S11E7: How AI Is Reshaping Communication and Trust

Welcome to Age of AI, Politics, Law and Society, hosted by Meddler Memma, Chris Lamont and Young Diogenes in association with the Global Governance Institute in Brussels. Peter, thank you so much for being here. Before we jump in, would you mind briefly introducing yourself and sharing a bit about your work? Sure, yeah, it's great to be here. I'm a professor at the University of Southern California and I teach business communication. Everything that we do focuses on helping managers and leaders be better communicators in the workplace. And so we cover all kinds of things, anything from presentations to written communication to leadership communication to the entire company, even getting into a little bit of negotiation. But those are kind of the areas we focus on. And technology impacts that in a huge way. And we've been focusing on technology-mediated communication for a long time. My interest in AI started about 12 or 13 years ago and was kind of a side hobby. And now we're in the last three years full-on focusing on how AI impacts communication in the workplace. Wait, 12, 13 years ago, that's before, you know, chat GPT. Are you saying there was AI before chat GPT and LLMs? Yeah, there's a lot of us who've been following it for quite a while. But if you don't mind, I could tell you the very moment when it became a professor. He's due. You know, when Google Glass came out, I was at a tech conference. And one of the use cases they described as possible with Google Glasses was that someone could go to some type of networking event and then the Google Glasses with computer vision would recognize people and actually give them cues about how to network effectively with the people they're seeing in the room. Yeah, it's going to remind them, oh yeah, you talked to them about their pet dog last month. Maybe you do a follow-up question on that. And so from that standpoint, I was just intrigued by the communication elements of that technology. And then there's all kinds of implications for not only how influential you are, but how authentic you are. All of that addresses issues of trust that we get into all the time in Earth. And that's actually a good segue because trust is exactly what I want to talk to you about, you know, how AI impacts trust. How much we trust AI, right? What impact that has generally societally, but also interpersonally in those relationships. We will address this a little bit later, but just thinking about your conversation. You just mentioned what first started this. If you no longer commit some of these conversations to memory because you trust that some element of augmented reality will actually help you through that, how genuine is that concern? Or is that consideration? Or is that relationship with that individual? That's the first question that kind of comes to my mind when it comes to trust. Yeah, well, I think you raise an issue that's important. I mean, we've entered a stage where autocomplete in communication has taken on a whole role. It used to be a few words. And now it's to the point where you can actually outsource all of your focus on the other person, the empathy that's normally there as you're constructing how you want to interact with that person. So it requires certainly a lot of intentionality, especially if you're a super user of AI. So there's all levels of use of AI, but we actually have 10 to 15% of the people out there in the workforce right now. And again, that's what we focus on is the workforce in particular who are super users. In other words, they're using it for all of their emails, they're doing it when they prepare for a meeting, they're generating ideas. And so if you're in that position, you really have to be intentional about how much of that is deriving from you as opposed to this outside entity, so to speak. Well, this is a good preview of what our conversation will be, I think for the next 30 minutes. There's three things I want to talk to you, is I've just kind of looked through your body of work and we've talked about this before offline. There's three big questions, at least in my mind. Question number one is about public attitudes and the paradox of trust in AI. And then question number two is the impact of that interaction that we just talked about between AI individuals and AI in society. What impact that has in interpersonal and societal kind of trust? And the third one is like, what do we do about that? The question of regulation, does a government have a role to play here? What about business and others? So those are the three big questions that I want to address over the next half hour. So let's start, maybe the first one. Let me share with you some sort of research on public views on AI today. I'm sure you're familiar with the Edelman trust barometer. So the most recent report, the 2025 report, notice that we're seeing interesting trends. For the first trend is that there's sharp global divides when it comes to trust in AI. So in China, for example, 72% of people say they trust AI. In the US, it's just 32%. That's really, really low. And then these divides go beyond geography. Trust in AI seems to be consistently lower among older adults and lower income individuals than women. So there is that divide. There is an income divide. There seems to be a gender divide. There seems to be a geographic and age divide. And then there is this additional kind of layer on top where broader trust in technology and technological companies is also going down. So a decade ago, 73% of Americans trusted technology companies, that number has dropped to 63%. So 10% in a decade. And then globally, only 44% of people feel comfortable with businesses using AI. And it's even lower in the United States. What do you make of these numbers? What do they reveal about how people