Episode 15: Will Your Next Boss Be An AI Agent?

Brian Milner (00:23) Well, welcome in everyone. Welcome back for another new episode of People over Prompts Podcast. I'm here with you as always, your host, Brian Milner. And I am here to talk to you about the future of teamwork in the age of AI. All things having to do with AI and humans working together. What does that look like? How is it going to change how we work together? We're trying to explore that from a lot of different angles here on this podcast. And we have one that we haven't really explored. Yet, which I when I saw the title and saw the the the kind of focus area, I thought, wow, it's amazing that we haven't really gone to that topic yet. We talked about whether AI is going to replace a lot of low-level jobs and entry-level kind of work, but what about the opposite end of the spectrum? So we titled this Will Your Next Boss Be an AI Agent? subtitle or what I've been thinking kind of behind the scenes is you know is AI coming for your boss's job? we hear a lot about it coming for the lower level jobs and for the programming jobs and for the mid-level kind of people but what about the managers? What about the leaders? it's an interesting question. There's a lot to unpack here. And as always, we try to explore this from different angles. So I I want to start with a Quinny Piac survey that that actually came out from March of 2026, this year, at the time of this recording. And they asked people if they were how they'd feel about having an AI program as a direct supervisor. by the way, what they they meant by that was assigning task, setting schedules, not necessarily the hiring, firing, or employee development, but day in, day out, here's here's your next thing to do and coordination, you know, that functions of a manager. And this was I I think it was about fourteen hundred adults in the US. So not worldwide, but just in the US. It found that fifteen percent were willing. Only fifteen percent were willing to have AI as a direct supervisor. Eighty percent checked the box for unwilling. Eighty percent said nope, that's the line I'm gonna draw, I'm out, if that's the case. I saw it kind of split up. It was a little bit more for millennials were more were okay more okay with it, whereas Gen X was a little bit less okay with it. but average right there in the middle, about fifteen. Neither neither side was really gung-ho about it. I think the millennials were about 19% and Gen X was maybe 12% or something like that. so it's it's interesting to see that division. It's interesting to see that just from way we feel about it, people aren't really willing at this time to do that. They aren't willing to have AI as a boss. There there's some conversation around this as well that I think i is valid and important to take into consideration here. there's one particular article that was in Forbes that was called What Happens When AI Starts Assigning Work to Humans? And it talked to in that article's kind of background, the CEO of Lindy, name is Flo Crivello, and The the quote here was, I don't think middle management should keep existing. And Fravello kind of argued there in that article for sort of what it what was called a sandwich approach. That humans were were kind of on the top setting goals, and AI was was kind of in the middle layer of the sandwich. They were coordinating. And then on the bottom layer were humans that were actually carrying out the work. Now, that's kind of interesting to me that even in that scenario, we're setting a dynamic where the the lower level work, the the kind of grunt work is still being done by humans. that's not been what I've seen. I've seen a lot more organizations that try to replace that layer with with AI agents. and again, I I think we have to even pause there and check our language because it's not really replacing any person or any person's role or responsibilities. It's it's replacing certain aspects, certain tasks. And that's the thing that I think is going to be important here as we talk about managers as well. So i i it does bring to mind an important question then when we think back to that survey as to to why why do you think people wouldn't want AI as a boss? Or if you're listening to this and thinking, yeah, I'm in that campus well, right, that's probably the majority of you or the vast majority of you. Why is that? What what precisely about having an AI as a manager in the sense of a signing out and coordination, would bother you about that? I think it's important for us to kind of pinpoint those kind of things. would your answer change if your current manager was bad? Was ineffective? Did a poor job? Would you be more open to it in that kind of scenario? you know, we hear over and over again that people don't leave jobs, they leave managers. So If that's one of the number one sources of people's frustrations with work, if you're one of those people who are frustrated with your manager, would that influence your decision? I think that's an that's kind of important thing to think about. how much of management is actually about managing people versus managing information? And and that's that's part of what we're gonna explore throughout the rest of this. So let let's talk about what a manager actually does, because a manager is not just one thing. And depending on what a manager is managing, obviously, they're gonna have different responsibilities and different things that they do. there's there's a particular study that kind of broke this