That preference is operationally important. Email, Teams chats, project updates, performance feedback, and executive announcements are not merely containers for information. In hybrid workplaces, they are much of the relationship. When employees cannot tell whether a manager or coworker actually chose the words on a screen, they can begin to question whether the sender owns the message too.
The submitted Digital Journal story frames this as a Canadian workplace issue. The underlying Omni Calculator report does not. Its respondents were U.S.-based employees surveyed through Prolific from June 8 through June 15, 2026. That does not make the findings irrelevant to Canadian businesses, but it does mean the article’s discussion of Toronto, Vancouver, Calgary, Montreal, Canadian AI institutions, and Canadian policy is context rather than evidence. There is no Canadian sample in the study.
What the survey actually found
Omni Calculator’s survey asked employed U.S. adults how they use AI to write, rewrite, or polish workplace communication. Thirty percent said they do so multiple times per day; another 24% use it a few times per week. Senior leaders and executives were the heaviest reported users, with 52% saying they reach for AI multiple times a day, compared with 22% of individual contributors.
That leadership split is worth more attention than the broad “AI is changing work” conclusion. Managers tend to write messages that carry consequences: priority changes, policy updates, performance feedback, organizational announcements, and explanations for decisions employees may dislike. A generic project update is annoying. A generic message about a reorganization, return-to-office mandate, missed promotion, or workload increase can look evasive.
Workers in the Omni survey most commonly used AI for writing email and summarizing long messages, at 47% each, with grammar correction or translation close behind at 46%. Only 14% reported using it for Slack, Teams, or chat messages. The distinction suggests that workplace AI has not yet turned every interaction into machine-generated prose; it is concentrated in the formal, high-volume writing that employees already find tedious or risky.
Still, formal communication is where voice matters most. The employee who normally writes concise, blunt updates can sound unlike themselves after a tool adds softeners, transition phrases, generic reassurance, and the familiar “please let me know if you have any questions” closing. The manager who normally speaks plainly can send a message so symmetrical and carefully neutral that staff cannot infer what they actually think.
Omni found that 25% of AI users said their messages now sound less like them and more generic. Forty percent said they had shed small typos and imperfections, 27% reported dropping slang or casual phrasing, and 30% said they had noticed themselves writing or talking in an “AI style” even when they were not using an AI tool.
Those numbers should not be treated as proof that language models are literally rewriting personalities. They are self-reported impressions from one online survey, not a longitudinal analysis of participants’ actual email or Teams history. But the results identify a real organizational risk: standardizing the delivery of a message can also standardize the social signals people use to judge its sender.
The “AI penalty” is not simply anti-technology sentiment
The broad claim that workers want “human” communication has independent support, although the research is more nuanced than the Omni survey’s headline suggests.
A 2026 peer-reviewed study in Computers in Human Behavior: Artificial Humans tested how people reacted to human, AI-assisted, and fully AI-generated messages in workplace-email and social-media settings. The researchers reported an “AI penalty”: AI-mediated communication was perceived as less trustworthy, less authentic, and less useful for knowledge uptake. That result aligns with Omni’s finding that 27% of workers said they trust a message a little less when they can tell a coworker used AI, while 25% said they could not tell what the sender really thinks.
But a separate study published in Business and Professional Communication Quarterly reached a less sweeping conclusion. In an experiment involving 887 working adults, researchers found that AI-generated workplace messages were generally viewed as professional, effective, efficient, confident, and direct. Sincerity and caring scored somewhat lower in some scenarios where AI involvement was disclosed, particularly for ChatGPT-generated messages.
Taken together, the record does not support a blanket rule that AI-written communication fails. It supports a narrower and more useful conclusion: AI is generally good at producing workplace-appropriate language, but it can impose a relational cost when the recipient believes it has replaced the sender’s judgment or effort.
That is why a routine status update and a sensitive employee message should not be placed in the same AI policy bucket. Tone is not merely an aesthetic setting in Outlook or Teams. In messages involving accountability, disagreement, recognition, apology, performance, or difficult change, the reader is evaluating intent as much as grammar.
Gen Z’s skepticism complicates the “digital native” story
The most striking part of Omni Calculator’s survey is that Gen Z respondents were more skeptical of detectable AI writing than older groups. Forty-four percent said that when they notice a coworker used AI, they think the person did not care enough to write the message themselves. That compares with 22% of Gen X respondents. Gen Z was also more likely to prefer a slightly messy human-written message over a polished AI-written one, 61% to 44% among Gen X.
This is not evidence that young workers reject workplace AI. The same age group uses digital writing tools constantly, and Microsoft’s Work Trend Index has documented extensive adoption of AI at work across generations. It is evidence that fluency with the technology does not translate into tolerance for its use in every social context.
For IT leaders, that presents a policy problem. A rollout that treats Copilot as an always-on “professionalism layer” may deliver clean prose while teaching employees that a polished message is expected by default. Omni found that 32% of respondents felt more pressure to make their work messages sound perfect since AI became common. Most did not report that pressure, but one-third is enough to create a meaningful shift in workplace behavior.
