Episode 14: Would You Work as Hard for an AI Teammate?

Brian Milner (00:24) Welcome back in everyone. Welcome. This is the People Over Prompts Podcast. This is where we talk about the future of teamwork in the age of AI. we're here to try to explore all of the the various dynamics of that. the human interaction, the human-computer interaction, and the psychology, sociology of it, how this affects us as teammates as human beings. I'm your host, Brian Milner, and I'm here as always. And hope your week's going well. We're we're we missed last week, so apologies for missing last week, had some unavoidable things and could not do that, but we're back here this week. And I feel like we got a really good topic to dive into with you. there's a few studies that kind of coalesce around a certain idea here. So the the general prompt that I'm I'm kind of posing for you here in this episode is would you work as hard for an AI teammate? The day is coming, I think, when we all will have AI teammates. So that's the question. Would you work as hard for an AI teammate as opposed to a human teammate? And there's actually been some interesting research on this. So as I said, there's several things I I'll I'll kind of bring up and put in the show notes afterwards for you to find that I think you'll find really interesting here. so the first one is a study that is really about if A AI teammates equal less effort. It's one from Jonathan Winter from 2025. He had about 2,000 participants in this. And he was trying to answer this question: would people put less effort into a team if they were paired with an AI or if they were paired with a human? Would that make any difference into the effort that they put in? And the way he conducted this that participants were going to complete arrhythmic reasoning problems under some kind of time pressure. They were assigned into a couple of different competitive conditions, which are really interesting. the first one is an individual competition. Basically, the idea behind this was that the participants competed against each other. against one another. Whether it was a human person or an AI you know agent, they were competing against others. But it was individual, right? Sol. so they didn't have a teammate, nothing like that. The second group was a team competition. And the team competition participants were paired with either a human or an AI teammate. And their scores contributed to a shared result for their team. Then there was a third condition here that they they tried to test as well. They they had participants that could choose whether to complete a task themselves or delegate it to their teammate. And there was some really interesting findings that came from this. the individual competition. Really didn't turn up that much of a difference. essentially it was equal. the number of problems solved competing against an AI or competing against a human basically statistically insignificant, right? Pretty much equal. The team competition was where things actually really sh kind of shook things up. What they found was they found a 26% higher performance. if the person had a human teammate. Right? If they knew that their other teammate was human, then they they had higher 26% higher performance than if they thought that their teammate was an AI. the delegation didn't really produce much kind of interesting results either. That that was similar to the the individual, not really statistically significant difference or variance there. It's that really the team competition part of it that produced these interesting results. Humans had higher performance if they felt they their teammate was another human. And they had lower performance if they felt their teammate was an AI. Now, this doesn't doesn't solve why, right? There could be many factors that contribute to that, but it is an interesting finding. I I think it's also important to kind of note here that the teammates did not communicate during their task. So they each completed their task, right? but there wasn't a communication back and forth. They weren't helping each other complete task. That was done intentionally because they were trying to separate out whether there was gonna be difference in communication that could have really made an impact, right? Communicating with an AI versus communicating with a with a human. So they just stripped that out of it, right? No communication. Just I have a teammate, they're doing their best, I'm doing my best, my teammates are human or my teammate's an AI. How does that actually change my results? 26% higher performance when the person had a human teammate. So That's a really interesting thing for us to noodle on. as we think about designing teams in this new AI age, where we we are going to have human teammates and AI teammates, where there's gonna be delegation, where there's going to be oversight of a team of AI agents, how does that affect the humans' interaction? Well, this This particular study had some pretty clear findings that it's gonna lower that human's performance. May might not affect the AI at all, but that's what we're here to talk about. How has this changed the human dynamic of it? Right? And if humans think it's an AI teammate, then they're not gonna perform as well. so it it it's kind of brings up some interesting questions about does does it change if we Have a higher trust in the AI's capability? is it about the capability of it? Is it about other more human factors that go along with it? yeah, this study doesn't answer those questions, but I I think it poses those. And that's kind of the next part I wanted to talk about is why why would we work for others, other humans harder? than we would for an AI teammate. There's actually some social theory here, some s some psychology that's involved in it that I think is really interesting. one of the concepts is called social obligation. This is the feeling that we we owe something to others because they're depending on us. So we work harder because we we don't want to let our teammates down. We we have this kind of sense of obligation, right? I don't want to be seen as the person that lets the team down. And the theory is that that may not be present with AI. Because I don't really care if AI thinks I let it down. Because it's not a human, it's a tool. So It doesn't have feelings. It's not gonna see me in a different light because I let it down occasionally. It's a smart calculator, right? It does things algorithmically, and it's a very, very smart calculator, don't get me wrong, but it