down into different tasks. It's the study's called Algorithmic Algorithmic Management in Scientific Research. And this was published in 2024. I'm just gonna cut I'm gonna use their definitions here to kind of make this easier to discuss. So they divide it up into five core functions for management. First, task allocation, breaking down the work, matching tasks to people's capabilities. That's the first. The second was direction. And they mean by that setting expectations, priorities, timing, and even performance standards. That's kind of the direction Bucket. The third bucket would be coordination, managing dependencies, integrating people's contributions, right? Then there's motivation. Motivation is encouraging effort, commitment, and sustained performance. And then the last one that they they kind of called out was supporting learning. And they they said that that kind of was represented by giving feedback. Coaching employees, skill development within employees, and just general growth of the employee within the organization. all those things are are kind of how they divided out the different functions. I I I think there's maybe some additional things that we might want to consider as well. So I'm I'm throwing some additional specifics on top of that that weren't necessarily covered like hiring and firing. That's that's something important to consider. They had performance standards, but what about performance evaluations? That's an important function, I think, a lot of times for a manager. Employee advocacy. I just was talking to someone the other day is talking about how they had a terrible manager because the manager just kind of checked in and out of work every day and they they never advocated for any of their employees to be promoted. Whereas other parallel managers did advocate for their employees to be in the promoted. So that's an important aspect. Resolving ethical dilemmas, organizational culture, and trying to set that culture. And just generalized accountability. You know, if I am a software development manager, I have accountability for the overall work of the people that are working for me. So I think accountability is increasingly an important thing to shine a spotlight on in this era of AI. Because remember, as we've said in previous episodes, AI is not accountable. we see this even legally in scenarios. There's a hypothetical situation I just saw that's not really hypothetical, where someone was chatting with a chatbot and the chatbot told this person there was an airline and the chatbot said the person was trying to figure out about a bereavement fare. And the chatbot said the chatbot said, Go ahead, book the fair, and you'll you can be reimbursed after. And then afterwards found out that the person couldn't be reimbursed and the person sued because they said, Hey, hey, look, I I followed the advice of your representative. Your your chatbot was your representative. And personally I agree with that. I I think that that's that's gotta be something we consider from a customer service standpoint is if you've ever been in that scenario, I've actually had that scenario as well, not this exact scenario, but I've had scenarios where I've interacted with a chat bot about a return and it said, yes, no problem, we'll do this and then it'll be taken off your account and then got an email from the company saying, No, sorry that it said that to you, but we we're not returning your your money and you cannot return this product. So the That seems to be a an ethical line, a gray line that's starting to emerge of are you responsible for what your chatbot says? and in our workplaces, in our our organizations, where does the accountability fall? It's gonna fall to a human somewhere. So is it the programmer? Is it the manager of the programmer? Is it the leader who made the decision to install the chatbot and put the chatbot in place? That's a really tricky question. And it's probably different organization by organization and situation by situation. So if we think about management then as not one thing, but as a conglomeration of different functions, then I'm gonna have to come back to kind of a central thesis of this podcast in general, which is maybe what we're doing. Wrong in all these scenarios is we're saying, here's our current system. Our current system involves a middle manager. So can AI swap out that human and take the place of the human rather than taking a a larger scaled back view of the entire system and saying, if we were designing this from scratch today. With AI, would that role be needed? Or would that the responsibilities, the things that that role does, would it be distributed in other ways? I'm not proposing in any way that we fire all managers, although someone did just point out to me the other day some research that Adam Grant called to light recently about a company where they did away with their management structure. So everyone didn't report to an immediate manager. And they did it as an experiment and then found that it worked so well, they continued to do it after the experiment. And that was back in I think 2019. So it was prior to the current proliferation of AI in our companies. so if that that did work and can work in certain scenarios, isn't it logical for us to consider today what we actually need? Right? What functions, like we just broke out, what functions are still needed, and maybe even more importantly, what functions can AI take and what functions are more suited for? a human, that a human st we