How to Humanize AI-Generated Content With Surveys
By Owen Fay . Posted on January 3, 2023

With everyone and their mother using tools like chatGPT and Jasper to create AI-generated content, the million-dollar question becomes: If everyone has chatGPT, how is your article different from others?

We’ve seen this story before. About 10 years ago, there was a craze around content farms, and everyone was creating sites with thousands of auto-generated content. Then Google came out with the Panda release and cracked down on auto-generated content. It is almost certain that Google will soon crack down on AI-generated content.

“Even if AI can write thousands of words, the message must still be humanized to make a lasting impact on its readers.” – Neil Patel

How do you humanize your AI content with exclusive data? Here’s what you need to know.

What Is The Problem With AI-Generated Content?

As one can imagine, the AI has been trained on a corpus of knowledge that is already existing. While this corpus of knowledge might be humongous, it is basically rewriting knowledge that already exists.

So from Google’s perspective, it raises an important question: How is your article different from others, and what value is it adding to users?

Suppose the AI is basically just saying the same things in a different form (no different than translating something in another language), what value is your article adding in Google’s eyes?

This fundamental question also raises the primary concern that Google will soon crack down on fully AI-generated content that is not humanized or that does not add unique content or perspective to the topic.

Google’s position on this has been very clear (as early as April 2022)

For us these would, essentially, still fall into the category of automatically generated content which is something we’ve had in the Webmaster Guidelines since almost the beginning.

And people have been automatically generating content in lots of different ways. And for us, if you’re using machine learning tools to generate your content, it’s essentially the same as if you’re just shuffling words around, or looking up synonyms, or doing the translation tricks that people used to do. Those kind of things.

My suspicion is maybe the quality of content is a little bit better than the really old school tools, but for us it’s still automatically generated content, and that means for us it’s still against the Webmaster Guidelines. So we would consider that to be spam.

Why Is it Important to Add Exclusive Content To AI-Generated Content?

By adding exclusive content to your AI-Generated content, you answer the question: How is your article different from others?

It solves the uniqueness and value-add factors that are important to Google. For SEOs familiar with the EAT (Expertise, Authoritativeness, and Trustworthiness) principles, exclusive data is exactly the sort of content needed to increase your EAT scores.

Let’s take a look at how adding exclusive content can benefit your EAT scores:

Expertise

Rather than simply regurgitating what others are saying, having exclusive content (like market research surveys) increases the level of expertise you bring to the topic. Google loves this. Adding thought leadership based on hard facts and data, especially qualitative statistically significant surveys, is the type of “expertise” that makes for a good article.

Authoritativeness

There is a big difference between saying, “I said so,” and saying, “I know so based on deep market research I have conducted.” When your exclusive content is based on clear research studies and market research data, it adds an element of authoritativeness to your article (even if you yourself are not an authority!)

Trustworthiness

There is one *huge* problem with AI-Generated content (at least as of Jan 2023!), and that is: AI has a truth problem.

“It’s a mistake to be relying on it for anything important right now. it’s a preview of progress; we have lots of work to do on robustness and truthfulness.” – Sam Altman, CEO, OpenAI

In simple terms: AI sometimes blatantly makes up stuff. This is very much true of tools like chatGPT as acknowledged by its CEO.

So if your article is supposed to be trustworthy in Google’s eyes, what does it say that the AI is outright making up stuff?

6 Benefits Of Surveys To Humanize AI-Generated Content

One of the best ways to humanize your content is to add human insights from surveys into your article.

Let’s take a look at what you get from conducting a market research survey on your topic.

1. Exclusive content from survey responses.

When you conduct a market research survey, you get hundreds (or thousands) of human responses *in their own words*. This is unique content and perspective filled with emotions, powerful words, and imagery.

Exclusive content from Survey Responses.
Exclusive content from Survey Responses.

You can insert unique quotes into your content by carefully selecting good responses. These quotes are not the same old information, just said in different ways. They represent the judgment and perspective of real humans – something that, hopefully, Google will appreciate in its EAT scores.

2. Statistically-significant data

When you conduct head-to-head surveys with a statistically significant number of respondents, you can make conclusions based on the results. This gives you an opportunity to make authoritative statements like “A is better than B” and have the data to back up your statements.

