Letting your managers “figure out AI on their own” feels efficient right now. It will cost you a lot more later, and the bill arrives quietly — as inconsistency, not as a single visible disaster.
Most companies didn’t choose this on purpose. AI tools showed up fast, budgets for formal training lagged behind, and managers did what capable people do when nobody gives them a system: they experimented. Some became power users. Others avoided the tools out of caution or quiet anxiety. A few used AI in ways that created real risk nobody noticed until much later. No one designed this outcome. It just happened, by default, in thousands of companies at once.
Why self-taught AI use feels fine until it isn’t
A self-taught manager who’s curious and careful might genuinely get good results. That’s the trap — it works often enough to feel safe. The problem isn’t any single manager’s skill. It’s the variance across an entire management layer, where every manager is operating on a different, untested understanding of what the tool can and can’t do.
One manager uses AI thoughtfully to draft fair, well-reasoned feedback. Another feeds confidential team data into a tool without realizing the privacy implications. A third doesn’t trust AI at all and quietly tells their team to avoid it, undercutting company strategy without anyone in leadership knowing. Same company, three completely different outcomes, and not one of them was trained on purpose.
The real cost isn’t a scandal. It’s inconsistency.
Most companies worry about the dramatic version of this risk — a data leak, a biased decision, a public mistake. Those things matter, and they happen. But the quieter, more expensive cost is consistency itself. When every manager’s AI judgment is self-taught, the business has no reliable floor for how decisions get made, how feedback gets written, or how risk gets assessed. Two employees on two different teams can get wildly different treatment, not because of policy, but because their managers learned AI differently from a YouTube video and a Slack thread.
That inconsistency erodes trust slowly. Employees notice when one manager’s decisions feel fair and well-supported while another’s feel arbitrary. They rarely connect it to “uneven AI training,” but that’s often exactly what’s underneath it.
Why “just send them a tools workshop” doesn’t close the gap
A single AI tools session teaches everyone the same features. It does not teach everyone the same judgment, and judgment is precisely where the inconsistency lives. Two managers can sit through the identical workshop and walk away with completely different instincts about when to trust the output, when to double-check it, and when not to use AI at all. Without structured practice and feedback on those judgment calls specifically, the variance survives the training untouched.
This is the same gap that has made leadership training disappoint companies for years: a one-day event creates temporary alignment in the room and almost no lasting alignment in daily behavior. AI just raises the stakes, because the inconsistent behavior now touches more decisions, faster, than it used to.
What actually closes the gap
The fix isn’t banning self-directed learning — curious managers should keep exploring. The fix is giving every manager a shared, practiced baseline: the same core judgment calls, practiced the same way, reinforced over weeks instead of taught once. Not just “here’s how the tool works,” but “here’s how we decide what to trust, what to double-check, and what never goes near AI in the first place” — built into real work, not a slide deck.
Done well, this doesn’t slow managers down. It gives the curious ones a stronger foundation and gives the cautious ones enough confidence to finally start using the tools the company already paid for.
The business case
Inconsistent AI judgment across your management layer is a hidden cost sitting on your balance sheet right now, even though no line item says so. It shows up as uneven decisions, slower adoption, and quiet risk nobody flagged because nobody was responsible for flagging it. A shared, practiced standard turns that hidden cost into a visible asset: a management team that uses AI the same way, for the same reasons, with the same judgment — no matter who’s reviewing the work that day.
So before assuming your managers’ self-taught AI skills are good enough, ask the question that actually matters: if every manager in your company is improvising their own AI judgment right now, do you actually know what’s running your business?
Recommended reading from jordanimutan.com:
2. Why Your Leadership Training Isn’t Working (And What To Do Instead)
4. Your Managers Keep Talking About Accountability — But No One Feels It
5. The “Invisible” CEO: Building a Startup Structure That Doesn’t Break When You Step Away
Most leadership training is forgotten after the seminar. 