The Psychology of AI Adoption: Why 95% of Pilots Fail with Dr. Gleb Tsipursky

Hey, everyone. Welcome back to Make It Happen Mondays, where we talk about sales, business, entrepreneurship, personal growth, mental health, and everything in between with guests who I truly respect and I think make a positive impact on the world around us. Today's guest is Dr. Gleb Tysperski. And I'll be honest with you, I butchered his name in the first 10 seconds of the recording, so we went with Dr. Gleb for the rest of it. He's written eight books, including two global bestsellers, which are Never Go With Your Gut and The Blind Spot Between Us. He's gotten something like 750 articles out in places like Harvard Business Review and Inc. Magazine. And his new one is the philosophy of AI adoption at work, peer reviewed, out with Georgetown University Press, built on over 100 consulting projects and 10,000 plus survey responses.

And here's why I wanted him on the show. I'm watching AI adoption right now in sales organizations be a disaster. Everybody buys the enterprise license, tells people to figure it out, and then wonders why nothing changes. Dr. Gleb told me something in this conversation that reorganized how I think about it. There's a study showing that 95% of AI pilots fail due to scale. And it's not because the pilots are bad. The pilots are great because you filled them with the people who already love AI. The failure happens when you try to roll it out to everybody else. And everybody else is made up of three different groups. He calls one, the AI alarmist, two, the pragmatic resistors, and three, the reluctant adopters.

The last group is the one that got me. Those are the people using AI in secret, getting their work done faster, and then hiding it because they're embarrassed. Shadow AI users. He said they spend the time they save hanging out on Facebook and listening to this podcast. So at least they've got some pretty good taste. We also got into the question I've been wrestling for months now, which is where the fundamentals are going to come from when AI can do all the entry-level grunt work better than any kid coming out of college ever could. His answer is that maybe 10% of what you and I ground through was actually informative, and each profession needs to figure out what that 10% is and teach only that.

We talk about whether college is even worth it anymore and what he'd tell my daughter, and then we get, unfortunately, a little dark at the end about where this whole thing is headed. I try to stay optimistic, but it's looking pretty bleak out there. And by the way, for the record, I wish I'd read his book before I tried to get my wife and daughter into AI about two years ago, because it would have saved me a lot of eye rolls. So I hope you enjoy this conversation as much as I did. Let's make it happen. Dr. Gleb, welcome to the Make It Happen Monday podcast, my friend. How are you doing today? I'm doing well, JP. Thank you for inviting me. Thanks for coming on, man. This is going to be an interesting conversation.

You've got a new book coming out, which I am very interested to talk to you about. It's about the psychology of AI adoption at work, because, man, I am right in the middle of that right now with the world of sales and tech and SaaS and watching it be just a disaster for the most part. So I'm actually looking to learn a little bit from you about how I can help my clients better integrate AI into their routines here. But before we do that, though, I always love to get some of the background here that you can't find on LinkedIn. And I know you are a very proud Ukrainian background. So I'd love to just learn a little bit more about kind of where you're coming from, like what you're growing up, your influences, like parents, those type of things, just kind of what made you you, and then we can get into your history of school and what you're doing now.

But talk to me a little bit about growing up and where you're coming from. Sure. Happy to. So my parents fortunately left the former Soviet Union. My dad is Ukrainian. My mom is Moldovan. They fortunately left the Soviet Union in 1991 when I was 10. And I grew up in the United States in New York City. So that's home for me. Now, in terms of what shaped my influence to do what I do, which is focusing on future of work adoption. So how people adopt the future of work. When I was 18, in 1999, I observed companies like Webvan, Pets.com, and so on being praised in the Wall Street Journal, Fortune, Forbes. And then only a couple of years later, they were criticized in the Wall Street Journal, Fortune, Forbes, when the tech boom turned into the tech bust, and all these companies on bankroll.

And so that made me realize that we really didn't know what we're talking about when we talk about company decision making. Even the top people in Fortune, Forbes, and so on, who are commenting on these, don't know what they're talking about. And so I decided to study this topic. How do people make good decisions in the future of work? I got a PhD at UNC Chapel Hill in the history of behavioral science. How do people make good decisions in historical and contemporary perspectives? I taught for a while at Ohio State of Decision Sciences Collaborative and History Department. And all this time, I also did consulting for companies. So that's been already working on this topic since 1999. So 27 years that I've been working and researching and learning about how do companies most effectively adapt to the future of work.

So now, you know, I was reading the Harvard Business Review on Fortune and Forbes in 1999. Now I regularly appear in them. I had an article just published in Harvard Business Review, for example. Just had an article come out in Inc. Magazine. I have something like 750 articles out in various venues. Then I wrote eight books now. So this is my eighth book. And if people heard me, they might have heard it. Never go with your gut. How pioneering leaders make the best decisions and avoid business disasters. Or the blind spots between us. How to overcome unconscious cognitive bias and build better relationships. Those are my two global bestsellers. The other books are good too, but not global bestsellers.

And so hopefully my new one is going to be a global bestseller. It's out with Georgetown University Press. It's a peer-reviewed book. It's based on over 100 consulting projects, something like 600 focus groups or focus groups, something of respondents, something like 10,000 plus surveys and 50 interviews with leaders from Microsoft, Uber, Experian, and so on. And that's the psychology of AI adoption at work, from resistance to results. Love it, man. You've been busy. Curious, how much do you use AI to do your public? Do you write all your own articles, all of those 700? Especially recently, have you used AI to write those? Or are you kind of a purist when it comes to writing? Because there's the adoption when it comes to...

