Generative AI has made polished communication available to almost anyone with a prompt. For business leaders, that promises faster correspondence, clearer presentations, and fewer hours spent rewriting difficult messages. It also raises a question that productivity figures cannot answer: does a better message reflect better judgement?
The concern deserves attention as workplace adoption expands. Gallup’s July 2026 report found that 52% of US workers used AI in their roles, including 30% who used it at least a few times a week and 15% who used it daily. Among AI users, 51% employed it for writing and editing.
These figures help explain why AI and leadership now intersect in everyday management. An executive can produce a reassuring announcement before deciding what reassurance the organisation can honestly offer. A manager can generate thoughtful feedback without examining the employee’s work closely enough to justify it.
The danger lies in allowing fluent language to conceal unfinished thinking. Emerging research suggests that organisations should assess AI-assisted communication through the quality of the judgement behind it, alongside the efficiency of producing it.
Better Writing Does Not Settle The Leadership Question
The case for using AI to improve communication is substantial. In a 2023 experiment published in Science, economists Shakked Noy and Whitney Zhang assigned professional writing tasks to 453 college-educated participants. Access to ChatGPT reduced average completion time by 40% and increased assessed output quality by 18%.
Those findings concern defined writing assignments, however. The tasks did not require the detailed organisational context or factual verification that many executive communications demand. They demonstrate gains in producing written work, rather than improvements in leadership effectiveness.
That boundary matters when companies decide what to automate. Drafting a meeting summary and explaining a restructuring involve different responsibilities, even when both arrive as emails. The second requires someone to weigh competing interests, acknowledge uncertainty, and answer for the consequences.
For international organisations, AI can also help leaders express ideas in a language they use less confidently. Organisations should not confuse linguistic fluency with managerial competence, or treat writing assistance as evidence of weak judgement. The relevant question is whether the sender understands and owns the substance.
Why Polished Messages Can Still Lose Trust
Research by Peter W. Cardon and Anthony W. Coman, published in the International Journal of Business Communication, examined perceptions of AI-assisted workplace writing among 1,100 working professionals. Participants evaluated the authorship of a message congratulating a team and setting new objectives.
Although AI-assisted writing could appear professional, medium to high levels of assistance raised doubts about the sender’s trustworthiness, including their sincerity, caring, and ability. The findings concern perceptions within a specific study; they do not establish that every AI-assisted message damages trust.
Writing for the World Economic Forum, Harvard Kennedy School adjunct lecturer Allison Shapira described a “believability gap” between a leader’s professional presentation and the confidence an audience places in that person.
A plausible explanation is that employees judge communication partly as evidence of attention. Specific praise suggests that a manager noticed someone’s contribution. A candid explanation suggests that a leader considered the consequences of a decision. When recipients suspect that this attention was absent, elegant wording may offer little reassurance.
Consider a hypothetical announcement about reduced budgets. Employees may want to know which commitments will change, how workloads will be managed, and when decisions will become final. A message expressing appreciation while avoiding those questions leaves the central leadership task unfinished, regardless of who drafted it.
The Risk Begins Before The First Draft
Communication helps leaders develop a position, as well as convey one. Preparing an explanation can expose a weak assumption, an unsupported promise, or a conflict between stated priorities and actual decisions. The struggle to write clearly sometimes reveals that the underlying decision is unclear.
AI can shorten that process, but the result depends on how the leader uses it. Asking a system to make a decision sound persuasive may bypass the examination needed to determine whether the decision deserves support.
A 2025 study by researchers from Microsoft Research and Carnegie Mellon University surveyed 319 knowledge workers, collecting 936 examples of generative AI use. Greater confidence in AI was associated with less self-reported critical thinking, while greater confidence in participants’ own abilities was associated with more.
The study relied on reported experiences and did not demonstrate permanent cognitive decline or establish that AI causes weaker leadership. It nevertheless identifies a relevant risk: confidence in a tool can influence how carefully people examine its output.
For executives, that scrutiny must extend beyond factual accuracy. A draft may contain correct information while overstating certainty, minimising disruption or implying a commitment the business cannot fulfil. Editing for tone alone will not resolve those problems.
Accountability Cannot Be Delegated With The Writing
Leaders have long relied on speechwriters, communications teams and advisers. Authorship has never required an executive to type every sentence personally. The enduring obligation is to understand the message, approve its claims and accept responsibility for what follows.
AI makes this obligation easier to overlook because a complete draft can appear before anyone has challenged its premises. A useful safeguard is to establish the substance first: what has been decided, what evidence supports it, who bears the consequences, and what remains unresolved.
The leader can then use AI to test whether the explanation holds together. It can identify unanswered questions, propose alternative interpretations, and expose language that suggests more certainty than the evidence allows.
Even these suggestions require human judgement. A simulated employee reaction cannot replace a conversation with employees. The system cannot infer an organisation’s unspoken concerns reliably from a brief prompt, particularly when management itself has supplied an incomplete account.
The final test is practical. A leader should be able to explain the reasoning without the prepared text and respond when someone challenges it. If the message collapses under ordinary follow-up questions, further polishing will achieve little.
Preserve The Work That Develops Managers
The implications extend beyond senior executives. Writing feedback, explaining priorities and handling disagreement give developing managers opportunities to practise judgement. If organisations automate these activities without preserving the thinking involved, they risk rewarding finished documents while overlooking how managers reach conclusions.
That risk should shape training. Before asking AI to draft a performance discussion, a manager can identify the relevant observations, distinguish evidence from impressions, and decide what support the employee needs. AI can then help organise the conversation without supplying the assessment itself.
Research described by MIT Sloan in June 2026 offers a useful example of deliberate review. Renée Richardson Gosline and colleagues worked with Commonwealth Bank of Australia to examine what happened when people explained why they were following AI recommendations.
MIT Sloan reported that these checkpoints reduced uncritical reliance and improved accuracy without significantly increasing task time. The work is ongoing research, so it should not be treated as a universal claim for every setting. It does suggest that a brief pause for reasoning can preserve scrutiny without sacrificing efficiency.
Companies could apply that principle to important communications by asking managers to explain the evidence, commitments, and uncertainties in a draft before approving it. Such a review tests understanding more effectively than asking whether the text sounds professional.
Measure Understanding, Not Message Volume
Businesses also need a broader definition of communication productivity. Producing an announcement quickly offers limited value if teams spend the following week interpreting it, correcting assumptions, or seeking answers that the original message omitted.
Useful measures therefore include whether employees understand a decision, whether responsibilities are clear, and whether managers can address questions consistently. These measures connect communication to execution and make it harder to mistake increased output for organisational progress.
The time saved by AI creates a choice. Leaders can use it to generate more messages, or reinvest some of it in listening, examining evidence and meeting people affected by their decisions. The latter gives communication a stronger foundation.
AI may make the first draft almost effortless. Leadership still requires someone to decide what is true, what is fair, and what the organisation is prepared to do. As professional language becomes easier to produce, the ability to substantiate it will carry greater weight.
Stay connected with Business Herald for the latest business news, insights, and updates.
Follow us on Facebook, Instagram, LinkedIn, and YouTube.
Join our growing community on WhatsApp and Telegram for real-time updates delivered directly to you.