it it is a tool. So there's no emotion, there's no lingering trust issues, right? So that's one of the the psychology theories here that I think come to play in why we might actually treat the treat an AI differently as a teammate. another is reciprocity. Basically, this is the idea or the expectation that relationships always involve a give and take. There's a shared expectation of give and take. I'll I'll help you today or I have your back. but I have your back because I know that you're gonna have my back when I need it. And again, that's not there with AI, right? Because there's no long-term memory. There there's no social history. it might approximate that. But it's really a context window. And that context window is much, much shorter than it would be with a human. So yeah, there's no expectation of reciprocity. If I have the AI's back, it won't remember that. And it won't try harder to have my back down the road because it has no emotion. Right? another big one, and this this is actually a really big one, and this it's one that we talk about a lot in teams as well. But it's this psychological theory of social loafing. And this is this can be a big problem on on just regular human teams. basically it's the tendency for people to put in less effort when working in a group than when working alone. Especially when they believe that others will pick up the slack. So when there's the the way this works with the human group is if there's a lot of ambiguity around who's responsible for what and who's gonna take which thing to do, there's a pool of things to do, and it's it's kind of cloudy or unclear as to who who's gonna do what, then There's a tendency sometimes in some people to back off of actually doing those things that are undone because there's a a general assumption that somebody'll get it. Right? it I don't have to worry about taking care of everything because it's not all on my plate. It's on our team's plate. And if it's on our team's plate, Then it's not solely my responsibility. So I'm sure someone else has this. That's kind of the theory behind social loafing. And that comes to play here as well. If I have an AI teammate, then I might assume that gosh, AI is super smart. I'm sure it has this. So I might back off of doing as much as I normally would because I have these three super smart agents that are working alongside me. Yeah, I'm contributing too, but I'm sure if I don't do this, one of them will. So that that's something that I think we'll we'll we'll see more of as AI gets smarter. Because right now it's it's very much instructional. You have to give it specific instructions to do things. So it's not as big of a factor with AI today, but as as the AI becomes more capable, I really believe that that's gonna be something that we have to look out for and actually pay more attention to is that that tendency towards social loafing that might happen in a in the human that's involved in that scenario. And the last one here that I think is relative relative to this and should be worth considering is a theory about shared struggle. this is you know the thing that lots of teams experience where they they have a bond and a sense of unity that they develop when they face similar challenges. when they when they have to overcome you know the same kinds of obstacles or or they have to really achieve some difficult thing together. you know this is kind of like soldiers in the foxhole becoming kind of a band of brothers a little bit. similar in a in a different way obviously but similar to the way a team might bond because there's someone that they all dislike right there's somebody who makes their life miserable but it makes all of their lives miserable so they joke about it they laugh about it and they bond over it or there's that really tough deadline that they have to meet and Yeah, they all have to stay up late at night working on this. And after it's over, they have this shared war story of, man, remember that project wasn't that horrible? Yeah, yeah, we barely made it through that. That kind of thing is the shared struggle. And again, when you start to think about a team that's a mixture of humans and AI. Where's the shared struggle? Right? Do I feel like my AI teammate is got a shared struggle with me if I let it run overnight? Heck no. Right? Why would I care? It's a computer. It has no feelings. It doesn't need to sleep. So no. It doesn't ever struggle in the way that I would as a human, because it doesn't have the needs I have as a human. Its feelings aren't going to be hurt if there's a particularly difficult manager. It doesn't care. Right? So there isn't that sense of shared struggle that the human would form this bond with an AI teammate. And though those are just some of the reasons, but I I think that might those might, in some combination, even be an explanation as to why. Humans might not perform as well with an AI teammate as they would with a human teammate. It's it's about the human dynamic. And even though AI is doing its best to try to imitate humans, I think at some level there's still a realization that it's not human. And when it comes to these kind of deeper bonds that we would form with a human, we just don't have those with an AI. Right? We don't have them with an AI because they don't experience the same things we do. So, no, I'm not gonna put my butt on the line for an AI teammate because I don't care if that AI teammate gets fired. But I care if my buddy gets fired. Right. I care if the person I bonded with over that hard project gets fired. I care because they have my back. And I in turn have their back. So I think there's some really interesting human dynamics at play here in just how teams operate. And this is just one more aspect of that. So there's some other studies though that I think still impact this that I want to just briefly fly f fly through for you. There was one study by a gentleman last name of Wu from 2025. they ran four experiments here with 3,500 ex participants. They were trying to determine does AI collaboration improve immediate performance, but Does it then reduce people's motivation when they return to working independently? So it's about motivation there. They they conducted these experiments. The participants completed professional and creative tasks such as writing promotional content, drafting performance reviews, composing workplace emails, even brainstorming, brainstorming like product improvement type things. And They they divided a central comparison here into two groups. One was AI