Statistically-Significant Research StudyStatistically-Significant Research Study

For example, You will see below how a mommy gardening blogger was able to select a favorite plant based on a head-to-head contest between two plants.

3. Wisdom of the crowds

Sometimes bloggers and writers tend to fall into the trap of “individual bias”. They write the article based on their own biases and the topic.

Wisdom of the Crowds - Poll the People Market Research PanelWisdom of the Crowds – Poll the People Market Research Panel

But by tapping into a large pool from a panel, you can tap into the “wisdom of the crowds”. This gives you the opportunity to base your article on the collective wisdom and, hopefully, with some unique perspectives.

4. Authoritativeness based on market research

We often make an authoritative statement (e.g. “A is better than B”) without having the data to back it up. When you run a market research survey, you can validate such statements and provide proof to back them up.

This is exactly the type of “Authoritativeness” that Google is looking for in its EAT scores. A statement backed by a research study and data to prove it.

5. Niche-Specific Expertise

By targeting the survey to only a specific niche of people (e.g. “Females between Age 18-25”) or readers of your blog, you can get niche-specific expertise.

If you can do this, you can gain deep insights from people who know your niche best. This could be readers of your blog, subscribers to your newsletter, or even targeted respondents from a panel.

When you micro-target like this, you can a wealth of information that is unique and not available anywhere else.

6. Trustworthiness via Human Intelligence

When you conduct a market research study, you are getting the perspective of hundreds (or thousands) of real humans with feelings and empathy. This is something that AI cannot do (and never will!). AI is great at understanding variables and factors – but is absolutely terrible at human factors like feelings, emotions, judgment, and empathy.

Trustworthiness via Human IntelligenceTrustworthiness via Human Intelligence

You can uniquely gather these emotions from a panel of human survey takers. When survey takers explain their choice, they pour their heart into the explanation, something we’ve uniquely seen when analyzing responses at Poll the People.

Case Study

An excellent example of using survey data to create exclusive content comes from Michelle Wilde from Indoor Plants Channel.

In her article, “16 Low Light Plants That Will Transform Your Dark Spaces“, Michelle uses the Poll the People platform to run a head-to-head survey between two of her favorite low-light plants.

Case Study: How to Humanize AI-Generated Content With SurveysCase Study: How to Humanize AI-Generated Content With Surveys

Let’s look at what survey data Michelle used in her article:

1. Survey Responses

Michelle took a select number of response essays from the results and included them as quotes in the article. This gives her article unique content in the words of the actual human panel of respondents.

Survey Responses for Content MarketingSurvey Responses for Content Marketing

In the words of Neil Patel, it “humanized” her article beyond just a list of plants.

2. Pie Charts and Survey Results

Michelle was able to provide an executive summary and pie chart about the head-to-head contest between the two plants.

Pie Charts and Survey Results To Improve Authoritative EAT ScoresPie Charts and Survey Results To Improve Authoritative EAT Scores

This authoritative statement based on actual data increases the authoritative scores and adds an element of trust to her article.

3. Derived Data like Wordcloud

Michelle was even able to include derived insights like Wordclouds, demographic data, and other insights she could glean from the survey results. This results in thought leadership about her topic backed by actual survey responses.

Wordcloud from Survey ResponsesWordcloud from Survey Responses

Also, notice how this adds an element of EAT to the article. In one shot, she managed to showcase her expertise, displayed authoritative data, and captured the trust of her users with this market research survey.

Quickstart Guide

For the impatient, use this quick start guide to quickly run a market research survey and get results within an hour.

1. Duplicate

Duplicate the survey run by Michelle and edit the question and options for your niche. Michelle ran a head-to-head survey (similar to an A/B test). If you want, you can also choose an open-ended template to ask a single question.

2. Copy-Paste Responses

Notice that each survey response has an essay explanation where the respondent must clearly explain their choice. These explanations serve as excellent content where you can get into the minds and emotions of the survey takers. Copy-paste the responses you like as quotes into your content.

3. Screenshots

Include any stats, Word Cloud, and graphics you find interesting. You can include high-level screenshots of the results. Or use the word cloud and demographic data (especially if you find any interesting insights).

