A management team is discussing a new initiative. The presentation is convincing, the numbers make sense and, after forty-five minutes, the person leading the meeting asks the inevitable final question: “Any concerns?”
There is a brief silence, someone says it looks good, another person nods and a third suggests that everyone should “stay close during implementation,” a wonderfully useful expression because it sounds responsible without requiring anyone to explain what it actually means.
The meeting ends with general agreement. Within ten minutes, one person is telling a colleague that the timeline is unrealistic, another knows Operations cannot deliver what has just been promised, and someone else is still wondering how he became responsible for something he does not remember agreeing to.
Everyone had noticed something. Nobody had brought it into the room.
Now imagine giving the meeting transcript to AI and asking it to identify hesitation, assumptions that went unchallenged, signs of disagreement or questions that should perhaps have been asked before the decision was made. There is a reasonable chance that the technology we are teaching to recognise human signals will point out something the humans in the room had already sensed.
Which raises a slightly uncomfortable question.
We are putting extraordinary effort into developing artificial intelligence, including its ability to recognise sentiment, interpret language, adjust tone and help us prepare for difficult conversations. We are teaching AI to become better at reading us, while in many organisations the humans still leave a meeting saying they are aligned and start disagreeing properly as soon as they reach the corridor.
Emotional intelligence is sometimes treated as the softer relative of the more serious business skills, associated with empathy, listening and being good with people, but in leadership it asks considerably more of us. Noticing the hesitation before the yes is only the beginning; we also have to interpret what we have noticed without immediately assuming we know what it means, manage our own discomfort with disagreement and create enough space for someone to say what may be easier left unsaid.
Sometimes that means interrupting a beautifully efficient meeting with: “We all seem to agree. Does anyone actually disagree?”
AI can increasingly help us spot patterns, challenge assumptions, suggest questions and show us something we may have missed. But recognising a signal and knowing what to do with it are different things, and the useful part still begins when a human being decides whether to ignore the hesitation, explain it away or become curious about it.
We don’t need to compete with AI on who can process more information or produce an answer faster. That contest has a fairly predictable ending.
But while we are investing so much in teaching artificial intelligence to recognise more about us, perhaps it is worth paying equal attention to what we do with the emotional intelligence already sitting around the table.
Because it would be a little embarrassing if AI learned to read the room before we remembered what to do once we had.