'We're becoming statistically boring': With AI, are people losing their voice?

A quarter of workers who use AI for workplace messages say their writing no longer sounds like them. More than half say they've watched a coworker's personality disappear from their messages. And nearly one in three catch themselves writing in an "AI style" even when they're not using an AI tool at all.

Those are among the findings of a survey of 921 U.S. workers conducted by Omni Calculator in June 2026 — and for researchers who study how technology shapes human communication, the numbers are not surprising. What is concerning is what they represent: not just a shift in how messages sound, but a slow erosion of the human fabric that holds organizations together.

"In communication studies, there's a strong body of research that believes it is communication that creates organizations," says Kate Cassidy, assistant professor at the Faculty of Social Sciences at Brock University.

"It's not that there's a structure and people communicate what they're doing within it, but rather it is the communication that makes the organization what it is."

So that leads to the question: “If communication does create organizations, how am I using AI to create our organizations today?” she says.

“And what can that mean for tomorrow? Time savings today may not make you what you want to be over time.”

Dehumanizing effect

Essentially, it’s leaning towards a form of dehumanization, says Shane Saunderson, assistant professor of information systems with the Degroote School of Business at McMaster University.

"They are statistically averaging machines and, so, necessarily, we’re losing the edges. We're losing people that like to use weird or less common words. We're sort of converging towards the middle. And so what that means is we're becoming statistically boring, we’re all kind of homogenizing ourselves.”

With technologies that seem even more human-like — that act and speak and use ‘I’ pronouns — it makes it that much easier for us to adapt to them, he says.

“If I am adapting to a machine-like thing and machine-like behaviours, then I myself am becoming a little bit more machine-like.”

Part of what makes that dynamic so insidious, Saunderson argues, is the mismatch between how AI presents itself and what it actually is.

"The way that these things process and analyze data and spit it back out, it's quite inhuman," he says. "In many ways, these are artificial alien intelligences. They are thinking in ways that are massively different from us, but masked in this sort of facade of a human-like existence."

'It goes to the mean'

The survey finds that AI use for work communication is now pervasive. About 30% of workers reach for AI tools multiple times a day, while just 13% say they never use them for work messages. Senior leaders and executives are the heaviest users, with 52% using AI for work communication multiple times a day, compared with 22% of individual contributors.

The most common uses are:

• writing emails (47%)
• summarizing long messages (47%)
• fixing grammar or translating (46%).

More than a third use AI to reply to awkward or difficult messages (38%) and to soften or rephrase their tone (37%).

And that could be a positive, says Cassidy.

“Sometimes, it can help you laterally to imagine other scenarios that you yourself maybe can't,” she says. “But, at other times, if you're simply outsourcing the thinking process of that, you actually needed to go through that empathizing to understand what they might want.”

The concern here is that the AI voice is set up to go to the mean, to the average, says Cassidy.

“It is a very specific voice. With prompting and with resources, you can make it sound different, but it doesn't have the situated knowledge that a person embodies in their writing in ways that maybe aren't obvious to them."

The problem is compounded, Cassidy says, in hybrid and remote workplaces where connecting with colleagues can be more of a challenge.

“If you layer on decontextualized voice into that, you're even farther from your colleagues."

The generational fault line

The survey reveals a sharp generational divide in how AI-assisted writing is perceived — and it cuts against the assumption that younger workers are the most enthusiastic adopters.

Gen Z is the most skeptical generation by a wide margin. When they can tell a coworker used AI, 44% think that person didn't care enough to write the message themselves — double the 22% of Gen X who feel the same. Gen Z workers are also the most likely to trust the message less (39% vs. 20% of Gen X) and to feel unsure what the colleague really thinks (37% vs. 17%).

Given the choice, half of all workers (51%) say they'd rather receive a slightly messy message a coworker clearly wrote themselves than a polished AI-generated one. Among Gen Z, that figure rises to 61%.

The pattern holds even for spotting AI in the first place. Gen Z is the most confident in detecting it, with 21% saying they can almost always tell — double the rate of both millennials and Gen X.

Making the effort despite AI

Saunderson says the skepticism toward AI-generated writing isn't simply about effort.

"There was a study that showed that when individuals are evaluating the authenticity of a statement and they are told that it is either human or AI-generated, they unquestionably will label the AI-generated stuff as lower," he says. "We want to feel special and coddled."

The survey data reflects this kind of ambient erosion. Among workers who have noticed a colleague's AI use, 57% say they've felt that coworker's real personality disappear from their messages. Of those, 21% say they miss the old version of the person.

Saunderson calls the broader dynamic "adaptive dehumanization."