Some people kind of dump their ideas, have AI structure it for them, and then tweak it on the back end. Others are, nope, I'm going to do all my own content. How do you use AI currently right now to write your own articles? So it depends on the venue's requirements. If a venue requires that I don't use AI, then I don't use AI. If a venue says that I can use AI, then I work with AI in a variety of ways. Help me research, help me look after I give to the initial ideas, help me draft initial ideas, help me spell check and edit. So it just depends on the venue and its requirements. Okay. So you're not kind of one or the other. You're just, you just use as, as directed or as able, right? But if you had your choice, I guess, if you were writing a new, um, a new article and you didn't have any limitations, right?

Nobody told you. Of course, of course I use AI. Why would I waste my time otherwise? But I guess, yes, well on research, right? But I think more of like the actual writing of the article. How do you say, same idea. Why would I waste my time if I don't have to, if the venue requests me to not to use AI, then I won't. And then I decide whether it's worth me spending extra time to, uh, write for that venue. But otherwise, why would I waste my time if I don't have to? It's like, well, am I going to do a long division if I don't have to, or am I going to use a calculator? I mean, of course I'm going to use a calculator. Am I dumb? Well, what's your take on the consumption of content produced by AI?

Cause I, cause I look at it these days where if I know AI, whether it's AI assisted or AI developed, you know, whether it's a video or content, doesn't matter. If I know it's AI, I'm actually kind of interested in it. But if I, if I, if there's a human who uses AI to pretend like it was their original thought and that type of stuff, and I kind of catch a whiff of it, I tend to discredit it a lot. That's my own personal view. Um, just because I don't like being, I, I'm, I think AI is like, Hey, I'm AI, right? Like, cool. Let's, let's learn. But if it's like a human, like, look at me and how smart I am, but I'm not using AI. That's where I tend to lose credit on that. Where's your take? Does it matter to you?

Well, I think if people say that they're not using AI and they're using AI, that's a problem, but it's not a problem because of the AI, it's a problem because they're not being straightforward. And so that's speaks to their, I guess, character and speaks to the, how do you see this? It would tend to undermine the quality, my perspective of the quality of the writing if I can't trust their character. So that's, so that for me is not about AI. That's about their character in the same way. Let's say somebody had a ghostwriter do their work and they don't disclose it. Then that kind of undermines the character. Or let me say this way. I don't judge people who use a ghostwriter ghostwriter to do their work as long as when they're asked whether they use a ghostwriter to do their work, they say that.

That's fine. That's fine. If somebody uses AI to do their work and they're asked and they say, no, I didn't use AI, that's a lie. Now, if somebody uses AI to do their writing and you don't ask and it's AI, then there's no problem because like you would, you would not, it's like the same idea. You know, it doesn't matter whether somebody used a calculator or long division, as long as they got the outcome and they're responsible for it. And so it's, it's a question of disclosure. If asked, if there's like an, if asked them that they didn't don't disclose or if they lie about it, that's the problem. Is it because I think true, like honing in on this, I wasn't going to jump to your at this point, but let's dive into it.

Cause I think there's a trust factor here with AI and just like, you know, I read a stat the other day that, you know, 50%, they actually compared AI to sales reps and we're about tied as far as what, what people believe coming out of our mouth and what people believe coming out of AI and how they do the research on AI, but then they ultimately want to talk to somebody to kind of validate some of the stuff that they learned. And I think that goes to adoption internal within the organization as well, right? You have a lot of people who are very nervous about AI and whether it's going to take their jobs and there's a trust factor here. So diving into kind of the psychology of AI adoption at work, how do you build, how do you introduce AI in a way that is trustworthy versus people thinking that it's going to take their jobs and all of a sudden, like from a leadership standpoint, how do you, how would you approach it to, to start building trust using AI so that people do adopt it and use it and, and are, and are interested versus avoiding it.

So let's talk about how AI adoption is different than other technology adoption. Let's talk about something like a sales enablement software, like let's say outreach, say like Salesforce or a CRM, like HubSpot or something like that. So stuff that sales folks use marketing folks use extensively. When you think about adopting that, what's the typical approach to that is a pretty top-down approach where you need to have top-down rollout and have clear SOPs where everyone uses it in the same way. And so everyone needs to use it in the same way, because if you don't put in clean client data into the software, you get that outcomes, you get dirty data, it's unreliable, right? So that's a problem.

So that's one thing that you, with a technology, typical technology adoption. And that's kind of how leaders are thinking about, well, this is the, this is how you adopt technology. Another thing about technology adoption is that Salesforce or HubSpot or whatever you use, I use ActiveCampaign for my company. It's hard to learn. There's a lot of new buttons to produce. You need to figure out how to adapt your workflows to. So it's a lot of hassle. It involves learning new things and it involves changing your habits. And so those are the things that leaders typically think about. So when they adopt AI, that's what they try to do. And that's what they kind of have a top-down approach. And they think that, well, it'll take people, it'll be difficult to learn and they'll need some time to do it.

