Everyone in the business acumen simulation workshop I conducted last week was using AI
assistants to research data, understand reports, and make decisions.
Some teams performed better than others, and a couple were absolutely extraordinary. They set all-time simulation records for stock price, revenue growth, profitability, and shareholder value.
Naturally, during the final debrief, I asked the top-performing team what they had done differently. I expected to hear something about strategy, financial analysis, or decision-making discipline.
Instead, one participant said something fascinating:
"We coached our AI agent much differently than we coached each other. The feedback and the way we communicated with the AI were much clearer and more clarifying, and it made a huge difference."
The room got quiet.
As we unpacked that statement, it became obvious that they had stumbled onto something profound. The skills required to coach AI effectively are surprisingly different from the skills required to coach people effectively.
In fact, in some ways, AI is exposing weaknesses in how we communicate with each other.
As we continued discussing it, something started to bother me.
For years, leaders have been taught that coaching is fundamentally a human skill. It requires empathy, trust, emotional intelligence, and the ability to adapt your communication style to different personalities.
Yet here was a team telling me that one of the reasons they were successful was that they coached an artificial intelligence tool more effectively than they coached each other.
The more we talked, the more I realized this wasn't really a conversation about AI at all. It was a conversation about communication. AI was simply exposing habits that have quietly crept into our workplaces for years.
We have become comfortable being vague. We assume people know what we mean. We avoid difficult feedback. We rely on titles and authority to drive action.
And because humans are generally pretty good at compensating for our communication flaws, we rarely notice the problem.
AI doesn't compensate. It simply gives us back exactly what we asked for. That realization led to five observations that I haven't been able to stop thinking about.
1. AI Doesn't Reward Ambiguity
When coaching people, we often speak in shorthand.
"Can you tighten this up?"
"Make it more strategic."
"Take another look."
Humans fill in the blanks. They use context, experience, relationships, and intuition to figure out what we mean.
AI doesn't.
If your instructions are vague, the output will usually be vague.
The highest-performing team discovered that specificity dramatically improved results. They learned to define exactly what they wanted, why they wanted it, and what success looked like.
The more precise they became, the better the AI performed.
Ironically, this is also what great human coaches do. The difference is that humans are often polite enough to compensate for our poor communication. AI simply reflects it back to us.
In many ways, AI acts like a mirror. If the instructions are fuzzy, the output is fuzzy. If the thinking is clear, the output improves.
That's probably a lesson many of us should apply far beyond AI.
2. AI Loves Feedback More Than Humans Do
If you've ever had an uncomfortable coaching conversation and watched the other person become defensive, frustrated, or emotional, you know that most people don't naturally enjoy criticism.
Even when delivered professionally and with good intentions, feedback can trigger anxiety, self-doubt, or resistance.
AI has none of those limitations.
The best-performing team continuously challenged their AI assistant.
"That's not specific enough."
"Show me another approach."
"Your assumptions seem weak."
"Rebuild this analysis using different criteria."
There were no hurt feelings, no ego, and no politics. The AI didn't spend the next week replaying the conversation in its head or questioning the intent behind the feedback. It simply improved the output.
As a result, the team created a rapid cycle of feedback, adjustment, and improvement.
It made me wonder what would happen if our organizations normalized feedback with the same frequency and objectivity we use when coaching AI. Many performance problems would disappear overnight.
3. AI Requires More Structure Than Most People
One of the team's discoveries was that AI performs dramatically better when given structure.
Instead of asking:
"Analyze this business."
They would ask:
"Analyze this business as if you were the CFO. Identify three risks, three opportunities, and the likely impact on revenue, margin, and cash flow."
The structure improved the quality of the thinking.
And once again, the lesson isn't really about AI.
One of the biggest leadership mistakes is assuming people know how to approach a problem simply because they understand the objective.
We tell someone to improve customer satisfaction, increase profitability, or develop a strategy, but we don't always provide a framework for how to think through the challenge.
Great coaches do.
They provide structure.
They provide models.
They provide a way of thinking.
Great coaches don't simply tell people what to do. They help people understand how to think. AI simply forces us to become more disciplined about doing what effective leaders should have been doing all along.
4. AI Doesn't Care About Authority
This was perhaps the most interesting observation of the entire week.
Humans often respond differently depending on who is giving the feedback.
A suggestion from a peer is interpreted differently from the exact same suggestion from a vice president.
Whether we want to admit it or not, titles matter. Politics matters. Reputations matter. Organizational hierarchy influences how ideas are received.
AI couldn't care less.
A brilliant prompt from an intern produces the same result as a brilliant prompt from the CEO. The team's success came from focusing on the quality of the thinking rather than the status of the thinker.
This observation stayed with me long after the workshop ended because it raises an uncomfortable question. How many good ideas never make it to the surface because people are evaluating the source rather than the idea itself?
In many organizations, a mediocre idea from a senior leader receives more attention than a brilliant idea from someone earlier in their career.
AI doesn't care. The quality of the prompt matters. The title of the person writing it does not.
There may be a powerful lesson here for organizations trying to create more innovative cultures.
The best ideas rarely care about organizational charts.
5. Coaching AI Is Really Coaching Yourself
This was another huge insight. The team initially believed they were improving the AI.
Over time, they realized the opposite was happening. The AI wasn't learning.
They were.
Every weak answer forced them to ask a better question. Every flawed recommendation forced them to communicate more clearly. Every disappointing output exposed assumptions they didn't realize they were making.
Every mistake forced them to think more critically about the problem they were trying to solve. In many ways, the AI became a mirror reflecting the quality of their own thinking.
And perhaps that's the most important lesson for leaders.
As the debrief wrapped up, I realized that the team's insight had very little to do with technology. The future isn't simply about learning how to use AI. Millions of people will eventually learn that.
The real competitive advantage may come from learning how to communicate with greater clarity, provide better feedback, ask better questions, and think more rigorously than the people around you.
The organizations that win won't necessarily have access to better AI. They will have leaders who have learned how to coach it effectively.
And in the process, they may rediscover what great coaching, great communication, and great leadership looked like all along.
