AI can build the material. It cannot run the room.

By McKinley Malbrough III. Updated September 22, 2026.

Here is the finding that matters for anyone who trains people: AI took the part of training that was never the hard part. Writing the deck, the handout, the quiz, the follow-up email. What it did not take is the live judgment in the room, and that is the part almost nobody was trained on.

What does AI actually change about training?

It collapses production. A session outline that used to take a week takes an afternoon, so content stops being the thing you pay for. 91% of companies plan to increase AI spending in learning and development in 2026, and 59% of global heads of learning and talent now rank AI as their top priority, ahead of leadership development and reskilling. When everyone can generate material at the same speed, the difference is what happens live.

So the value moves to the facilitator. Reading the room, adjusting in real time, deciding what to do about one specific human in front of you. That is the skill I teach.

Why do live sessions still finish when recorded courses do not?

People finish when other people are waiting on them. Self-paced online courses finish at roughly 10 to 20%, and cohort programs that run live report 85 to 96%. Paying for training nobody completes is not a savings. It is a write-off with a receipt.

What can AI not do in a live session?

It cannot notice that the person who talked for the first twenty minutes went quiet after one comment, and work out why. It cannot tell the difference between a room that agrees and a room that has given up arguing. It cannot decide what a manager should do about a specific employee with a specific history, because that decision belongs to a person who carries the consequences.

It also cannot handle the moment a session stops being a training. Someone discloses something serious in front of eleven colleagues. What you do in the next thirty seconds decides whether that person is protected or exposed. I teach a protocol for that moment, and no model runs it for you.

How do I use AI in my own practice?

I draft with it, then I make it show its work. I ask the same question three ways and watch whether the answer holds. When it gives me the polished, agreeable version, I push until it gives me the version with the risk still in it. That habit, not the tool, is what I teach facilitators.

Full method, with examples from my HR practice, is on AI and HR at Urban Human Resources.

What this means for urban organizations

You do not need to buy a learning platform to train your people well. You need two or three people inside your organization who can run a live session that changes what happens the following Monday, and who know how to use AI to prepare in a fraction of the time. That is a training problem, not a software purchase.

What this means for urban professionals who teach

Your value is not the deck you built. Anyone can generate that now. Your value is the room: what you notice, what you name out loud, and who leaves knowing something they were never going to be told. Build that skill on purpose and AI makes you faster instead of replacing you.

Sources: Gartner and iVentiv data compiled by VirtualSpeech, January 2026; Ruzuku, State of Online Courses 2026. Data current as of September 2026.

See the programs Book a free consult