They need to change their habits. We need to have a top-down approach. But that's not how AI is actually best adopted. That causes a lot of problems. So there's a recent study from MIT showing that 95% of AI pilots fail to scale. So when they fail to show return on investment from the investment that leaders are putting into. So that's a big problem. So that more companies are spending money, time, they're not showing return. So why is that? Well, because the approach that they're using is not correct. First of all, the emotional blockers are different. And second of all, what you need people to do to adopt it is different. So if you think about the top-down approach with SOPs, that just doesn't work.

AI is a very flexible tool. It can be used in very many different ways to essentially do all the workflows that sales folks do with their computer. Everything except having meetings, virtually or in person, can be done using an AI tool. Now, it doesn't mean it will be done as well. It doesn't mean it will be done as effectively. But it can be done. So what does that mean? So that means that you can't have just like an SOP procedure. You can't like describe all of your work. IT doesn't know what you do, doesn't know all the details. And you need to customize it to your own workflows. So it takes a lot more engagement from individual salespeople, marketing folks, to actually learn how to use an AI tool and then build it into their workflows.

That's very different. So people need to be much more engaged with AI tools than they are with typical tech tools. That's one. On the other hand, people are much less engaged with AI tools. Why? Because of the emotional blockers. The emotional blockers aren't, oh, it's a hassle to learn. It's a, you know, I need to change my habits. AI is very easy to learn. You can talk to it. You can ask it. How do I adapt? Here's my workflow. How do I adapt to it? Like, tell me. And it'll tell you. It'll guide you through it. Quadranthropic is great at that. And so you can talk to it. Literally, I can just turn on the voice option if you want to. It's great. But that's not where the emotional blockers are coming from.

There are three big emotional blockers that I see in companies. And the three big emotional blockers, the first one is fear. So that's the first and biggest emotional blocker. Sales folks, marketing folks, and others are worried that AI will replace their jobs. They're worried that by using AI, they're training their own replacement. So that's a huge blocker. I talk about that in Chapter 6 of my book, The Psychology of AI Adoption at Work. And specifically in Chapter 6 of my book. That's the psychographic profile called the AI alarmists. So that's the first one. Now, there's a second blocker. And that's a sense of identity. What's that about? Well, it's not threatening to the identity of salespeople to look up information about clients in salesforce versus to look it up in various Excel spreadsheets.

But it's a threat to their identity when an AI tool writes a great sales email, when it writes great marketing copy, when it creates a great client brief. So those are things that are serious threats to their identity, to the person they are, to who they feel proud of, to the things that are central, core to what they do as sales professionals, as marketing professionals, and all other roles with finance, with HR, operations. It applies to all of them, development, of course, thinking about engineers and programming, it applies to all of them. And so there are many people who get a sense of meaning from their work, and who are threatened to their sense of identity. And those are the people I call the pragmatic resistors.

So that's the second tech graphic profile associated with identity threat. That's what's most meaningful and important for them. And so that's the second group. Now, there's also a third group. The first two groups don't adopt AI. They're resistant to adopting AI. They're hostile to it. The third group does adopt AI, but they're reluctant to talk about it. I call them the reluctant adopters. What happens for them is a sense of shame. So the people feel socially stigmatized for using AI to draft their emails, to create marketing copy, to do their marketing briefs, to do financial analysis reports. People feel a sense of shame because they are socially stigmatized by their colleagues for doing so.

So it's a serious issue. And that group does use AI, but then they don't talk about it. They're shadow AI users. And so as a result, you have people who are using AI. They get their tasks done quicker, and then they spend their additional time, instead of contributing to company bottom line by doing more tasks, they spend it on hanging out on Facebook and listening to Make It Happen Monday's podcast. And that's not a waste of time. Might not be a waste of time, but not very productive for the company bottom line. And so that's why you see the lack of a return on investment, these three groups, when the pilots are scaled up. So then how do you approach it? If you're a leader right now where you're on your AI journey, right, and say you haven't adopted it fully, but you at least got everybody the enterprise version of Gemini or the enterprise version of ChatGPT or whatever, and everybody's still kind of doing their own thing with those personalities.

And then there's the other ones who are wide open about AI and very proud of it. So how do you start to lay the baseline for trust about how and why you're introducing AI into this equation versus just, hey, we got to do it, so figure it out along the way? Because that fundamental baseline of trust is critical on here. So how do you approach that from a leadership standpoint? Yeah, so thinking about the other personalities, there are certainly other personalities, but they're not, in many companies, they're not very prevalent. There was a recent few research surveys showing that something like 52% of respondents are more anxious about AI tools than they're excited about them. Only 9% are more excited than anxious, and the rest are equally excited and anxious.

So if you think about, and I call the people who are excited about the AI, I call them, there are two psychographic profiles there, the AI efficiency seekers who are excited about the efficiency you get from AI. And there are people who are excited about the AI evangelists, who are excited about the AI tools themselves. But they're not, they typically aren't very large within companies. So if you think about that 9%, who belongs to that 9%? The leaders of the company are excited about AI adoption, because obviously it can save them money. And then you have a small proportion of employees who are excited about AI. Now, certainly some companies have a larger proportion of efficiency seekers, AI efficiency seekers, and AI evangelists than other companies.

But what I tend to see, and again, looking at the research, 52% of all respondents are more anxious and excited. Those are the big free psychographic profiles that I talked about. And that's what happens with an AI pilot. When it gets rolled out, like when you start an AI pilot, the leadership team, let's say, starts an AI pilot. The people who are part of the AI pilot are going to be the efficiency seekers and the evangelists. But then the problem is when you try to scale it up and you run into the pragmatic resistors, the reluctant adopters, you run into the AI alarmists. So that's where you see the problems. How do you deal with it, as you ask? Like, how do you address the trust? Well, each of those groups needs a different approach.