Step-By-Step Guide

If you want more control than the above quickstart option, here are ways you can control each step of the process.

1. Create a Survey

There are two types of surveys that you can use to create survey data for your article content.

1. Head-to-Head Contests

Use this template (recommended) to run a survey comparing two options. This is similar to A/B testing and pits two options again each other. This template is perfect to compare two options, deciding winners, and comparing against competitors (e.g. “Which insurance company do you prefer?” to compare “Geico” to “State Farm”)

2. Open-Ended Questions

Use this template to ask an open-ended question from a panel of human panelists. This template is perfect for asking panelists their opinion on a given topic. (e.g. “What factor is important to you when buying car insurance?”)

All templates have been created by experts in market research. They include features like randomization and anti-cheating that are built into the Poll the People platform.

Also, notice that every template requires the survey taker to provide a detailed explanation explaining their choice. This means that you get the reasoning behind the choices that people make. This is something that AI can never replicate (the emotions and judgments of real humans).

2. Run a Survey

Once you have selected your survey, the next step is to select your audience. For this, you have two options:

1. Panel (Recommended)

This option uses human panelists from our in-built panel of 500,000+ survey takers. Each survey taker is highly motivated and needs to say why they made the selection they did — this way, you’ll know they didn’t arbitrarily choose one title over the other. Most surveys using this option are completed in about 1 hour and cost about $1 per response. Running a survey with 200 responses for head-to-head contests is recommended for the best results.

2. Free

In this option, you collect responses from people using your own social network or readers. After creating the survey, you create a link that you can then distribute on social media or send via email to your newsletter readers.

While this option is totally free, it does take a lot longer and requires you to take the additional effort of distributing the link and getting people to take the survey (similar to how you have done with Google Forms or SurveyMonkey)

We strongly recommend using our in-built panel because it will yield quick results (in less than an hour) and also save you the trouble of gathering responses on your own.

3. Incorporate Survey Data Into Your Content

The last step is the fun part: Looking at the survey results and gathering insights and content from it.

Let’s take a look at what we can glean from the survey data that will add value to your article content.

1. Survey Responses

The first thing you will notice is the Facebook-style response feed. Each response from the survey taker is shown in the feed and has a detailed explanation essay.

Survey Responses for Humanizing AI-Generated ContentSurvey Responses for Humanizing AI-Generated Content

Read through the responses and see which explanations would be good to include as quotes in your article. If you like the one you see, mark it with the “Pin” option. In the end, you can filter all the pinned responses, which you can then copy-paste into your article as quotes.

2. Pie Charts and Survey Results

The next item you want to include is any high-level statistical insights (e.g. who won the A/B head-to-head contest).

You can do this by including an explanation of the results and the statistics like confidence, winning percentage, etc.

Pie Charts and Survey Results for To Improve EAT ScoresPie Charts and Survey Results To Improve EAT Scores

One thing to watch out for: If you didn’t include enough responses (typically 200), you might not get statistical significance. So do watch out for that.

Taking screenshots of the pie charts and detailed results is also an option to include in your article content.

3. Derivative Data like Wordcloud

One final option is to try and gain insights from any derived data. There are tabs for word clouds and demographic data. Make sure to analyze those to see if there are any interesting patterns that emerge. For example: Do younger audience vote differently from older ones?

Demographic Insights As Content For SEO ArticlesDemographic Insights As Content For SEO Articles

Conclusion

AI and tools like chatGPT can jumpstart your content creation and marketing process. But in a world where everyone has chatGPT, you still need exclusive content to differentiate your articles from the others. This is true even if you use human editors to edit/improve your chatGPT-generated content.

To ensure your AI-generated content is unique and exclusive, ensure you are augmenting it with exclusive data that nobody else has. This could be content like infographics or expert quotes, but it could also be exclusive research data that you get via surveys.

Surveys can give you unique niche-relevant content, and a head-to-head survey can also give you quantitative data that you can include in your content. This approach not only gives you exclusive content that nobody else has but also lends an element of thought leadership in your attempt to humanize your AI-generated content.

Leave a Reply

Your email address will not be published. Required fields are marked *

%d bloggers like this: