Recently I was asked whether AI is costing us the human touch in internal communication. The research says the risk is real, and it is more specific than ‘AI feels cold’.
1. Trust predicts AI readiness better than the technology does
McKinsey found that employees with low trust in their organisation’s support during AI transformation are 1.5 times more likely to feel anxious about AI-related changes. About one in five employees report AI-related anxiety, and that rises to one in four among middle managers. McKinsey also warns that reassurances such as “no jobs will be lost” may ease anxiety for a while without building trust.
Likewise Gallagher‘s 2026 report points the same way: organisations sending high volumes of communication saw a 30% jump in seeing leadership trust as a significant risk. Link to the report:
Many leaders answer anxiety with a better-worded announcement. The evidence points to a different fix: trust has to be built through behaviour over time.
2. How a message is made can change how the sender is judged
A 2025 study in the International Journal of Business Communication surveyed 1,100 professionals. Only 40–52% saw supervisors as sincere when they used high levels of AI, compared with 83% for low-assistance messages. Employees rated the messages as professional. They still doubted the sender. The researchers concluded that relationship-oriented messages, such as praise, empathy and motivation, are better handled with minimal technological intervention.
AI is fine for the reminder and the routine update. A message about restructuring or recognition needs the leader’s own words and voice.
3. Psychological safety is the enabling condition
Frazier and colleagues’ meta-analysis in Personnel Psychology drew on 136 independent samples, over 22,000 individuals and nearly 5,000 groups. It found that psychological safety is strongly linked to information sharing and learning behaviours. Most of the studies were cross-sectional, so causation can’t be claimed.
For AI, a 2026 preprint of 2,257 employees at a global consulting firm found that psychological safety reliably predicted whether people adopted AI tools, but not how often or how long they used them afterwards. People who can ask questions and admit confusion take the first step. This preprint hasn’t been peer reviewed, so treat it as early evidence.
4. Good internal communication is two-way
Rita Men’s research shows that transparent communication, marked by substantial information, accountability and employee participation, contributes largely to employee trust, commitment and satisfaction. In a survey of 402 employees, transformational leadership shaped symmetrical (two-way) communication, which in turn strengthened employee–organisation relationships.
Open rates measure reach. They can’t show whether someone felt safe enough to say ‘I don’t understand’ or ‘I’m worried’.
5. The Indian context needs care
The KPMG–University of Melbourne study of over 48,000 people in 47 countries found 76% of Indians willing to trust AI, against a global average of 46%. The same study reported that 57% of workers have concealed their AI use. That second figure reached me through secondary reporting. Trust in the technology is not trust in how the organisation will use it, and hidden use is a sign that employees don’t feel safe being open.
What this means for leaders
- Match the channel to the stakes. Use AI for logistics and a face-to-face conversation for anything involving jobs, performance or change.
- Treat AI use by leaders as something to discuss openly. Don’t hide it.
- Equip middle managers first. They carry the most anxiety and shape how everyone else experiences the change.
- Measure understanding and willingness to speak up, alongside reach.
- Build two-way channels where questions are welcome.
AI can help deliver the message. Leaders make it believable.
Where have you seen this gap between sending and understanding?