So let's talk about each one in turn. For the AI alarms, the most are alarmed about their jobs. And they have reason to be. There's definitely research showing that AI is already causing some job loss for young people. So there's research out from Stanford, you know, research paper called Canaries in the Coal Mine, showing that young people in industries most exposed to AI, so people 22 to 25, they looked in June, 2025, they found that there is a 15% employment gap compared to young people in industries that are least exposed to AI. And so there's 15% worse employment prospects for young people in those industries that are most exposed to AI. They rerun the numbers for June, 2026, and they found that there's a 19% gap.

So increase year over year, and it will keep increasing, of course, it will hit the broader job market. There's a letter from over 200 economists, including 16 Nobel laureates, saying that we'll face major AI job displacements in the next few years. So what a leader needs to do is talk to his leadership team or her leadership team and have a concord, an agreement, a policy, that if AI makes us more efficient, more productive, we will not focus on job cutting, we will focus on growth. We will not cut costs for the personnel. We will instead focus on our additional resources on growing the company. And it's definitely doable. So there's additional research out from Stanford showing that companies that are AI forward in the same industry compared to companies that are AI backward are actually growing their top line 9% faster, revenue 9% faster, but they're growing their headcount by 6% faster.

So they're still growing their headcount, even if they're still, if they're using less resources to do the same things, 6% versus 9% to grow their top line, but they're still growing faster. Because they're taking market share away from companies that are not so AI forward. And so that's an incredible commitment that you can make that commitment. So that's one thing that you should do that I talk to leaders about doing. Now, the other thing that they should do is they need to communicate to these AI alarmists that, Hey, so what's happening right now is that we're not going to focus on, we're not going to cut people because of AI, but we're going to definitely keep people here who are the most productive, who are the most effective, who are the most efficient.

And just honestly, the reality is that those are the people who are using AI best. The people who are most engaged, so we're talking about being engaged with AI tools, the people who are figuring out how to integrate AI effectively into their workflows and be more productive, as in so. And so those are the people who are going to be kept around. If you can do that job without AI, great, be productive and effective. But, you know, you'll need to stay up into the wee hours realistically to do that, or you can use AI tools. So you need to flip the script for the AI alarmists. They need to not think that right now they're thinking that by using AI they're training their own replacement. Instead, they need to realize that they are going to be replaced if they don't use AI effectively.

And so that's the first group, that's the AI alarmists. Let's talk about the second group, the identity threat. Now, what they're currently threatened by is that the company is naturally focusing on automating the things that are most economically efficient, the biggest impact, most impactful things, writing the sales email, doing extensive research, you know, doing marketing copy, doing the tax preparation analysis and the actual tax report, the things that people feel are core to their identity. That's really important. So instead, what I do when I start working with a company is we do a survey on, hey, what are the things that you hate the most? What are the most frustrating things? What would you like to get off your plate?

So that's kind of one area. Plenty of things that everyone wants to get off their plate. Second area is what are the blockers, the things that you love? Where do you have to look across a whole bunch of sources to write the email to a client, to create a PowerPoint presentation to a client, to create great marketing copy, to create a great marketing brief? Those are the blockers to the things you love. And then I work to develop AI tools, prompts and agents that would help people actually do those things, automate those things, the things that they hate, and the things that block what they love. And then we do a training on that. And so that really helps get people who have experienced identity threat into using AI, because now they're addressing the things that are really important to them.

So for the third group, what you want to do, first of all, is leadership modeling. Leaders need to talk about how they're using AI extensively. All the ways that they're using AI, all the things that they're doing, their email writing, their performance evaluation, their profit and loss analysis. That's one. Second, the training. So the training that works for group two also works for group three, because you're expanding the norms, what's permissible, and you're addressing that social stigma. So that's two. And the third thing you need to do is create a channel where people can share about their AI accomplishments. There are cool AI tools and what they accomplished using them. And leaders need to positively reinforce this sharing.

They need to recognize people who share that and say, hey, great job, and so on. So that helps shift the social norms of AI usage and expand what's permissible and what's doable and address that sense of shame. So that's what leaders need to do for all three groups. So who's the personality that you tie into for these pilot groups? Like you said, if you use the AI evangelist and those type of people who are already bought into it, it's not going to translate to the other groups. So if you're choosing a smaller group to kind of test some AI functionality with, who's the best? Do you include somebody from each one of those profiles, or is there a certain set that you look at and say, I need this exact type of profile that's going to be able to connect with the people that are the doubters and the people that are afraid of it, and also the evangelists?

So those are two distinct things. So you definitely want the AI evangelists and the efficiency seekers to be the ones involved in the pilots. But before you roll it out to the rest of the company, you want to do the three things that I talked about. You want to address the AI alarmists, the pragmatic resistors, and the reluctant adopters. Because if you don't, that's where the gap will be. The pilot is, the things about pilot, so what did the study find? The pilots themselves are great. Like, you want to have the pilot survive if it wasn't great. The problem is the scale. The problem is the rollout. So it fails when people in companies, when they try to roll it out to the broader employee audience.