understand and evaluate AI? So we do research constantly. Ever since Strat2pt's come out, I would say every two months, we're serving knowledge workers in the United States. And one thing that you can see consistently is there's really mixed views of AI. Well, on the one hand, people recognize all of the personal benefits for themselves, but they also recognize all of the potential consequences. So one of the areas where most of the concern comes is with things like job security. And what does the future look like for me and my kids? And do I really want my leader or manager using this kind of technology? So we actually see that, you know, not only see the division across a society, but in an individual themselves. A person will hold very mixed views of AI. And I'll give you a few examples of that. If you ask knowledge workers, so people are spending most of their time at computers, do they see the personal benefits? And that, in my view, equates to a certain level of trust. Do you trust it to help you write an email better? They will say yes. 60 to 70% will say, "Yeah, I can trust AI to help me write a better email." And at the same time, you might find that 40 to 50% are really worried about the role that AI is going to play in the future of work for them. So again, you have these mixed attitudes. A person loves what it's doing for them right now. They're really concerned about what it's going to mean for their work next year and two years and three years. And they're really worried about the way their leaders and managers are going to use it as well. So in many ways, I view it as an emotional roller coaster for a lot of people who are using AI a lot. There are a lot of ups, a lot of downs. And in all of that, there's uncertainty. And whenever you have uncertainty, that creates problems for trust itself. Okay. And let's dig a little bit deeper on this because you're referring to some of your research on this and surveys. You and your colleague, Brian Marshall, if I'm not mistaken, you conducted a survey with about 800 full-time professionals. And one of your key findings was that the more active AI users, as you're just pointing out now, are actually comfortable with AI in human-like roles, like teammates or even performance coaches. Less active users, however, feel much more uneasy about that. Again, based on your research, what do you think explains this divide? Secondly, how should leaders, managers, and educators prepare people to adapt, especially as AI becomes more embedded in society? All right. To the first question, I think it's interesting you raise this issue of coach or therapist. That's certainly a human-like role. And our survey showed pretty clearly the people who are using this a lot are more comfortable with those roles. I would also note that if you watch some of these podcasts with AI leaders, and I'm specifically referring to, for example, open AI leaders like Sam Altman, and you look at like the COO, and then other people who are at some of these other companies, Anthropic or Google or wherever, they often get this question of, what do you want your kids to know about AI? And one of the consistent answers across all of those, one of these threads, is I want my kids to use AI like a tutor. About a month ago in Harvard Business Review, there's a blog post where there's a fellow who's been following the most common uses that people have for ChatTPT. In this year, the number one use was therapist. And so the survey we did was fall of 2023. It was almost a year into the post ChatTPT AI world. That was the warm-up. In 2025, we're increasingly seeing people who are saying, I'm comfortable with that role. And there's several reasons that you see that. Number one, there's an enhanced ability by the AI to actually understand you, so to speak. I want to use that word understand loosely, because it's not understanding the human sense, although many people interacting with it feel like it's the human sense. That's right. But it has some kind of memory of your interactions, it learns about you, and then it is able to personalize its interaction with you based on that. And there's an enhanced ability by AI to accomplish that goal. But one of the other things that's really critical, and you'll hear a lot of people talk about this, that use this as a performance coach or a therapist, is they feel like the AI doesn't judge them. And since it doesn't judge them, so to speak, and again, when they use that word judge, they're ascribing a human function. They feel less vulnerable, they're more willing to open up. They feel like what they see is potentially more objective than if it were a human. Is there a risk that people confuse polished delivery with credibility? Are we trusting the style and not the substance? Well, I think there is a danger. Historically, professionalism from a communication perspective has been really closely tied with the credibility of the communicator. And in the research, particularly that I've done with Anthony Coleman, we just had an article that came out today, it was labeled professionalism and trustworthiness in AI assisted writing. And it actually shows that those two constructs, professionalism and credibility that historically have been so closely knit, are no longer as closely knit. In fact, you will see people who judge communication as professional, but not credible, if they know that AI is involved. So certainly the presence of AI is creating a division between those two notions. And it's just really intriguing to watch it occur. And I kind of suspect that people are going to start craving more and more authenticity in their communications. And as a result, they're going to become more and more skeptical of something that sounds polished or professional. I had that happen to me just the other day. I was asked by both Chris and