‘If my manager has to use AI to write a message like this, why are they my manager at all?’: Academic discusses findings of study, sharing best practices
When it was discovered recently that a coding course at Staffordshire University in the U.K. was being taught largely by AI, it didn’t go over well with students.
They felt that AI wasn’t augmenting teaching; it was replacing it, without adding genuine value. The signs of AI involvement included AI‑generated slides, an AI version of the lecturer’s voice reading them aloud, American English clumsily edited into British usage, generic explanations and stray references to U.S. legislation, according to the Guardian
However, the university defended its teaching methods, saying, “Academic standards and learning outcomes were maintained” and the AI tools “do not replace academic expertise.”
With AI being used for all kinds of content creation, it’s a reminder of the risks — and best practices — for HR and management when it comes to internal communications.
Importance of trust with AI-written messages
That dynamic was recently addressed in a 2025 study in the International Journal of Business Communication by Peter Cardon and Anthony Coman: “Professionalism and Trustworthiness in AI‑Assisted Workplace Writing.”
The researchers surveyed 1,100 working professionals and showed them an AI‑generated congratulatory email from a supervisor, using either low, medium and high levels of AI assistance.
People tended to see the AI‑generated business messages as “clear, efficient and highly professional,” but when the amount of AI writing increased, trust-related perceptions declined.
Anthony Coman, lead author of the study, distinguishes between cognitive trust — “Can the person do the job?” — and affective trust — “Does this person care and do they care about me?” Both are vital in a manager–employee relationship, he says: “I need to both believe my boss is competent and leading me in the right direction and that they care about the team and care about me personally — and then I will trust them.”
The risk, says Coman — instructional associate professor and writing coordinator at the UF Warrington College of Business — is that a manager may think, “‘Oh, it’s okay, I’m busy today, I know I mean this, so I’m just going to offload this to AI.’ But recipients judge that more harshly.”
One open‑ended response in the study summed up the sentiment, he says: “If my manager has to use AI to write a message like this, why are they my manager at all?”
Types of messages matter with AI trust
The type of message also matters. Coman’s team found that employees were more accepting of AI for purely transactional or informative communication. Things changed when a message felt relational — intended to congratulate, motivate or show care.
“If a reader perceives this message as being purely transactional or informative, they're more willing to accept higher levels of AI assistance,” he says. “If they think that the message is in part to motivate or congratulate or convey some sort of a relational message, then they're much less accepting of AI use.”
The findings suggest that people who intend to leverage AI-assisted writing for workplace messages “should know that doing so may carry a risk of reputational harm,” say the researchers.
“While using AI may improve the overall perceptions of the professionalism of their writing, some recipients may doubt a supervisor’s authenticity, sincerity and caring.”
A separate “blind” study found that people may rate AI-generated messages as professional, sincere, effective and caring — but if they were told it was highly AI-generated, their perceptions of the sender went down “quite a bit,” says Coman. So, it’s more about the use of the tool, not the actual message.
‘Begin with your brain’
Coman offers tactical advice at the message level, citing the importance of human connections and trust. Managers should “begin with their brain and draft with their brain first,” he says, instead of starting with a prompt box, particularly for relational messages.
If a leader is “worried about putting the polish on it,” he says, then AI tools “can be a fantastic help — but to start with the tool carries a lot of risk for leaders.”
Coman also points to converging research on “skill atrophy” and time costs when people over‑rely on AI and it doesn’t end up saving them the time expected.
For performance reviews and similar high‑stakes discussions, he says he “would not encourage” managers to use AI — beyond perhaps background data crunching.
Coman points to separate research showing that employees are much more willing to accept an AI tool for automated tasks such as scheduling their shifts, for example, than evaluating their performance.
‘Wanted: authenticity’
Another stream of research offers a more optimistic but still cautious view. In a qualitative study in Public Relations Review titled “Artificial intelligence for internal communication: Strategies, challenges, and implications,” researchers conducted in‑depth interviews with 20 senior communication professionals focused on three areas: the general impact of AI on internal communication, the key challenges organisations face, and best practices to harness the power of AI.
On the benefits side, the practitioners reported that AI improves “efficiency, information flow, listening capability and employee experience.” Communicators are using tools to summarize meetings, generate ideas and outlines, draft or edit content, and analyze internal data.
They touted the benefits of automating repetitive tasks and hyper‑personalization to tailor content to individual employees and “enhance this interaction quality and build a relationship with the employees.”
Co-author Linjuan Rita Men — professor of public relations and director of internal communication research in the College of Journalism and Communications at the University of Florida — describes how some use AI to process large volumes of internal data, then respond with customized answers that improve “internal listening,” she says.
But one of the most persistent worries is authenticity linked to a higher reliance on AI, says Linjuan, along with the lack of “human experience and human touch in the communication.”
Co‑piloting with tech tools
Linjuan says another important theme that emerged is the “widely adopted consensus” that co-piloting is the best approach. In talking to communicators, the concern is that the use of AI to generate content will “potentially lose the human touch,” she says.
Rather than letting machines or AI do all the work, communicators should use these tools to “enhance the efficiency and efficacy of the work, but always have a human in the center.”
This co‑piloting mindset also reframes the skills question. If AI will automate parts of content generation and analysis, Linjuan argues that communicators need to “upskill and reskill to focus on more of those critical and also strategic and creative aspects, and then also really use our human experience, which AI tools don’t have.”
To disclose or not to disclose
One of the other challenging questions for both academics and practitioners is transparency: when should senders disclose AI use in internal communication — and will that help or hurt?
Linjuan’s recent experimental work on CEO “digital twin” videos suggests that disclosure can have a surprisingly large impact. In that study, which will be presented at an upcoming conference, her team compared three conditions: a real CEO video, a highly realistic AI‑generated digital twin, and a digital twin with an explicit disclosure that AI was used.
While the real CEO and the undisclosed digital twin didn’t really impact the perceived authenticity, the one with disclosure “massively influenced” the CEO authenticity, she says, adding more research is needed to better understand the impact.
And context matters: If AI is used for a “broad, generic” announcement by the CEO, that might be perceived more positively than a CEO communication in a crisis scenario or an apology, says Linjuan.
So, when it comes to disclosing the use of AI, context matters.
“It’s a complex issue… it’s an evolving situation,” she says, citing as an example the stigma around using AI.
“People feel that it's cheating. But then when this becomes part of our lives, it is enhancing our work, we're using it effectively to enhance the quality, then maybe the acceptance level for using AI might have changed, just in society in general.”
AI becoming default approach
Coman agrees that the question around AI disclosure is “tricky,” since high use of AI could damage trust with the reader.
Plus, many tools, such as Gmail’s suggested replies or Microsoft’s Outlook Copilot, are embedded by default, he says: “At what level should you begin to cite that you used AI or declare that you used AI?”
But clear policies around AI usage, he suggests, are essential — whether or not they require routine disclosure — so that teams aren’t left guessing at norms.
And even if you don’t have all the answers, as we’ve seen in the past with adoption curves around new tech, it’s important to embrace the new tools, says Linjuan.
“Like it or not, it is going to stay and it's going to be the future. So, if you do not engage in the game, then you'll be left out of the game.”