Because, again, you have the big gap, the emotional gap between the 9% and the 52%. So that's the emotional gap that needs to be spanned. And what my book is about is how do you span that emotional gap. There are plenty of people who love AI. I love AI. I'm an AI efficiency speaker. I'm not an evangelist. I understand it's words. But I'm definitely an efficiency speaker. I save myself tons of time by using AI. I probably got that from the conversation about writing and so many other things that I'm doing. But I understand that my intuitions, my gut reactions are not the right. Like, that's why one of my books is called Never Go With Your Gut. My gut is not the same as other people's guts. And my gut lies to me.

So just because I feel good about AI doesn't mean other people do. And that's a fundamental issue that so many leaders don't realize. Just because they're excited about AI tools doesn't mean that other people are excited about them. And just because an AI evangelist is excited about an AI tool like an employee doesn't mean that all of your employees will be AI evangelists. If you're working at a startup and you only hire people who are excited about AI, sure. But the regular folks, this is for the large majority of employees, they're not. And so this is what leaders need to bridge. This is the gap that my book is addressing. I love it. I wish I'd read it before I tried to introduce it to my daughter and my wife.

Because about two years ago, I was like, oh, my God, this is incredible. And they were both like, no, absolutely not. And it's been about two years for me to get them on track here. But I want to go back to the hiring piece, especially since with your experience at the universities and such. My fear is, you know, you talked about 15%, 19%. You know, I just hit 50, right? 50 years old. All the grunt work that I did coming out of college, right? Like the hand-to-hand combat, basically. It gave me the experience. It gave me the business acumen. It gave me, you know, the scars, if you will, so that I can be in a situation where I don't have to use a script. I can, you know, I'm pretty dynamic with conversations, whatever.

But now, pick a profession. Almost any aspect of that profession can be done better as an entry level, at least for now, and even better moving forward. But the entry level component of it is it can be done better, faster, cheaper than almost any individual kid coming out of school. So, with that, two things. One is, and this is the question, you're the best person I think I've had on my podcast to be able to answer this. One is, where do the fundamentals come from anymore, like when that scenario? But even more importantly, what are the fundamentals anymore? Because I think we're in this transition phase where I, like, we can still go back to core fundamentals to understand the output of the why and all that stuff.

But I think in a very short period of time, in the next couple of years here, we're going to have to fundamentally shift the way that we teach people when knowledge is a commodity, right? And I think it's different than the calculator and long division because this is just so, like, any information as long as you know how to use it the right way. So, where are we right now and how do we teach this new generation to be able to leverage this tool without outsourcing their thinking to it? Great question, JB. And I talk about that in my book under the concept of metacompetence. So, what's that about? So, when you think about, let's say, the time when we switched from doing long division to using a calculator, certainly people lost a number of skills.

They lost some abilities to, you know, grind out the math and learn how to use it. But they gained other skills that they can gain more complex thinking and evaluation of the results of their calculation that they can do using calculators that they previously had to do using long division. And there were concerns about using calculators. And so, what needs to happen, and this is something that I work with companies on when we talk about the psychology of AI adoption work, is changing the apprenticeship path. Changing what you talked about and that grinding out. A bunch of what you did when you ground, doing grinding out was wasted. Like, it's not useful to do a bunch of grinding out. But a significant aspect of the grinding out is learning.

Like, there were a number of things that you correctly pointed out that you learned. So, what I talk about in the book, and I talk about that in chapter 13 of my book, about how each profession needs to identify what aspects of that apprenticeship path are actually valuable and useful. And what aspects are just useless grind, you know, recopying things that were copied and copied again, but don't really need to be done. So, there are plenty of things that you don't need to do. Like, let's say, if you look at education, do students really need to spend all that time in education memorizing things and so on, when they can instead look them up on Google or ask ChatGPT or ask Claude Dropout to do that?

No, they don't. Our current education system is broken in many ways. Instead, people need to be taught to think. They need to be taught to use AI tools. They need to be taught how to use digital information. And then they need to focus on much more complex things in school than they currently are. And similarly, many apprenticeship paths into the job need to be changed. So, the grind part of it needs to be taken out. Instead, what needs to happen is that each profession needs to focus on what is the actual key aspects of the profession that people need to learn and recognize. And practice to get to the level of where you are and where I am. What do they actually need to do? What do they actually need to learn?

And so, they need to practice a number of things, certainly, that would get them to where they are. But those might be not the things that are involved in grinding out that, like, it's going to be 10% of what's currently involved, that are actually conducive to learning. What they need to learn is recognizing the outputs that are high quality so that they can use AI tools and then recognize when is the output good enough and whether it's not good enough. And then how to edit the output. And so, that is the key skill of using AI tools in tasks. Then they need to be able to understand how to create workflows and manage those workflows. And that's what an AI agent is. An AI agent is a replacement for a holistic workflow.

And people need to be taught how to manage that agent, see where it goes wrong, and make sure that they can have control of all of the aspects of the workflow, have accountability for all of the aspects of the workflow, including each of the outputs that it creates. So, those are the skills of the future. Those are the things that people need to recognize. And in order to get that, you need to only do 10% of the work that you and I put into to get to where we are right now. But how do you develop judgment without experience? So, you recreate experience. That's what I'm saying. The 10% that actually matters. That's going to be the things that are formative, right? Not all of your experiences, like how many emails that you write that are just completely wasted versus the significant emails where you can learn the nuance.