assisted and then solo, meaning participants completed the initial work with with ChatGPT and then they completed work independently on their own. That's group one. Group two reversed it. actually it's not reversed, sorry, it's completely different. They did solo, then solo. So basically just solo. they completed bow tax task with no AI. So first group with AI first, then solo. Second group, solo entirely. So they could have sort of a comparison group, right? Here's what they found. AI improved immediate performance. So participants generally produced better work on tasks where chat GPTs, assistance was available to them. But their motivation decreased after the AI was taken away. So when they transitioned back from AI assisted work to solo work, they self-reported lower intrinsic motivation and actually even just higher levels of boredom in their work. And the benefits did not reliably carry over. Participants generally did not retain the same performance improvements that they had with AI when working alone afterwards. So interesting little study there about motivation. And you know, kind of the key takeaway there is that the better output doesn't necessarily mean more motivated employees, and that if we do work with an AI, that doesn't necessarily upskill us to then go back to doing work alone and do it better. in fact, maybe the opposite is true, though although this study didn't prove that. That it could actually be the case that we decrease our skill, right? That would take another study though. so then then the question is, what's the solution? Should we should we humanize AI? Because that's clearly that the approach that the AI companies have taken. Is that the the solution here is to humanize it as much as possible, make us forget. That the thing we're working with is not actually a human, that it's a a program. Well, there's a study done by this this other group from 2026. It's three studies, and again, I'll link all these in your your show notes if you want to follow up on this research yourself. But they're try they tried to figure out are people more motivated to contribute when they experience stronger social presence from their teammates, including AI. And what's meant by social presence here is how much another participant feels like a real present social actor during an interaction. So that's what they meant by social presence. They they divided us, they actually conducted three separate studies here to find this out. One was an online experiment containing an identified AI teammate with a supposedly human teammate. The second one was an online experiment testing whether increasing AI familiarity changed the social presence. So the human was made aware of the AI and helped get a little instruction about what was actually behind the scenes with the AI AI. How was it work how did it work? What were its limitations? That sort of thing. That's a second group. And then they also had a laboratory experiment where they used some some facial expression analysis and showed some images of people and did some skin response measurements and and did some questionnaires afterwards. So the the outcome of this, participants worked in a simulated five-member team, and they they used the text chat to brainstorm their ideas with the team. The other apparent team members were simulated actors using predetermined messages. That that was meant to control the interaction and keep the differences. And how AI generated responses might occur from really distorting the end results of this. Okay, so predetermined outcomes here. In that second study, again, remember, the second study is when they gave higher familiarity of what the AI actually did. So this was the interesting outcome. The second study, the researchers increased that familiarity and helped the participants understand what it did. It found that the social presence was associated with motivation to contribute, both directly and through people's willingness to depend on teammates and their commitment to the team. So, yeah, interesting finding there. Making the AI more familiar, and this is where it kind of stood out to me. Helping the participant understand what the AI was and what it did and its limitations and boundaries substantially narrow the difference in perceived social presence between the AI and human teammates. So the conclusion that they they kind of reached here was that AI's role, operations, and limitations could actually support social presence. without relying solely on on appearances and personality. So we we try to answer that question, should we make the AI more human-like to try to mask the fact that this is actually a computer? And this study was actually saying no, that it's actually a better outcome if you make the participant more aware that they are working with an AI. So interesting little outcome there, right? and I think that kind of leads us to the the the big question here, which is when we think about redesigning teamwork in general, which is I think really kind of the point of this this podcast and and I think where a lot of effort and energy is gonna be put in the next few years. I think it's important to understand how these things come to play. you know, our first study, an AI teammate identity can't can actually affect human effort. The second study, productivity can coexist with lower motivation. And the third one, the social presence and familiarity affect commitment. So I I I don't have the answers for this yet because I think that's still ongoing research. But I think there are some really important questions that we need to be thinking about. And hopefully research over the next few years will help us resolve these things. One, does commitment to a team require commitment to other people, other humans? Think about that. Does commitment to a team require commitment to other people, other humans? Can we have a commitment to a team as a human being if our other team members are AI agents? Or is this such a fundamental upheaval of the concept of team that it just doesn't work unless there is another human? I mean from my background, coming from kind of the agile and and scrum world, that makes me think about the value of commitment when an agent joins the team. Right? Part of the reason that we think commitment is important there is because you you kind of have to look each other in the eye and say, Hey, I'm committed to this. Are you committed to this as well? And there is sort of that social pressure of, all right, well, I'll I'll have your back in this if you have my