Keep Your Expertise in Use When Working with AI
⏱️ Read time: ~4 min
I read one section of a leadership development proposal aloud and asked myself: is this how I would actually say it?
I had used my AI-powered content library, built from the last seven years of my own work: thought leadership, proposals and actual deliverables. I had done the homework to give AI rich context and clear direction. I also used a slide-generation tool and asked AI to help with literature research. The output was good, really good. That was exactly when I noticed how tempted I was to move on without checking it.
That small moment brings me back to a question I hear often in leadership development: as we use AI more, how do we keep human judgment in the work?
That proposal was not ready just because it looked finished.
I sat down and read every word. At one point, I read sections aloud and listened for a simple thing: is this how I would actually say it? Does it reflect my experience, my point of view?
That check changed the work. The tools helped me move through research, structure and slide drafting. I still had to decide what belonged, what was missing and what I was willing to stand behind.
It reminded me of leading work as a management consultant. Even when others helped with the research and drafting, I still had to review the work and decide what I could stand behind. AI has not changed that responsibility.
I did not spend the whole time watching AI work. I invested real time in reviewing and refining. The gain was not that I somehow skipped the work. It was that I could do a more substantial piece of work, with more support and research, without building a human team around it.
This is the part of AI use I keep returning to. The risk is not only that we hand a decision to a machine. It is that a polished answer makes us feel as if the decision has already been made for us.
I notice this in myself. If the last few outputs were useful, my attention drops. Add a full calendar, constant interruptions and the pressure to move faster, and skipping the check starts to feel like a sensible shortcut.
That is why “a human in the loop” is not enough as a principle. A person can be present and still be on autopilot. The human needs to stay in the lead: setting direction, bringing context and making the decision.
For me, that means building review into the work, rather than relying on how careful I happen to feel that day. I read the complete output, not just the opening. I read some parts aloud. And I ask whether the content is accurate, relevant and something I would be comfortable saying in a room with the people it is meant for.
If you lead a team that uses AI, the useful question may not be, “Which tasks should stay human?” It may be: “Where must people exercise judgment, and what routine will help them do it every time?”
A review step should be proportionate to the work. A draft social caption and a recommendation that affects a person or a client do not need the same level of scrutiny. But for any work where quality, trust or a consequential decision is at stake, the check should be part of the process, not an optional final glance.
I also worry about what happens if we stop practising that judgment. If we stop learning how to assess AI output, we may get less practised at challenging it. We should design work that keeps our expertise in use, not work that quietly trains us out of it.
Key takeaway
AI can expand what we are able to do. But review is not leftover work after AI has finished. It is how I make sure the result reflects my expertise, responsibility and voice.
In scuba diving, we say, “Plan the dive and dive the plan.” For AI, decide what will be reviewed, when and by whom, then stick to the plan.
Elena
P.S. On Tuesday I am hosting a live session to showcase my slide generation system and how it compares to the likes of Gamma, Canva, Claude Design. Haven't signed-up, yet? Now's the time 😊