How many conversations that you have that were wasted versus the significant formative conversations, right? So, that is the key. That's what I'm talking about, the 10%. The formative ones. Each profession needs to figure out what's formative for it. And then that's what it needs to teach. So, I'll give you an example, right? When you're writing an email to get a meeting with somebody, right? You talk about trigger events and, you know, somebody opened up a new office, somebody launched a new product, all that fun stuff. And the merger and acquisition, right? That's the one that a lot of people, oh, you went through a merger and acquisition. Let's talk. AI is going to be able to write a beautiful merger and acquisition template, especially to the uneducated, okay?

And it's going to sound fantastic. And so, some kid, 22-year-old kid, is going to use AI to find a target account, write an email, like use it, it'll write this beautiful email. And because the rep has zero context of what it actually means to go through a merger and acquisition, they're going to look at that email and say, holy shit, that's better than I could have ever written. So, copy, paste, sent. And when that email hits somebody like my desk, who's been through a merger and acquisition, right? Or say I'm even going through a merger and acquisition. It's going to land. And depending on the stage of the merger and acquisition, what my role is in that merger and acquisition, it's either going to land like a fart in charge because I'm going to be like, do you have your mind?

Like, I'm actually worried about my job and getting fired, right? Or even worse, it's going to land. And then I'm going to say, yep, you know what? Let's set up this call. And I'm going to get on the phone with some 22-year-old kid and be like, yeah, talk to me about how you can help somebody like me through this merger and acquisition. And they're going to go, uh, uh, uh, uh. So, how does that kid, from a training standpoint, learn about to be able to judge that email that AI is writing for it, to say, this isn't the right time for that? Or, you know, like, what do I do if it does work? How do we address that? Right. So, previously, what would have happened is that the kid would have had the role of doing all the research.

Um, to figure out all the context in the merger and acquisition, right? That, that would have been their role. Then the sales, the senior sales rep would have actually written the email and made the decision to do that, right? So, the kid is wasting his time, um, doing all the research. So, what needs to happen is that AI is replacing, so, for that aspect of things, just walk through that scenario. The AI will do all the research. The senior sales rep will still make the decision. And the senior sales rep will explain to the kid why he's making the decision, what's happening, why is it happening at this stage, and so on. And so, the kid learns. That's what I'm talking about, the difference between the 100%, like, the grind, the doing the research to, like, foil the context, versus the 10%, which is the learning about why, the considerations, and so on, about the merger and acquisition.

So, that would be for that workflow. And that applies to all the workflows, HR, operations, like, anything, like, marketing, like, anything that you can think of. It's 10% of what we need, of what we learn. The formative part is not the research that the kid was doing. The formative part is the kid observing the senior rep doing that email, right? And eventually, playing, like, going to the meeting and shadowing the senior rep and, you know, holding her briefcase when the senior rep is doing the presentation, right? Like, that's kind of the phrase holding the briefcase. And so, that is the part that is formative. So, each profession needs to figure out what's formative and focus on that. Because I, you know, with all this, I am insanely optimistic for people who are curious, who have some business acumen behind them, and who have experience, and are creators, right?

Because now, the world is yours. There's no limits. If you can think it, you can do it. You can make it work. And if you're curious, you don't take what it gives you, you question it, source check, give me an example, whatever. But I think we're living in Pareto's principle right now, and I'm going to be very generous with Pareto's principle as far as the 80-20 rule. I think there's only about 20% of our population across the board that is that curious, that wants to do it. So, the other 80%, like kind of the B players, I always used to think, you know, the distribution of talent, you got your A, Bs, and Cs, right? The Cs were tolerated previously in our world because whatever, we just need them.

The Bs was always the focus to try to shift them to A's, and the A's were always going to be the A's. In this world moving forward, my gut tells me that, and I was probably wrong, that's why I'm asking you, but the Cs are irrelevant because they're getting, those are the ones getting replaced, and that's where the layoffs come because that was dead weight anyways. The Bs, though, that big middle part, to me, there's two sections of that middle part. The one that is the B players who are always going to be the B players that just want to do their job, right? They just want to, the nine to five, they just want to do their job, they don't want to cause any waves. And then there's the upper echelon of that middle percent that wants to get better and wants to be an A, right?

The ones I'm most concerned with is the kind of the fat part of that circle, those B players. So, how... Nine to five, yeah. Yeah, exactly. So, how do we, you know, how do we address that as a society? I mean, forget about sales here for a second. How do you address, like, how do we retrain these people to not get replaced by AI? Because I look at a lot of function. I mean, quite frankly, I look at a lot of sales reps right now and I tell them that they're almost irrelevant in the process moving forward because I'm not worried about all the AI tools that are coming for sales. I'm worried about when the client wakes up and realizes, why am I talking to you when I could just ask ChatGPT and it gives me far better answers at a moment's notice.

And so, what do we do with this middle group of people who, I mean, you know the stats. I think still only, like, 8% of our population here in the United States even uses ChatGPT, like, the free version of it. And there's, like, another 92%. And I think there's a wave coming for that group of, like, for people who aren't paying attention. And it's going to hurt and it's going to hurt bad. And I don't see how we're not headed towards universal income and all these different things to subsidize that. So, where are you on this? So, this is where I'm talking, like, what we can control, like, is within each company, right? So, within each company, what we can control, like, we can talk separately, let's talk about the broader societal level.