The pressure is likely to land unevenly. Employees working in a second language, junior staff, people writing to senior leaders, and workers in highly regulated environments have obvious reasons to seek help with tone and clarity. Removing AI assistance from those situations would be counterproductive. The mistake is treating that legitimate support as a mandate to outsource the entire message.
The real problem is authorship, not typos
The Digital Journal piece leans heavily on the idea that small imperfections prove a message is authentic. That overstates the case. Typos can signal haste, poor proofreading, or inaccessible communication just as easily as humanity. No employee needs a policy encouraging misspellings, vague requests, or error-filled client correspondence to prove they wrote something themselves.
What recipients are detecting is usually not a missing comma. It is a lack of identifiable authorship.
A useful message has evidence of the sender’s perspective: a specific decision, an honest explanation, a reference to a shared event, a clear request, a stated trade-off, or language that reflects the relationship between the people involved. AI can assist with structure, readability, translation, and tone, but it cannot supply genuine ownership unless the human supplies those ingredients first.
Microsoft Research’s March 2026 work on “mimetic alignment” points to the technical challenge. The researchers found that systems trying to communicate on someone’s behalf need to capture how that person actually communicates; generic output built from shallow persona prompts does not reliably do that. Even their more individualized approach produced substantial variation across people and scenarios.
That is a warning against the simplistic promise that a corporate AI assistant can “learn your voice” and solve the problem. Voice changes by audience and stakes. The version of a manager’s voice needed for a customer escalation is not the voice needed for a direct report’s bereavement leave, a software engineer’s incident update, or a team’s celebration after a release.
What a workable Copilot policy should preserve
Organizations do not need to choose between unedited human writing and unreviewed AI output. They need to distinguish assistance from substitution.
A practical policy for Microsoft 365 Copilot, ChatGPT Enterprise, Gemini for Workspace, or Claude for Work should make several expectations explicit:
- Employees can use AI to clarify, translate, organize, summarize, or improve grammar, while remaining responsible for facts, tone, and the final message.
- Managers should personally draft or materially rewrite communications involving performance, compensation, disciplinary action, job changes, reorganizations, safety incidents, customer apologies, or other decisions that affect people directly.
- AI-generated summaries should identify source material and preserve uncertainty, dissent, action owners, and deadlines rather than flattening a meeting into a polished but context-free recap.
- Teams should avoid using AI to manufacture empathy. A generic expression of concern can be worse than a brief, plainly written acknowledgement from the person accountable.
- Administrators should provide approved tools and data-handling guidance, because workers will otherwise use consumer AI services on their own. Microsoft’s Work Trend Index found that many workers had already brought their own AI tools into work environments without formal direction.
The final point belongs in the same conversation. A company that tells employees to “be authentic” but does not provide secure, governed tools is likely to get the worst combination: unsanctioned AI usage, inconsistent quality, and no shared expectation about when a message requires a human hand.
The emerging fault line is not between people who use AI and people who do not. It is between communication where a person remains visibly accountable for the words and communication where an automated draft becomes a substitute for attention. As Copilot and its rivals become ordinary fixtures in Outlook, Teams, Word, and Google Workspace, the organizations that preserve that line will keep the efficiency gains without making every difficult message sound as though it came from the same cautious machine.
References
- Primary source: Digital Journal
Published: August 7, 2026 at 10:50 PM UTC
As AI writes more of our messages, are we losing our workplace personalities?
As Canadian organizations continue adopting generative AI technologies, leaders may need to think carefully about where automation adds value and where human expression remains essential. The future workplace will almost certainly involve more AI. The challenge will be ensuring that, amid the...www.digitaljournal.com - Related coverage: prnewswire.com
Millennial Branding and Randstad US Release First Worldwide Study Comparing Gen Y and Gen Z Workplace Expectations
/PRNewswire/ -- Millennial Branding, a Gen Y research and consulting firm, and Randstad, the second-largest HR services and staffing company in the world,...
www.prnewswire.com
- Related coverage: gokorbyt.com
Korbyt and Reworked Report: Fixing Workplace Communication Gaps
Why do employees ignore workplace messages? Korbyt’s latest survey uncovers the impact of poor communication—and how to fix it with smarter strategies.www.gokorbyt.com - Related coverage: assets.ricoh-usa.com
- Related coverage: apps.apple.com
- Related coverage: nairaland.com
- Related coverage: grady.uga.edu
Too Polished to Be Human: Understanding Employees’ Detection and Perceptions of Leaders’ Use of Generative AI in Workplace Communication - Grady
Ruoyu Sun, Mengqi (Maggie) Liao, Zheng, X., & Li, W. “Too Polished to Be Human: Understanding Employees’ Detection and Perceptions of Leaders’ Use of Generative AI in Workplace Communication,” paper to be […]
grady.uga.edu
- Related coverage: gallup.com
