A Debate about Going From Knowing to Becoming: Leadership Micro-Experiments and AI

[00:00:00] Welcome to the debate. So, last year, corporations around the world spent literally billions on leadership training. And the result? Pretty much zero. Exactly. By almost every measurable metric, virtually zero lasting behavioral change. It's like, imagine sitting on dry land, reading this comprehensive, beautifully illustrated textbook on the biomechanics of swimming. Right, like studying the exact angles of the freestyle stroke.

[00:00:27] Yeah. The precise timing of the breath, the physics of water resistance. You could probably ace a written test on the subject, but then someone just shoves you into the deep end of the pool. And the water is freezing. The resistance is entirely different than your brain anticipated. Absolutely. Suddenly all that theoretical knowledge just evaporates. You're thrashing around trying not to drown. Because, you know, understanding the biomechanics of swimming doesn't actually make you a swimmer.

[00:00:52] No. It doesn't. Capability only develops when you're actually in the water, feeling the resistance, making choices, adapting in real time. Which is exactly the disconnect at the heart of the persistent crisis in leadership development. Right. We spend massive amounts of money teaching people the, uh, the biomechanics of psychological safety or strategic delegation. But it doesn't translate into action.

[00:01:15] Exactly. As Dr. Jonathan H. Westover outlines in his article, From Knowing to Becoming, this investment fails for that exact reason. Adam Grant's central premise here is that in today's era, agility matters far more than static ability or accumulated knowledge. Right. It's about how quickly you learn and adapt in the moment. Yeah. Grant argues that in an ideal world, every leader would run a new behavioral experiment every single week. Which brings us to the core of today's discussion. We have two converging trends trying to solve this crisis, right?

[00:01:45] The practice of micro experiments and the integration of artificial intelligence to support them. So the specific question we are tackling today is this. Does the integration of AI into micro experiments truly democratize and solve the leadership development crisis? Or does it merely automate the translation of knowledge while completely failing to address the deeper, you know, human cultural barriers to true transformation?

[00:02:06] I'll be taking the stance that AI enabled micro experiments represent a massive structural breakthrough that finally closes the knowing doing gap by scaling continuous learning directly into the daily flow of work. And my position is that while micro experiments are definitely valuable, relying on AI to drive them fundamentally overestimates what technology can actually do. Because true development relies on social sense making and psychological safety, which technology just cannot manufacture.

[00:02:34] Sure. So to kick this off, the reason this AI push is happening right now is because our current architectural model for leadership learning is just broken. Completely broken. We've been treating leadership like something you can cram for, right? We pull managers out of their actual jobs, stick them in a hotel conference room for three days, fill their heads with frameworks and send them back to their desks. Hoping their individual motivation will somehow change their daily habits. Right. We just cross our fingers. Yeah.

[00:02:59] But if you look at the widely cited 70-20-10 framework by Lombardo and Eichinger, it breaks down how humans actually learn at work. The on-the-job stuff. Exactly. 70% of a leader's learning happens on the job through tough assignments. 20% happens through relationships and feedback. And only 10% comes from formal training. Yet we have a multi-billion dollar industry entirely obsessed with that 10%. Because as you probably agree, it's the easiest thing to measure. Oh, absolutely. You can track attendance at a seminar.

[00:03:28] You cannot easily track a manager struggling through a really tough conversation on a random Thursday afternoon. Precisely. And Josh Burson's research shows the average employee only has about 24 minutes a week to devote to formal learning anyway. So if development requires stepping away from the work, it will always be bottlenecked. Right. We need a mechanism to operationalize experiential learning. That's 70%. And this is where AI comes in. For decades, academics have pointed to David Kolb's experiential learning cycle.

[00:03:56] Which sounds academic, but it's really just how humans naturally learn. Exactly. First, you have a concrete experience, like you bomb a presentation. Second, you reflect on it. You know, why did everyone look so bored? Why was everyone on their phones? Yeah. Third, you conceptualize. Ah, I spoke for 40 minutes straight without taking a single breath. And finally, active experimentation. Next time, I'll stop and ask a question every 10 minutes. We've always known that fourth step to active experimentation is where the magic happens. But we've never had a way to enforce it at scale.