back. And the the social loafing doesn't happen because we talk about it daily so it doesn't fall through the cracks. We don't ever assume someone else has it because we're very transparent about that conversation. So how does that change when there's AI teammates? And what's the tipping point? How many humans versus how many AI agents? That's something I'm really interested in trying to find out. another question is: is the answer to make AI more human-like? That's possible. I think the study that we looked at shows that may not be the case. What may be the solution is to redesign our team around these differences. And how we perceive them, how the humans are made aware of who it is they're teamed up with, that that could have a big impact on it. Another one that I think will be more prevalent is are we measuring team success only through output? Because if we ignore things like team cohesiveness, motivation, then those components of how a human fits into a team, I think somehow get lost in the shuffle. And we st the danger, right, is that you either start treating the AI more human and try to make it more human, or you do the reverse. You start treating the humans more like AI, more like computers. And I think that's a very, very dangerous possibility here, right? I think there is going to be plenty of organizations that make that mistake, that end up treating their humans more like AI, and really denying what makes them human, which is their advantage over AI. So I think it's gonna be an interesting time period here. And one of the reasons I I I wanted to start this podcast was to kind of just kick up the dirt a little bit on some of these issues and say, Hey, if you're not aware, they're coming. Right? These are the things that are happening today and are just going to bloom, right? They're gonna blossom over the next few years because. Teams are going to be increasingly more and more hybrid. human AI teams, hats. And when that happens, how does that affect the psychology of the humans on the team? How does that affect performance? How does that perfect affect motivation? Right? So that's the kind of stuff that we talk about here on this show. And if that's interesting to you, if this is stuff you think, wow, yeah, this is the kind of stuff that really fascinates me as well, you found a home. this is this is the place for that. that's my goal is to kind of bring these issues to the surface with every episode. Sometimes it'll be with a guest that I think you'll you'll enjoy hearing, sometimes it'll be with just me bringing up some research that I I think you'll find interesting. But either way, it's gonna be focused in this area. What is the future of teamwork here in the age of AI? if you like this and you think this episode is something that maybe some a colleague of yours or a friend of yours or relative or somebody might also find interest interesting, then the number one a ask that I have for you is to just point them in our direction. Just say, hey, you know what? I actually heard this podcast the other day that was really interesting. It talked about, some of these aspects of what changes when when humans and AI team up. point in our direction. we don't advertise anywhere else, and this is the only way we grow as a podcast is just your word of mouth. So that's really something from the bottom of my heart. I appreciate when you guys do that. also, of course, you know, the typical social media plea, like and subscribe to this. that that sounds so trite, but it it it really does help people find us. it helps people when they search for this topic. And I'll tell you, there's not a lot of other podcasts that are really trying to focus in this area. So I I do want others that are interested in this to find it as well because I think as as we maybe form a community around this, we can start to solve some of these questions and really start to shape what our teams need to look like moving into this brave new world that we're all finding ourselves in. so yeah, and subscribe to that. That'll make sure you don't miss any any of these. the other thing I'd ask from you is if you have any suggestions for me, any guests you think that I should interview, if you're there's a research that you're that you have found that you think the the the audience here would particularly like to hear, or topics you want me to cover, you can email me directly and this is my email it comes straight to me. It's Brian with an I. Brian at agilityevolved.com. That's my company, agility evolved dot com. We put on certification classes. I do CSM C SPO advanced Product owner advanced scrum master classes. I do private classes. I've been doing a lot of private classes recently for organizations that allow me to customize that material. We can weave in a lot more AI discussion when it's those private classes for companies. So if that's something you're interested in as well, if you think that might be something that your company could benefit from, hey, by all means reach out to me. you can even on our website at agilityevolve.com, you can schedule a free. 30 minute consultation with me and we'll just have a conversation. You and I. We'll talk about it and see if there is some way that I can actually help you. But if you have anything that you want to email me, Brian at agilityvolve.com. That would really be great to hear from you. I I always love to hear from people who are listening to the show. I I've got a I still, you know, I'm within this window, so I'll make this plug for this as well. I am speaking, I'm gonna be a keynote speaker at the regional scrum gathering in Central America coming up. That's going going to be in Guatemala. It's in Antigua, Guatemala. That's November 4th through 6th. So so if you're in any way considering something like that, then I highly encourage you to do that. it's a beautiful place. And we can just hang out. we can have some conversations there. You can come to my talk and we can talk about some of this stuff afterwards. I'm going to be talking there about this very topic about talking about how this changes our team dynamics. So yeah, if that's something that you you think you might be interested in, you can go to the Scrum Alliance website, Scrum Alliance.org, and you can find out information about registering for that conference. That's about all I'll have for you this week. I hope everyone's week is going well, like I said at the start. And we'll talk to you next time. Another episode here of People Over Prompts Podcast. Have a great one, guys.