So, within each company, when you're hiring people, like, the people who are going to be, A, are already the efficiency seekers, right? They are already the AI evangelists. And you can, one approach is to hire those people. That's going to be very expensive because that is going to be incredibly expensive for a company. So, realistically, what you want to do is move the people who are the AI alarmists, who are the pragmatic resistors, people who are the reluctant adopters, you want to address those people. And, you know, that's kind of one for the people who are there. So, you're talking about the people who are there. For the young people who you're hiring, you need to focus on how you're going to change the apprenticeship path.

And so, the apprenticeship path is exactly what we're talking about. And so, if you think about, let's say, the structure of education, so just, because this is about education, like, this is about people learning, right? There's going to be, if we imagine a world without college, without school, there's still going to be plenty of people, the A's, who would aspire to get an education. There will be still books around. We're talking about the world of, like, the, you know, the 1800, right? The world in 1800. There's, you know, there's no public education. People who are A's would still get educated. Find a way to get educated, yep. Exactly. The people who are B's, they'll just be farmers or, like, you know, shoemakers or something like that.

So, how do you get the B's into education? Well, you create a system. You create structure. You create public education. You create a way for them to do so. And this is what I'm talking about. So, how do we change education to make sure that these people get educated into what they need to do? Now, we can't, we can talk about the social, broader social level. But what leaders can control, sales staff, sales leaders, CEOs, executives. What they can control is what happens when a company. What their domain, right? And this is a very realistic path within their domain is to do exactly what we talked through with the sales staff. So, not have the grind part, not, like, have people go through the same process that they went through.

But figure out what is the formative part of it. And we identified, like, you and I talked through, like, I'm not a, you know, you are much more of a sales expert than I am. But that seems like the formative part of sales, right? So, we just talked it through here. I'm sure smart people can talk it through, like, you know, I'm not a genius, you know, maybe you're a genius. No, definitely not. This is something that's very doable. This is something that you can talk through. You can figure out what is the formative part of legal work. What is the formative part of it? Like, that's what I'm saying is you need to figure out the formative part of each profession, which is not going to be more than 10% of what it used to be.

And then focus on conveying that 10% and training the young professionals in that 10%. And that's how you'll get people who are currently in the world of bees. That's how they can survive in the world of AI disruption. Okay. Do you have kids? No. No. If you did have a kid, like, my daughter's 15. She's got two more years in high school. I'm wondering if college is even going to be a thing here in two years, quite frankly. But if you were looking at colleges for a kid right now, what would be the... Because I've always had a problem with our traditional education system in the first place. I think going $200,000 to $300,000 into debt for a $50,000 a year job is absurd. Now it's even more ridiculous knowing that that $50,000 a year job probably won't even be there in four years.

So what would be the characteristics of a college that you would look for to, if the kid was to go to college, right? Other than, you know, entrepreneurship and the things that I could do, you know, I could control. You know, in a structured learning environment, what are some of the characteristics of what colleges need to be doing right now, in your opinion, to be putting these kids in a position to be successful coming out of whatever four-year degree, depending on what the opportunity is? Realistically, if I had a kid, I'd recommend that she or he go somewhere cheap, local, if they want a degree that is... where their job is going to be dependent on degree. So some regulated profession, I don't know, if they want to go into teaching or social work.

Now, those are professions that will still be around because they're regulated. And there's, you know, guidance from the government that, you know, you need to have a teacher to teach you, you know, even though AI can do a great job teaching, right? So if you want a job like that, that's regulated, like nursing, I don't know, go somewhere cheap and don't load up with that. That would be my advice. If you don't have a job that's dependent on a degree, if you want to go into business, for example, don't go to college. You can learn all the things that through AI that you would have learned in college. So as long as you are sufficiently, how do I say this? Can sufficiently take initiative for your own learning, ask AI, have it mapped out the roadmap for you.

Read the books and, you know, the advices or have the AI summarize those in an effective way that you learn and give you tutoring and guidance to get that. You can, you can do that. Like there's no need to go to college. If you're not going for a regulated degree, if you're not going for a job that requires a college degree, I would not recommend going to college. If you are going to that job, look for somewhere cheap. Yeah, that's kind of where my head's at right now. Who knows in two years? Let's finish up on this one because I know we're coming up on time here. What is your thought? Let's macro out here, right? The anthropic security guy who just left and says, you know, there's a greater 10% chance that we're all going to die.

Where's your head with that? Because I'm a little, I'm very concerned, obviously. I personally do not think that our government, our government works way too slow. We're not going to collaborate with China on some type of regulations like we did with nuclear war. And by the way, nuclear war, the only reason that those regulations came in was because of the disaster that it caused and what we saw it do. And so are we waiting for a disaster to happen in order to have to regulate this? And if that's the case, is it going to be too late? Well, to be clear, what happened is not that the guy who left said that there's a 10% chance. It's the guy who stayed. It's the head of, it's the head of anthropic research.

And so that's very important that it's not someone leaving and saying, oh, it's like the company itself. Like the company, high-level leader of the company itself believes that there's over a 10% chance that in the next decade, AI will kill itself. And, you know, he's still working on advancing AI. It's like, come on. Yeah. Some people have weird motivations. So, you know, I can't speak to people's, you know, I wrote the book on adopting and the psychology of it, but I certainly would be working on advancing AI models if I thought that there's an over 10% chance. And he is working on that and he thinks that. So, you know, in terms of what I think about it, I think the recent hacking of hugging face by open AI models is very concerning because open AI specifically forbid their models from doing so, forbid them from getting access to the internet.