[00:04:26] Okay, but you're framing this as a structural problem of instructional delivery. You're basically saying, if we just have the right scaffolding to enforce that active experimentation step, problem solved. Well, largely, yes. But I argue the true bottleneck isn't a lack of instructional scaffolding. Learning, especially leadership learning, is fundamentally a social and emotional process. It is not merely a cognitive or administrative one. It is absolutely administrative when you're talking about a global company with 10,000 managers.

[00:04:53] I mean, a human coach simply cannot hold 10,000 people accountable for running micro experiments every single week. Sure, the scale is hard. Right. So, AI perfectly scaffolds that leap from conceptual knowledge to daily action. Take a platform like Permios, which was developed by the Leading Through Institute. Okay. It translates a really fuzzy general insight like a, I need to delegate better, into a specific bounded micro experiment. It shifts the goal to a tangible test. But how does it do that in practice?

[00:05:21] It prompts the manager and says, in tomorrow's 10 a.m. project meeting, instead of assigning the next steps, ask the team to propose who should own each action item. Simply observe your own comfort level. So it gives them a script? Not a script. It gives the leader a precise, observable action step right in the floor of their actual day. Well, let's break Kolb down a little further. Because his model really isn't the end of the story here. Research by a scholar named Russ Vince expanded on Kolb's learning cycle.

[00:05:48] And I think this is crucial for anyone trying to understand why AI falls short. How so? Vince looked at Kolb's neat little four-step cycle and basically said, wait, where is the fear? Where is the anxiety? Ah, the emotional side. Exactly. He emphasized that learning in a group setting is fraught with emotion. When you try a new behavior, you are terrified of looking incompetent in front of your peers, or worse, your boss. Sure, it's scary. Right. So an AI can push a notification to a leader's phone, prompting them to try a new delegation tactic, sure.

[00:06:18] But it cannot manufacture the prerequisite for that experiment, which is a culture of psychological safety. I think you're treating psychological safety like a binary on-off switch. Like it has to perfectly exist before any experimentation can happen. It has to exist to some degree. But that's a trap. If we wait for perfect organizational cultures, literally no one will ever develop. AI actually helps build a leader's internal framing to take that risk in the first place, regardless of the macroculture. I just don't see how an app builds courage.

[00:06:47] Think of AI as a GPS for leadership. It turns the vague, intimidating destination of being a better leader into turn-by-turn directions. Okay, I am struggling with this idea of an AI GPS for leadership. Because when I use a GPS in my car, I'm just blindly following instructions. Turn left here. Merge right there. It's an analogy. But isn't that the exact opposite of what we want leaders to do? How does an AI feeding managers turn-by-turn directions prevent them from just becoming robots following a script?

[00:07:16] Because the AI isn't feeding them a script to read. It's feeding them a framework for curiosity. It's doing the heavy lifting of translation, which drastically lowers the cognitive barrier to entry. Do we have evidence that actually lowers that barrier, though? We actually do have data on this. Research from the Center for Creative Leadership showed that AI-powered conversational tools effectively shift leaders from framing problems as external organizational issues to personal growth opportunities. Using what kind of prompts?

[00:07:43] One of the most effective prompts they tested was simply the AI asking a leader who was facing a tough situation, what about this challenge will require you to grow? That single nudge delivered in the moment reframed the leader's entire approach. So the AI acts as the GPS by navigating the leader toward introspection, which makes the subsequent action planning incredibly rich. A GPS is a brilliant piece of technology for getting across the city. I'll give you that. But it is entirely useless if the driver is terrified of getting on the highway. Meaning the emotional stakes.

[00:08:13] Yes. In organizational life, the highway is paved with real emotional stakes. Let's ground this in reality for a second. If you're listening to this and you manage a team, you probably know exactly what this feels like. The pressure of a real Tuesday morning. Exactly. You go to the workshop, you get the AI prompt on your phone on Tuesday morning telling you to, let's say, let your team lead the brainstorming session. Right. But at 9am, a massive client crisis hits. Your brain just floods with cortisol. What do you do?