would still want a human to do certain things. So let's talk about where AI might be better. Right? Which of those functions and which of the things might be better from an AI standpoint. well when we think about AI, I think it's important to think about their i its potential strengths. It it's got a strength and speed, it does things faster, it it has more capacity. So it can handle multiple things at once. it's consistent. relatively consistent. Large language models don't give you the same answer every time, but that's because that's not how it's generated. Right? It's it's a predicta predictability calculation that's being made. So, but it's relatively consistent. AI has broad information access. It can it can know a lot of things. It can have access to a lot of information. Right? it can coordinate multiple things simultaneously because it can be in lots of places at once. It can actually have eyes on things, virtual eyes, right? that it can monitor the progress of things. So it can coordinate multiple things simultaneously. And it it it is particularly good at identifying patterns. And these are patterns that oftentimes humans might overlook. So I know there's a lot of times that I will feed research into AI to try to have it help me to find patterns. Here's a bunch of data. What is the pattern within this data? Here's a bunch of you know velocity charts from my team, or or here's measurements of the team's performance. How does that reflect you know what patterns, what trends are we seeing, what should I be paying attention to? Right? Those are kind of things that AI can pay better attention to a lot of times than we can as humans. there there's there's one particular study that I I found that was interesting to this. It's called Can an AI agent lead human teams? By the way, I'm gonna put all the links to these in our show notes. So if you are a research nerd like me and want to go dive deep into these things, I'll give you the research papers. You can go into it yourself. and you know, side note there, I highly encourage that approach. When we're dealing with these kinds of topics. I I I can make up stats and tell you anything. But if you can read the research yourself, then you can say, hey, you know what, Brian actually got that wrong. you know, he said this research paper said this, but that's not the the same thing that the actual paper says, or or you know, that's not the conclusion I I draw from that research. that's a really important critical thinking skill, I think, in today's world. So anyway, back to our our our study here. The can can an AI agent lead human teams? they did two experiments. It was a smaller study, 153 people, and they had 51 three-person teams. Now, this is a little bit of an interesting experiment because they used multiplayer games. to to kind of measure this. So three person teams in multiplayer games, they they they the participants here competed a collaborative search and containment task within this multiplayer game. And the teams received directions from an expert human a rules-based AI or an LLM-based AI. So they broke it into three different sets of teams. The human-led teams performed better overall. The LLM led teams showed competitive performance on one measure but struggled in difficult visibility conditions. So when there was a fog of war type thing, the AI struggled more to give directions in that area. important to kind of understand, a little side note here from this study. What they did was they used voice commands. So imagine that you're kind of on the ground in a multiplayer game and you have a headset on and you're hearing directions from someone who is able to see the whole field, right? But you're only seeing your one section of it. That's kind of the scenario here. And they used voice because they wanted it to be a blind test. They didn't want people to know. whether it was human or AI. So they digitized voice, they used AI voice to give the directions, and that was the findings, right? The human led teams actually perform better overall. now if we go back to Cravelo, then there's something else that came out of that study that was kinda not the study but the the article. Where he talks about how much of middle management is coordination. It's a lot of those coordination tasks. And like we looked at and talked about here, I I think AI is particularly adept at doing coordination type activities. So from Cravello's perspective, he's saying Hey, you know what? maybe this isn't something that we need to have humans do as much anymore. Maybe this is something that that AI can take over. And if we're looking at that slice, right, if we're looking at coordination, then that would might make a lot of sense. But i i isn't AI that's that is science work intelligently actually leading people? I think that's an important question. Does effective management require understanding the circumstances behind the data, not just the data? And I would argue that it that it does in a lot of scenarios. Could AI handle coordination while teams retain control over decisions? That's an important question as well. Depending on what kind of work you do, I think that might control your answer, right? While we looked at what we think AI does better, let's talk a little bit about where human beings do a better job, where human managers still have an advantage. so s some potential strengths from a human perspective over AI. Empathy. AI has no empathy. Adaptability. That's probably our strongest quality over AI, is