As Generative AI Spreads, Does the Value of "Human-like Writing" Increase? Research Reveals an Unexpected Paradox

Generative AI is dramatically transforming the task of "writing" in business.

Sales emails, proposals, press releases, recruitment pages, crowdfunding descriptions, social media posts, product introductions, and executive messages. Sentences that used to be carefully crafted by copywriters, PR personnel, or employees skilled in writing can now be generated by AI in mere seconds.

Even those lacking confidence in their writing can input key points and receive text that is reasonably readable, logical, and with few typos.

At first glance, this seems like a welcome change for companies.

However, as more companies adopt generative AI, another issue may arise.

"As everyone's writing improves, no one's writing stands out."

This paradox was highlighted in research published by researchers from the University of Manchester in the UK and the University of Alberta in Canada.

The researchers are not concerned with whether AI can write good sentences.

Rather, it's the opposite.

Because AI can write sufficiently good sentences for many people, communication between companies may become similar, making it harder than ever to convey "differences."

"Democratization of writing skills" is occurring

One of the significant features of generative AI is that it greatly lowers the technical barriers previously required for writing.

Even executives who are not good at writing can instruct AI to create persuasive text for investors, and AI will produce plausible content.

If you list the features of a product, it can be turned into a product description for an e-commerce site.

Input the company's philosophy, and it can be transformed into a message for a recruitment page.

For small businesses or sole proprietors who cannot afford to hire writing experts, this is a significant advantage.

There have always been individual differences in writing skills.

Even if selling the same product, one company's description might be clear while another's is not.

One entrepreneur might be able to tell their vision as a compelling story, while another, despite having great ideas, cannot articulate them well.

Generative AI narrows this gap.

Researchers are examining the changes that occur beyond the "leveling of the playing field" commonly associated with AI.

The issue arises when all competitors start using similar "writing assistants."

What happened when 26,724 texts were enhanced by AI

In the study, to illustrate this concept, 26,724 texts used in actual crowdfunding campaigns were analyzed.

For each, an AI-enhanced version was created and compared with the original human-written texts.

The results showed more than just simple improvements in the text by AI.

The **"variation" among the texts themselves decreased.**

Particularly in terms of how abstract the texts were, the variation in AI-enhanced versions was 43% smaller than in the original texts.

Furthermore, the variation in text length and quality was reduced by about 80%.

In other words, AI doesn't just bring those who struggle closer to the average.

It potentially gathers the entire body of text, including those who excel, into a "certain range."

This reveals a very interesting property of generative AI.

AI generally aims to create "readable text," "logical structure," "appropriate explanations," and "inoffensive expressions."

As a result, each text may seem improved when viewed individually.

However, when viewed in the tens of thousands, they collectively converge to similar points.

The very criteria for "good writing" may change

So, what is the problem with AI making corporate writing generally better?

The important thing is that the standards of the readers also change.

In the past, simply being able to produce readable and organized text was a strength for a company.

However, with AI, anyone can create text above a certain standard.

Thus, features like "readability," "no typos," and "well-structured" become minimum requirements rather than strengths.

This is somewhat similar to smartphone cameras.

Before high-performance cameras became widespread, the ability to take beautiful photos was valuable in itself.

But once everyone can take high-quality photos, that alone doesn't capture attention.

Composition, theme, moment, people, and story become more important than the photo itself.

The same could happen with writing.

"Just writing well" is not enough.

What is being said?

Who is saying it?

Why does the company need to say it?

Is there real experience or evidence behind it?

The value of these aspects will relatively increase.

Trust in "plausible words" may decline

What generative AI excels at is creating plausible text.

"We transform the industry with innovative solutions."

"We create new value with a customer-centric approach."

"We contribute to realizing a sustainable future."

These sentences are grammatically correct.

They wouldn't be out of place on a corporate website.

However, the more generative AI becomes common, the more readers will become accustomed to such "well-structured but vague words."

Eventually, the attitude of "just saying it isn't enough to trust it" may strengthen.

Researchers also point out that companies may increasingly need to provide evidence to support their claims.

Instead of writing "We offer the best service," show customer examples.

Instead of saying "We are supported by many," show numbers.

Instead of declaring "We will change the industry," explain what has actually been changed.

In a society where words can be mass-produced by AI, ironically, the value of "things beyond words" increases.