[00:08:40] You revert to command and control micromanaging because it's faster and it's what you know. But wait, that's exactly why the AI nudge is necessary. Without it, you definitely revert to micromanaging. The nudge doesn't overwrite the survival instinct if the environment is hostile though. Look at Amy Edmondson's work. She defines psychological safety as a shared belief that the team is safe for interpersonal risk taking. Which is fair, but… If your boss tacitly punishes failure or if your performance review only cares about flawless execution,

[00:09:09] no AI prompt asking, how will you grow, is going to convince you to risk an imperfect micro-experiment in a high stakes meeting. So you think culture always blocks the tool? Absolutely. Microsoft's cultural shift under Satya Nadella is a perfect example of this. He didn't just deploy a new technology platform to get managers to act differently. He explicitly championed Carol Dweck's growth mindset. Right, moving to a learn-it-all culture. Yes, he completely overhauled how performance was evaluated to move the entire company from a know-it-all culture to a learn-it-all culture.

[00:09:38] The technology was totally secondary to the emotional and cultural overhaul. I agree that Satya Nadella's transformation is the gold standard, absolutely. But I think your argument ignores the reality of scale and access for the other 99% of companies out there. What do you mean? Well, the historical alternative to AI for guiding these micro-experiments has been the executive coach. The kind of structural scaffolding required to design bounded experiments, rehearse difficult conversations, and reflect on the outcomes… It's expensive. It was a luxury.

[00:10:07] Limited to senior executives whose companies could afford to pay a coach $500 an hour. Which is exactly why the industry is so desperate for a cheaper alternative today. But it's not just about cheapness. It's about democratization. What AI does is provide just-in-time nudges and asynchronous support to frontline managers who historically got nothing but a slide deck. A very boring slide deck at that. Right? Oh. BCGU, that's the capability building arm of Boston Consulting Group, is redesigning their entire learning delivery around this exact principle.

[00:10:37] Okay? They are making applied learning the crux of the skill building journey right in the flow of work. By using AI? Yes. By using AI, you give every single leader, whether they are a CEO or like a first-time shift supervisor, the structural support to experiment within their specific micro-climate. We've literally never been able to do that before. Well, I think automated democratization is a very slippery concept, right?

[00:11:02] So, giving a frontline manager an AI chatbot is not democratizing the executive coaching experience. Why not? It's offering a facsimile of like the administrative part of coaching. So, let's look at McCall, Lombardo, and Morrison's foundational research on how successful executives actually develop. Okay. Okay? So, they found that genuine development stems from like navigating hardships and engaging in developmental relationships. And it requires friction. It requires human accountability.

[00:11:31] But you can still have accountability with AI tracking. Not the same kind. And this is why cohort-based accountability models like Center for Creative Leadership's Frontline Leader Impact or, you know what, their CCL boost, those are so critical to actual growth. But okay, those CCL cohort models you just mentioned, they explicitly integrate AI to make that cohort time more effective. They use it, yes, but as a secondary tool to prep the participants. The AI doesn't drive the transformation itself.

[00:12:01] It facilitates it. It's less like writing sheet music and more like a flight simulator. The AI can put you in the cockpit, it can simulate the turbulence, and it can run you through the checklist of what to do in the engine stalls. Which is incredibly useful preparation. Sure. But it cannot replicate the actual visceral fear of crashing a real plane with real passengers. You need the human element for that. I'm not saying you don't need the human element at all. The AI handles the initial problem formulation, sure.

[00:12:28] But the real sense-making happens when a leader has to look three of their peers in the eye and say, I tried this experiment, it failed, and I felt completely incompetent. But relying entirely on that cohort meeting leaves a massive gap in the learning process. That peer cohort meeting happens, what, on a Friday afternoon? Usually, yeah. What happens in the moment of friction on Tuesday? This is where AI moves us from episodic learning to a continuous learning system. By pinging them on their phone? By building what philosopher John Dewey called the continuity of experience.

[00:12:57] Dewey argued that every experience takes up something from prior experiences and modifies the quality of subsequent ones. Okay. By tracking trajectories and recurring patterns across dozens of micro experiments, AI helps leaders shift their core identity. They move from merely doing a leadership task differently to becoming a continuous learner. Wait, wait. You're assuming an algorithm tracking data points is enough to forge a new human identity. That completely ignores the psychological weight of how identity is actually formed.

[00:13:24] I don't think it ignores it. It augments it. How exactly does a software program tracking my inputs change who I am? By making the learning cycle inescapable and visible. The leading through institute models this with their no-feel-do-become cycle. Right, the traditional model. Well, traditional learning stops at no. You know you should listen more. AI pushes a leader through do by prompting the micro experiment, but critically it captures the reflection immediately after the event. Which a journal could also do.