our ability to adapt and change. AI is programmed, right? And that programming doesn't allow it to go outside of it. humans are better at personal relationships, at at things like mentoring their employees, at at resolving conflict within their teams. Humans handle ethical moral judgment kind of scenarios much better. because it's not when we're in those scenarios, it's not always a hard and fast rule. It's sometimes we have to bend it because this scenario makes us think in a different way. So humans can do that better. And Maybe the trump card of all of them is humans are accountable. Humans can own things and be accountable for an overall success or failure. Whereas the AI can't. Right? there's a another study here that I'll quote. It's called the automation of leadership functions. Would people trust decision algorithms? it came out in 2021, and again this will be in the show notes. There were 333 workers that were evaluated in this study. They they had hypothetical workplace situations that involved an automated or a human leader in this study. The researchers examined trust and managerial decisions, including disciplinary and mentoring situations. So you can already see where this kind of goes off the rails a little bit, right? Automated leaders were perceived as having greater integrity, which is interesting, and transparency. Both those things I found extremely interesting. Automated leaders had better integrity and transparency, or at least the perception from the participants was that. Human leaders were perceived as more benevolent and adaptable, which tracks a lot with what we just said about the strengths of humans. These were perceptions though, so it's important for us to keep that in mind as far as the limitations of the study, right? It's asking people what they perceived. So it's not proof that one was one thing versus the other. It is what the humans thought about. But I think that is increasingly important because what we perceive as humans in interacting with AIs is controls a lot of things that happen downstream from that. So don't write it off just because it's perception from the humans, right? so kind of an a a interesting distinction here that came from this. a fair manager might apply rules consistently, but a good manager may also recognize when following the rules would s produce some sort of an unfair outcome. Now we're back into that ethical moral standard, right? Yeah, and that's what we're so it's important for us to understand there is still that realm for humans that I think we do a better job at. so what happens when AI becomes the boss? Well, there's another study here from Tool to Supervisor uncovering the impact of AI roles and employee performance that came out this year, 2026. And they they used a field survey and a laboratory experiment to examine how employees responded when AI occupied different positions in that workplace. some of the main findings they f they they came away with was when AI was perceived or AI's position was that of a tool, then it was seen as a resource. That the employee controlled. When it was seen as a peer, then it was associated with greater empowerment and performance. When it was seen as a subordinate, then it was associated with greater empowerment and performance, just like the last one. Supervisor, though, it reduced performance through lower empowerment and greater abdication of responsibility. So again, so we have that. kind of who accepts the responsibility for things, right? I think one of the most important discoveries that came out of it was they they said in the study that where AI sits in the hierarchy of the organization or of the structure of the work can affect people's behavior, not just what they think, but how they actually behave. And it's it's not just what the AI is capable of doing that's important here. Right? It it's the effect on the humans. and that that kind of leads us to thinking about if if we designed this today, then back to our original question: if we're starting over and we're designing this from scratch, would this look the same? We're accustomed to designing technology around existing jobs and organizational structures. But what if AI makes some of those structures unnecessary? Or what if it's not unnecessary, but it's just different now? Again, if we're designing from scratch, we might not do it the same way. So there's kind of three possible futures that I would propose here. One is AI replacing the managers. AI handles task allocation, coordination, supervision. Much of that daily interaction, humans at higher levels would still set goals. Kind of the AI sandwich model, right? I think another possibility is AI augmenting managers. That AI handles information processing and the routine aspects of coordination. Human managers spend more time than on the things they're strong at coaching, judgment, relationships, mentoring, developing people, creating a learning environment. And then the third is AI enabling self-managing teams, where there's no manager at all. AI supplies coordination and decision support directly to the teams when they need it. Hey, we need someone to help coordinate this. Great, we have the AI to do that. I don't need to go to a manager. I use it as a tool to do that. If that's the case, we're potentially removing the need for some layer of management there altogether when we do that. The third option there, that last option, it's particularly relevant, I think, to agile organizations. And I know I'm getting into