What AI Can’t Do
A Middle Manager's Guide to Using AI With a Human Touch
Last week’s piece talked about a vise: you as the middle manager are the reason AI adoption succeeds at your company, and you’re the role most likely to get cut in its name. If you did the Reflect exercise, you’re probably looking at a page with a lot of things circled. These are the things that landed on you because there was no one else to do it.
Most of that circled list isn’t leaving your desk this week. (Bummer!) So asking what is mine to hand back isn’t as helpful as asking if there is a way to do things in a way that costs less of you.
The Only One with EQ in the Room
AI does not understand your team. It can flag a spike in overtime hours or draft a warm-sounding check-in message, but the research on this is blunt about the limit: AI “does not possess genuine emotional intelligence... it lacks consciousness, empathy, and subjective experience.”1 It cannot tell you that someone’s gone quiet this week because of what’s happening at home, not because they’ve checked out. That is still yours to do because you’re the one who knows your team, who sits with them, who has seen them struggle and grow. You hold the relationship with your team members.
You are emotionally intelligent. You are aware of where you are strong and where you need to build additional skills; maybe you still catch yourself reacting instead of calmly reading the situation before you, but you’re working on it. Even a still-developing EQ is a category the machine doesn’t have access to at all. Between you and the machine, you’re the only one of the two of you capable of having any true EQ in the first place.
Despite that, the pressure seems to be running in the other direction right now. Leadership wants AI to take more off your plate, and eventually that might include some of the people work. Knowing exactly where the line is between these things keeps you from outsourcing judgment to a machine that isn’t equipped to make that call.
Checking the Plan & the People
AI is useful, but maybe not where we were expecting. Hand it the structural work: run your new onboarding process past it and ask where the steps contradict each other. Feed it last quarter’s project plan and ask where the timeline and the stated priorities don’t match. Ask it to find the gaps in a curriculum, a policy draft, a rollout schedule. This is AI’s sweet spot. It can hold a hundred pages of a plan in view at once and catch the inconsistency you’d only find on a third read, if you ever had time for a third read.
What that gives you back is bandwidth. Once the structural check is off your desk, what’s left for you to do is the thing AI still cannot do at all: watch how the plan actually lands on your team. Is the new process making one person’s job harder in a way the document doesn’t show? Is the team nodding along in the meeting and then not adopting any of it? That assessment requires being in the room with your team. There’s only so much you can pay attention to, so pay attention to the things that only you can do. As Forbes contributor David Morel puts it, a human still has to give an AI-driven outcome that “final pass”2 before it can be trusted. The same logic holds whether the outcome is a hiring recommendation or a rollout plan.
This is the emotional intelligence trait of Reality Testing. Let the tool check the plan against itself. You check the plan against the actual, specific humans who have to implement it.
The Fear in the Room
Your team is anxious about AI implementation, whether or not anyone’s said so out loud. A 2024 McKinsey survey found seventy percent of employees fear AI will shrink their role, but eighty-three percent say they’d embrace it if leadership explained why and what it meant for them. Leadership consultant Shaara Roman puts it more bluntly: culture is the biggest risk in AI adoption, not the tool itself, and people who don’t believe they belong in the future won’t help build it.3
Fear isn’t always labeled as fear. Fear shows up as the person who’s suddenly slower to try the new tool than anyone expected when normally they are an early adopter; they aren’t resistant to change but to them, using AI well feels like evidence they can be replaced by it. Fear shows up as the person who over-adopts instead, running everything through AI and flagging it constantly, because looking enthusiastic feels safer than looking unnecessary. Fear shows up as the question in a team meeting, “is this going to change our headcount?”
You know from last week’s article that you are the variable in how this plays out: teams that believe their manager genuinely backs AI use are nearly nine times more likely to say it’s actually changed how they work. Their fear about this is now sitting on your desk, next to your own.
You can manage your team’s fear without absorbing it. Not with platitudes and reassurance you don’t believe and that most people can tell are fake. You also can’t perform enthusiasm you don’t feel to keep morale up. Instead, say what you know, admit what you don’t, and let your own steady use of the tool, checking it, correcting it, deciding where it doesn’t belong, demonstrate your values to your team.
Calm is contagious the same way fear is. You don’t have to manufacture either one. You just have to take a deep breath and stop amplifying the fear that’s already in the room.
Reflect & Act
Reflect
Think of something you’re currently checking by hand that’s really a structural question: does this plan hang together, are there gaps, does it contradict itself somewhere? Now picture the hour you’d get back if AI did that pass instead. Would you actually spend it watching how the thing is landing with your people, or would it just get swallowed by the next task in line?
Also: has anyone on your team asked you, directly or sideways, whether AI is coming for their job? What did you actually say back?
Act
This week, hand one structural check to AI, a plan, a process, a draft, something with an internal logic to test, instead of doing that pass yourself. Use the time it frees specifically to listen to your team. How is the AI implementation impacting them? What are they excited about? What are they learning? What is bringing fear into the equation?
Say one true thing to your team about where AI use in your company stands right now, instead of a reassurance you don’t fully believe or an enthusiasm you’re performing.
ESCP Business School, “Artificial Intelligence and Emotional Intelligence: The New Frontier of Human-AI Synergy,” ESCP News, February 17, 2025, https://escp.eu/news/artificial-intelligence-and-emotional-intelligence.
David Morel, "Emotional Intelligence Is More Important Than Ever in the Age of AI," Forbes, January 13, 2025, https://www.forbes.com/sites/davidmorel/2025/01/13/importance-of-emotional-intelligence-in-the-age-of-ai/.
"When AI Joins the Team: The New Leadership Playbook for Intelligent Workplaces," Disruption Magazine, accessed July 8, 2026, https://disruptionmagazine.digital/news/when-ai-joins-the-team-the-new-leadership-playbook-for-intelligent-workplaces. (McKinsey 2024 statistic cited within this piece, not McKinsey's own publication — the original McKinsey report isn't linked from the source.)