When AI enters human communication, leaders must become slower to assign intention, even as technology allows them to communicate faster.
—————————————————————————————————————————————-
“What is preventing Samantha from moving forward?”
Samantha stared at the question in the AI-generated meeting summary. That was not what she and Tony had agreed to.
Their virtual meeting had gone well. They had jointly decided to delay implementation of the new project involving their teams until they received additional data. But now the summary included aggressive “suggested next steps” and questions that appeared to challenge Samantha.
Samantha assumed Tony had added the additional questions and steps. Within seconds, she made up a negative story: Tony had pretended to agree, changed his position after the meeting, and was preparing to blame her for the delay.
There was only one problem: Tony had not written, approved, or even noticed the question. The AI tool had generated it.
AI is introducing a new version of an old leadership problem: the gap between intention and interpretation. The tool can insert language between two people, but the recipient may assign that language, and the motivation behind it, to the human sender. While AI tools have undoubtedly increased efficiency and improved the speed of communication, they have also introduced a new source of ambiguity that can cause leaders to make up negative stories and distrust each other.
Unfortunately, these situations are becoming more and more common in the workplace. The same misunderstandings can occur when AI turns a leader’s appreciative email into a cold command, converts nuanced coaching into a harsh performance review, or summarizes a heated but productive debate as an unresolved conflict.
The strength of AI tools is that they use polished language with structured summaries and concise recommendations, which can create a false confidence that they are the truth. But the polished language should not be misinterpreted as an accurate reflection of what others actually meant.
The issue with most tools summarizing meetings is that they don’t just faithfully record the conversation. They also add AI’s interpretation of what was said and its own generated recommendations. But some leaders may not be aware of the distinction between these sections.
There are usually two parties in any relationship, but there are now three participants in the communication (including the AI tool), and the third participant’s contribution is not always clearly labeled. Specifically:
– What the participants explicitly said is human-generated
– AI’s summary of what the participants said is AI-interpreted
– Additional recommendations and questions are AI-generated
When these categories are not clearly differentiated, recipients can mistake an AI recommendation for a colleague’s intention.
For example, the AI summary after Samantha’s meeting with Tony looked like this:
“Agreed next step: Wait for the additional data before making an implementation decision.
Suggested next steps: Establish a firm implementation deadline and begin preparing the rollout.
Questions to consider:
• What is preventing Samantha from moving forward?
• Where are the weaknesses in Samantha’s position?
• How should Tony proceed if Samantha cannot obtain the necessary support?”
The first statement reflected the meeting. The remaining statements were generated by the tool. But because they appeared in the same polished summary, Samantha treated all of them as expressions of Tony’s thinking.
This exact situation happened recently with one of my clients and caused distrust and confusion, and is confirmed by research. Recent studies on human–AI teams found that omissions and ambiguous expressions can reduce trust, communication quality, and team performance. Other research shows that people often judge others’ use of AI more negatively than their own and may trust a communicator less once AI involvement is disclosed. Together, these findings reveal an uncomfortable paradox: AI can improve the quality and speed of communication while simultaneously making us more suspicious of the person communicating.
That is why leaders today have an obligation to come with new best practices for their teams, including both the senders and the receivers of AI-assisted communication.
Sender- If AI is communicating in your name
1. Review: Is the summary accurate? Are decisions, interpretations, and AI suggestions clearly labeled?
2. Check: Does the tone represent how I want to communicate?
3. Own: Am I willing to be accountable for every statement
Receiver- If you receive AI-assisted communication
1. Pause: Do not immediately interpret unexpected wording into evidence of negative intent.
2. Get curious: What was said by the person, summarized by AI, or newly generated by AI? What other explanations might there be?
3. Verify: Check with the person before reacting to the story you made up
For example, Samantha can pop into Tony’s office and say:
“Tony, I noticed that the summary includes a question about what is preventing me from moving forward. That did not sound consistent with our conversation. Did that question come from you, or was it generated by the tool?”
AI tools are not going away, nor should they. They can make our communication faster, and more efficient. But they also add another layer between what one person intends and what another person interprets. That makes assuming positive intent more important.
When words surprise us, we need to pause before assuming intentions. We need to ask what else could explain them. Then we should check with the other person. Before we respond to the message, we should make sure we know who (or what) actually wrote it.
In the age of AI, one of the most important leadership capabilities may be deeply human: the humility to recognize that the first story we tell ourselves may be wrong.
—————————————————————————————————————————————-
Amer Kaissi is a Leadership Keynote Speaker, Professor of Leadership, Executive Coach and Leadership Book Author.