that agile realm, but I I think it's it's important to understand that. We've always advocated self-management in in the agile world for days or for decades. not for days. we we've advocated advocated self-management for decades, that teams are better when they don't have to go through layers of approval. Well, if we empower teams more in that way. And take away some of that management layer of, hey, we have to go get someone else's permission. I need to have this coordinated with another team. Great, there's something here that I can use as a tool that just does that for me. No delay. Right? So you cut out that kind of version of the waste. And if AI removes some of those practical obstacles, such as just that complexity that comes from coordination. then what does that mean for the team? What does that free the team to do instead? What does that free up the organization to do instead of having that function? important distinction here though, removing managerial responsibilities from a person's job doesn't necessarily eliminate the organizational need for those responsibilities in some way. Someone has to be accountable. So someone still must own the consequences and the consequential decisions and be at the end of the day accountable for the outcomes. And that's maybe one of the more important areas for a manager, right? So our our questions then, that I think are still unanswered, that we're gonna have to solve at some point. If AI can handle coordination, Do we still need the same management hierarchy? Or is this yet another layer of our organization that we should be rethinking from the ground up? Right? I had a previous episode talking about how AI is an operating system. Right? It's not just a tool or a replacement for someone, it's an operating system. It changes our entire way that we operate. So maybe this is another one of those areas. Should AI give employees more autonomous more autonomy or give organizations more control over employees? I I know I have an opinion on that one. I think the more autonomy is going to be what succeeds. I think Agile has shown us that. So that's a no-brainer for me. If we started a company tomorrow, and this is maybe the biggest question, with today's AI capabilities, would we invent the same kind of manager that we've always had? I think a lot of times we we step into that role and say, by default, we're going to do this because this is what they teach in business school, or this is the way that business has operated for decades now. But This is yet again another one of those areas where I think we have to step back, remove ourselves from the situation, take a 50,000 foot view, and say, How has this changed the picture? How does AI now cause us to rethink some of the fundamental things that we've thought about our organization? Is that function a person now? Or is that function something that's better as a tool that AI can do? Do we have a new conglomeration of functions that becomes the new manager? Right? That becomes that new role that maybe previously was only part of what the manager did. But if we roll up all the things that humans do best into a person, I think there's still a need for that person. Right, I st I think there's a still a a a universe of things that the AI can't do. So I think there is some need for that leadership. But the more we can give autonomy to the team to have you know authority, control over their destination, over their outcome, then we remove that delay, right? We remove that layer of approval. And if we remove layers of approval, then that just speeds up a team. So it's not speeding up the team because AI is doing something faster. It's speeding up the team because of management structure. You get that part? Right? I think there's multiple things here that we have to take into consideration when we think about incorporating AI into our workflows, into our businesses. This is the kind of stuff we talk about here. So if if this topic is something that interests you and you like these kinds of topics, hey, there's a couple things you can do that would really help us out. Like they always say, like and subscribe. that would really help us. It actually does make a big impact because then people can find our show and join us on this journey. tell a friend, tell somebody that you know, a coworker especially. Right? If there's someone in your organization. And your organization is going through that process of saying, Hey, how do we handle this? Maybe some of these episodes can help, right? I'm trying to give you the facts, I'm trying to give you the research so that you can make your own conclusions about how that fits into your organization. So if you know someone in your organization or a friend or colleague that could benefit from some of this, point them in our direction. We'd really appreciate that. I also would always love to hear from you. Give me feedback on the episodes. Tell me who you think I should have on the show. If we have someone on as a guest, tell me what topic or aspect of the human AI working relationship I should cover next on this show. Always open to those suggestions and really appreciate it when I get those. And otherwise, I think that's about all I have for you this week. we're gonna keep doing this as much as we can and exploring different ways for AI to help us in our team structure. So I hope you're having a great week, and we'll look forward to talking to you next time on another episode of People Over Prompts Podcast. Have a great one.