Pros and Cons of Using ChatGPT for Online Medical Advice"
"Is ChatGPT the Future of Healthcare Content? Exploring the Benefits and Drawbacks"
"The Limits of Artificial Intelligence in Healthcare: Examining the Role of ChatGPT”
"ChatGPT and Healthcare Websites: Balancing Convenience with Accuracy and Safety"

5. Meta tag generator
You can ask ChatGPT to create an SEO meta tag for a web page. For example, I continued the conversation above and asked it to write a meta description for the same blog post. The result was a pretty good meta description, but it was too long for Google’s liking.

a. Prompt: Write a meta description for a blog about using ChatGPT for healthcare websites using keyword phrases: “what is ChatGPT,” and “how to use ChatGPT?”

b. Response:
Learn how to use ChatGPT, the powerful language model, for healthcare websites. Discover what ChatGPT is and how it can enhance your website's user experience. Read our comprehensive guide now.

Our Stance on AI-Generated Content: We Don’t Use It, and Here’s Why 

Your Money Your Life (YMYL) content—pages or topics that could potentially impact a person's future happiness, health, financial stability, or safety—should be written by a medical professional or someone with extensive experience in the core topics of that content.

OpenAI's model is trained on data up to 2021, meaning new medical procedures, treatments, recent innovations, or conversations in the last two years are not a part of the ChatGPT knowledge base. 

This increases the risk of inaccurate medical information.

While we see the benefits of using it as a knowledgeable digital assistant in certain circumstances, as a service provider for healthcare businesses across the country, we are responsible for delivering well-researched, fact-checked, accurate, and reliable information that is backed by medical professionals.

In December, OpenAI’s Chief Executive Sam Altman tweeted, “It’s a mistake to be relying on it for anything important right now.”

And as of right now, we couldn’t agree more.

Updated to add:
In a recent announcement, Bing (a direct competitor of Google) announced an integration of OpenAI’s GPT-4 model, which provides a ChatGPT-like experience for its search users. This model is trained on current crawl data, which means it does not have the 2021 data limitation that ChatGPT has.

Though this feature is available to a limited audience at this time, including a New York Times columnist who received some disturbing results, this reminds us that AI-generated content is becoming more mainstream.