The feeling that "AI texts all look the same" is spreading on social media

What makes this research intriguing is that the discussion in the paper aligns with the sentiments occurring on social media.

 

In marketing and writing-related communities, it has long been noted that

"You can immediately tell if a text is written by AI."

"Every company has the same tone."

"It's well-structured, but nothing stays with you after reading."

Such comments are repeatedly posted.

In a May 2026 Reddit post related to Digital Marketing, while acknowledging AI as a powerful writing tool, the issue of "posts that are clearly AI-written feel unnatural and lack human touch" was raised.

Among the comments, there were consumer reactions stating they stop reading once they sense AI-like qualities in ads or social media posts.

Of course, Reddit posts are not scientific public opinion surveys.

They are merely the opinions of some users.

Nevertheless, it is evident that when companies use AI, issues arise not only in "text quality" but also in "how it is perceived."

Even more striking is the experience shared by a marketing professional in a June 2026 post.

In their company, copywriters were reduced, and AI replaced text creation.

However, in reality, the marketing staff themselves had to add specificity and human touch to the "competent but lifeless text" generated by AI every week.

Rather than eliminating the work of writers entirely, the task of "making AI-generated text more human-like" shifted to another person.

This post does not deny AI itself but evaluates it as "excellent as a machine for creating initial drafts."

The problem is considering that draft as the finished product.

On the other hand, there are opinions that "AI can still create brand individuality"

The discussions on social media are not entirely critical of AI.

In a June 2026 Reddit discussion on brand strategy, there was an opinion that "AI does not destroy brands but amplifies the clarity of the strategies that brands originally have."

In other words, if a company has clear standards for

what values it holds,

what words it uses,

what words it avoids,

who it speaks to,

and how it differs from competitors,

it is possible to maintain individuality even when using AI.

Conversely, if the brand itself does not understand "who they are," AI will fill in the blanks with average marketing expressions.

As a result, the text will resemble something seen elsewhere.

The same discussion introduced the idea of not treating AI as a comprehensive "author" but using it for brainstorming, expanding from long to short texts, organizing text, and considering counterarguments, while humans make decisions on brand positioning and key claims.

This aligns with the researchers' assertions.

The important thing is not "whether to use AI or not."

It is the ability to judge where to follow AI and where to diverge from it.

The paradox that the more AI is used, the more human skills become important

When generative AI first emerged, much was said about the "democratization of specialized skills."

Even without the ability to write, AI can compensate.

Even without design experience, images can be created.

Even without programming skills, apps can be developed.

Indeed, this is happening in real-time.

However, this study looks beyond that.

In a world where anyone can produce work of a certain standard, the value of "the ability to produce work of a certain standard" itself decreases.

Instead, the value of the ability to judge what should be created, understand differences from others, read culture and context, and determine one's position increases.

Researchers particularly emphasize cultural understanding and storytelling ability.

For example, even with the same product, the words that resonate differ when selling to students versus executives.

Expressions accepted in Japan may not be received the same way in another country.

In one industry, the word "innovative" might be attractive, while in another, "track record," "safety," or "reliability" might be more important.

Generative AI knows a vast array of text patterns.

However, the judgment of "what should be conveyed to these people at this moment" still requires human understanding.

It's not that "companies not using AI" will win

It should be noted that this research does not argue that "generative AI should not be used."

The researchers themselves do not suggest avoiding the use of AI.

Rather, they emphasize the importance of effectively using AI while intentionally diverging from AI's suggestions when necessary.

This paper does not predict the future with certainty.

The researchers present "possible scenarios" based on existing research, AI usage trends, and comparisons of crowdfunding texts.

AI itself is rapidly evolving.

In the future, if AI can learn and reference company-specific texts, past statements by executives, brand guidelines, and customer data, it may be able to generate far more unique texts than it currently can.

In that case, the issue of "AI = homogeneous text" might be somewhat alleviated.

However, if all companies start using advanced brand AI, the competition standards will rise in another form.

Thus, the essence of the problem is not AI itself.

It is that AI continuously raises the standard of "ordinary."

From "writing well" to "writing what only you can write"

Before generative AI, writing skills often meant "the ability to write clearly."

However, in the era of generative AI, this definition may change.

Organizing text.

Paraphrasing.

Creating headlines.

Shortening text.

Correcting grammar.

These are already