"If I am adapting to a machine-like thing and machine-like behaviours, then I myself am becoming a little bit more machine-like," he says. "The implications don't stop at the individual — they start to bleed out into the organization... The very culture of the organization is going to start to become more mechanized, more automated, more cold, more heartless."

The skill attrition problem

Both researchers flag a risk that goes beyond how messages sound today: what happens when human communication skills are no longer being practiced.

The survey finds that 30% of workers have started writing in shorter, cleaner sentences in their own unassisted writing as a result of AI use, and 11% have picked up AI-associated words and phrases such as "delve" or "moreover."

Cassidy points to a distinction that often goes unrecognized.

"The skill of taking something that's already written, going in there, removing things is far different than the skill of I'm writing it myself," she says.

Skill develops through practice, she says: "Empathy develops through empathizing. So, the time saved is, is perhaps then time lost on something that you need down the road."

Cassidy is especially concerned about junior employees.

"Newcomers to an organization need to learn the culture by being immersed in the culture," she says. "They need to talk to colleagues and to supervisors to get their situated knowledge. And so if they're turning to AI — ‘Oh, I don't want to bother people, I'll just do it myself’ — they are missing the mentoring opportunities. They're also not connecting to the culture… if somebody more senior comes to you and says, ‘No, this is how we write these kind of emails’ — that's an important learning moment."

Saunderson draws on a 1983 paper by Lisanne Bainbridge on the “Ironies of automation” to make the same point. When a process is automated, people stop practicing it.

"We're losing social skills," he says, "and so in moments where interpersonal crises, where reputational issues, where organizational culture and clash that is going on within any group start to have these challenges — who picks those up? Who solves them if we're starting to lose that ability?"

Leading by example

Both researchers single out senior leaders as a critical piece of the puzzle — and the survey data suggests those leaders should be paying attention:  Executives and senior leaders are the heaviest AI users, and the most likely to view the practice as efficient and smart (43%, compared with 24% of individual contributors).

"The senior people are setting the example," says Cassidy.

“They are the ones who play a very important role in the culture. Employees' well-being has to do with their feelings of belonging and connection. And so much of that is also connected to success of organizations. If caring emails — emails that should signal regard — are signalling 'I have something more important to do that is not this,' that very much sets the course of an organization."

Key questions before using AI

Both researchers emphasize that the answer isn't a blanket prohibition on AI for workplace communication — it's intentionality.

Cassidy offers a three-question framework she says workers and managers should apply before sending any AI-assisted message:

• Does it reflect the actual situation?
• Does it reflect the particular relationship with the recipient?
• Can I stand behind what it says?

"If you can't answer those three questions, you're increasing the chance [of a problem]," she says. "And if you then ask, ‘What is the consequence of that?’ I think that tells you a lot."

For any organization — especially for an HR team who's trying to think about their values, their vision, their culture and what they mean as a group of individuals — generative AI “represents a huge risk,” says Saunderson, “because it could be actively undermining the culture that you're trying to build without even you being aware of it.”

He offers two questions he says organizations should ask before automating any communication task or building a large language model workflow.

"Number one: ‘Are you already really good at this? If so, why do you need to automate it?’" The second question, he says, is the one most people skip. "Do you enjoy this? Is this something that if you automate, you're actually taking part of your love away?" The time saved by offloading a task you value, he argues, may not be worth what's lost.

What HR should be doing

On the training side, Cassidy has a concrete recommendation that goes beyond the usual focus on cutting out bias and avoiding hallucinations.

"I would be pulling out example messages and having people talk about ‘What do we want this message to look like? Do we want it to sound like a person, or do we want it to sound like the organization? And if it does sound like the organization, what kind of cues are we putting in our messages to sound like that?’"

Every employee, she says, “because of the temptation of AI,” needs to understand where to use their own voice or the organization's voice, and what that organizational voice and values look like.

Cassidy also recommends that organizations move beyond the binary of "used AI / didn't use AI" as a disclosure standard. "That is not enough because it doesn't say — is this text true? Is it accurate? Do you care? Is it a good reflection of our situation? Is it in line with our values, our mission, our goals?"

For Saunderson, the solution to what he calls "anthropomorphic deception" — the tendency of human-like AI to blur the line between machine output and genuine human thought — comes down to a single word: Transparency. “It’s the thing that will set you free," he says.

If you have a largely transactional organization of developers that just needs to get stuff done, then maybe it's acceptable for these systems to act like they're human because the developers treat them that way anyway, says Saunderson.

“Whereas, if it's an organization that really focuses on its people, its values, its humanity, then maybe any use of that kind of technology should come with certain caveats and reminders as you use it so that we don't delude ourselves and get lost in that sort of dehumanization.”