[00:13:53] Yes, but over six months a leader might run 20 micro experiments. The AI system connects the dots in a way a human brain usually misses. Give me an example. It shows the leader. Hey, every time you experiment with asking your team for a solution instead of dictating it, your reflection nodes indicate a spike in your own stress that day, but a spike in your team's output that weak. Okay, that is an interesting pattern to see. Right? The AI holds up a mirror to a pattern the leader was just too busy to see. It ensures the no-feel-do-become cycle isn't left to chance.

[00:14:23] The leader builds the muscle of noticing, testing, reflecting, and adapting, until that process simply becomes who they are. I hear you, but Dr. Westover addresses this exact system in the source material, and he offers a very explicit warning about it. He does say we shouldn't rely on it entirely. He acknowledges the elegance of that continuous learning loop, yeah, but he insists that the AI is merely the connective tissue. The human remains the author of their development. Of course.

[00:14:48] If organizations over-index on the AI system, they risk losing the relational depth that actually transforms a leader's identity. But Westover is arguing for a human-AI partnership. He isn't saying we should throw the technology out. He's saying it needs to be paired with human insight. In a theoretical world, yes, a partnership sounds lovely. But in practice, organizations look for efficiencies. If a CFO sees that an AI can theoretically handle the reflection cycle and track the data... They'll cut the rest of the budget. Exactly.

[00:15:16] The immediate temptation is to cut the expensive human cohort programs entirely. And that is incredibly dangerous. It's a risk, I'll admit that. Westover specifically notes that AI cannot replace the vulnerability of sharing with peers or the emotional weight of a real leadership moment. Sure. Identity isn't shifted just because an algorithm pointed out a pattern in your anxiety levels. Identity shifts when you process that anxiety in a social context. With other humans. Right. When you realize your struggle is shared by the other managers in your department.

[00:15:46] And when you feel supported by a human system that values your long-term growth over your immediate performance. The AI ensures the story keeps being written, yes. But it is the human context that makes the story worth writing. Look, the organizational temptation to cut costs is very real. But that doesn't negate the power of the tool itself. It doesn't negate it, but it complicates it. When you look at the sheer scale of the knowing-doing gap, I mean, the millions of managers who currently have zero structural support for their development.

[00:16:14] We're just drowning in the deep end of the pool we talked about earlier. It's a lot of people. The AI system is a lifeline. By embedding scalable, highly structured learning directly into the daily flow of work, we translate vague insights into turn-by-turn behavioral tests. We are finally bridging the knowing-doing gap that has plagued this industry for decades. It's certainly a step. The mechanics of AI-enabled microexperiments are simply too powerful to dismiss as just connective tissue. I'm not dismissing it.

[00:16:41] The AI is an incredibly powerful administrative and translation tool. But true leadership development is a profoundly social endeavor. It is entirely dependent on organizational psychological safety and human connection. It needs both. An AI can prompt an experiment. It can track the data. And it can show you your own patterns. But only a culture that embraces failure and a peer group that offers genuine emotional support can give a leader the courage to actually run that experiment when the stakes are high.

[00:17:09] Well, one thing is undeniably clear from Westover's research, and it seems we both recognize this reality. The era of episodic, event-based workshop learning is dead. Completely dead. Sending leaders to a three-day hotel retreat and expecting them to return fundamentally transformed is a failed model. Agreed. The future of leadership development lies in continuous lived behavioral experimentation in the real flow of work. Absolutely.

[00:17:34] The challenge for organizations moving forward will be navigating the complexity of building this human-AI partnership. And they will have to build it very carefully. In our rush to scale development through cheap technology, we have to ensure we don't accidentally engineer out the messy emotional human friction that actually causes growth in the first place. So true. We invite our listeners to explore Dr. Westover's article from knowing to becoming to determine for themselves whether the future of leadership agility will be driven by the sophistication of our AI tools.

[00:18:03] Or the depth of our human relationships. Or, ideally, a delicate balance of both. It's a tension every modern organization is going to have to face. Because at the end of the day, you can study the biomechanics of swimming all you want. You can even have an AI perfectly map out the exact angle of your freestyle stroke. But eventually, you have to get in the water. So we can see here at the end of the day, the