And their models still found a way to access the internet, still found a way to hack hugging face. So that really does not bode well for the top companies being able to control their AI tools, which is what the concern is. And so that's why the person who left, you know, left and made that statement and the company clearly believes, you know, that that is a concern. At least Anthropic is vocal about it, which is one of the reasons I tell people that for using AI, you should definitely use Claude. So that's the most ethical company, they have the strongest guardrails and open AI is not nearly as open about things that clearly people inside of belief. And clearly their models are very, not, not as well behaved as the Claude Anthropic ones are.

Yeah. And then forget about Gronk. Anthropic has some problems. Yes, of course. Forget about Gronk. So it's very concerning. And I think you're right. Like the U.S. government is not doing well in regulations. It's not moving well. So it's, it's very worrisome what will happen in the next 10 years. And to the extent that, you know, folks have an impact on their politicians, they should be definitely giving their politicians to focus much more on regulating AI when these things are happening. And I, that's the, that's the challenge that I have right now. It's like, it almost feels inevitable, right? Like everybody's well, we got to win, win what artificial super intelligence. Like, I don't know.

There's a, I don't know of a single species that controls a species that's smarter than it. You know what I mean? And other than actually there is, there is one and it's, and it's a child. So the only hope, like, you know, a baby is obviously not as smart and not as, as capable as, as, as the parents. And that baby still dictates what those parents do for the most part. So I think our only hope is, is if AI actually looks at us as its children and, and there's something in its quote unquote DNA that says, okay, like we're, we're going to be here for good. But I would, the people who are running this, if they had any empathy at all, I'd feel at least somewhat comfortable and it doesn't feel like they do.

You know, cause when you ask any one of those, those leaders, where does humanity fit in this equation in the future? They all pause before they answer, like the, the easiest answer in the world, should humans exist in the future? And every one of these psychos pauses before they answer the question. And that terrifies me. And I, but I feel like as a, like, but as a, as a person who uses it and who, who sees some of the benefits, it's like, I feel like I need to do something. I don't know what to do, call my representative, but it's this inevitability that we're all kind of lemmings going for the, the cliff that we're all walking off of here. And I, I don't know what to do about that, quite frankly.

Yeah, I hear you. So I think there is a difference though, between other companies on Anthropic. So Anthropic, Dario Amadei has been public about the dangers of AI. And so Anthropic is like the one company that is actually talking about these issues and they're trying. So I wouldn't say all of them, and Anthropic is like one of the two major labs in, in this space. So what I recommend that people do is use, spend their dollars on the ethical company and just hope, like one of the reasons that Anthropic, Anthropic was founded by people who left open AI. And with their mission is to build open, build AI faster so that they can create those sorts of constraints. They're having a constitutional approach and they're saying, well, we want to try to build AI that doesn't kill everyone that cares about humans in the way that, you know, you care about children.

And they're trying that exactly, and they're trying that, but they are not being blind and publicly speaking about that, hey, they might not do it. So their motivation is like, we're racing toward, their public statements is we're racing toward AI because we think we are the ones who can build it in a way that it doesn't kill everyone. But we still think that there's a serious risk. I'm not saying that they're right, that they'll be able to succeed, but they do talk about this. So I just want to be fair to, like, Anthropic about, like, where are they at? Well, that's the reason, I'll tell you the main reason I went from ChatGPT to Anthropic was because of their U.S. government contract, period.

Yeah. That shift made me, like, I was like, okay, at least they have some type of ethics over here. Like, I don't know what they are, but they're better than open AI. Yes. Awesome. Well, look, I got a million more questions, but let's leave it for the book. So, Doctor, tell people where they can find out more information about you, the book, and everything else so that they can find out and get this to be a bestseller for you. Well, if they want to connect with me, they can do it on LinkedIn, but make sure you tell me you heard it on Make It Happen Mondays because I don't accept people who just connect with me on LinkedIn. There's a lot of LinkedIn spam. You want to get the book? Go to Amazon, Barnes & Noble, your local bookstore.

It's traditionally published, Georgetown University Press, peer-reviewed, so it's available everywhere. If you want to get a free sample, go to my website, DisasterAvoidanceExperts.com forward slash AI book. Again, that's DisasterAvoidanceExperts.com forward slash AI book. Awesome. And I'll put those all in the show notes. So, Dr. Gleb, thank you so much for coming on, man. I appreciate it, and I appreciate the work you're doing out there. Thank you, JB. Have a great day. Absolutely. And everybody, hopefully you enjoyed the conversation as much as I did. And like I said at the end of all these podcasts, go out there and make somebody smile today. Because no matter how bad your day is going or how bad you think it went, if you make somebody smile today, you know you had a good day.

And the world means a lot more of that right now. So, thank you all very much, and I will see you on the other side. Hey, thanks again for listening to the show and for all your support throughout the years. This is one last reminder to check out my training programs at JBarrows.com or leave a five-star review on your favorite podcast platform or G2 if you've already gone through my training or sign up for my newsletter. Any of those will help ensure I can continue to bring you free content and great conversations like the ones you just heard without having to annoy you with sponsors. Again, let's make it happen in 2026 together. We'll see you next time.