YD to come up with some ideas about what the next five years kind of look like. And I had a pretty clear and clean list of ideas of what that would be. And then both the reactions were, did AI produce that or was that you? And I thought that was an interesting reaction to it, because now coming across as articulate and polished may actually come across as an authentic, right? Yeah. And it's interesting, we have another survey we did where we asked people who were employers or in business roles to tell us what roles they think are most important for MBA graduates. And historically, oral communication and written communication are right at the top of that list. Well, interestingly enough, the last time we ran that written communication has moved way down the list. And oral communication is right at the top. And it's right, it's precisely for the reason you've just stated is that oral communication is more trusted in this moment, it's more spontaneous. It's a reflection of people's true sentiments and ideas, whereas there's that skepticism of something that's written. Although I will say this, like the cues that people get for written communication are coming down the line for oral communication as well. I mentioned the Google Glass, mentioned early on, we're going to see modalities of AI that are just as powerful with verbal or oral communication as they are with written communication. So the trust issues are going to come for that as well. And so this is actually what I'm going to say in a little bit. I want to move beyond the public perception of the big data kind of stuff on this, just something a bit more personal that you have written as you have just pointed out quite a bit on and how AI is reshaping trust between people, this interpersonal aspect of that. And then I want to move on to the societal trust, but first with interpersonal trust. In one of your recent articles, you warned actually that AI-mediated communication could threaten the relational foundations of human interaction. In other words, it could threaten interpersonal trust. When someone uses AI to write a condolence message or a performance review or a love letter, write any of those. The first question on mine is, who wrote this? Was this you? Can you talk a little bit more about how AI-mediated communication could threaten the relational foundations of human interaction? Well, I'm going to point again to workplace communication. And again, it's going to be broadened to interpersonal communication outside of work as well. But one of the recent projects we did, we basically just asked people's opinions about AI-assisted communication from a manager. And we have all of these open-ended comments. And the themes for those open-ended comments really drive at three key areas of trust. Number one is a construct called benevolence. Do you care about the person? The other construct is, do you think the person's competent? Do they have skills or abilities that should place them in the role they're in? And then the third is kind of this character, ethical ethics. And in all three dimensions, you see the comments very suspicious of people who are using AI to communicate, especially managers and leaders. Number one, they don't trust their character. So that's one of those bedrocks of trust, right? They don't know if it's really that person's own values that they're lining up with or AI telling them to line up with these values, right? Number two, do they really care about their employees? We see all kinds of skepticism about the fact that managers really care about their employees. And then finally, the competence. And that gets into questions of, should this person even be my manager or leader if they need AI to do their job, right? So I think those are pretty useful bedrocks of trust to think about. And I think we're clearly navigating a new world. This is new terrain as people are using AI increasingly within their relationships. But if you think about those three areas of trust, I think they're helpful. So Peter, let me just ask, because you talk about how a lot of these attitudes also evolve over time, right? They're not always been the same. First of all, do you think that it's the right attitude to distrust somebody who's using technology? And second, do you anticipate that stay in the same way as more and more people begin to use AI in ways that may be perceptible or imperceptible to them? So five, seven years from now, those bedrocks of interpersonal trust may actually be reconfigured because obviously people in our are a bit more comfortable with them. I imagine if you have the super users in the workforce, they perhaps are less bothered by that use of AI because of themselves use it. Am I wrong in that? And do you think that those indicators are going to stay there and we're going to see resistance in the long term? So a few quick thoughts on that. The super users, yes, it's true that they're more likely to extend trust to others as they're using it a lot. That said, there's a huge divide between people's own perceptions of their own AI use versus others. And people give themselves way more benefit of the doubt than they do others. So I think that's really critical to understand and it's really critical for super users to understand that so that they have that self-awareness to know that they may not be extended the leniency that they give to them. So that's part of the response. Yes, things change over time. I think your first question, though, related to this issue, should we just be simply trust in the first place? And who would argue, yes, even pre-AI, you could see trust at a societal level dropping over many decades. And so you referenced the Edelman barometer. There's many other measures in Gallup and other places. And the US society in particular, 20 years ago, it was considered a high trust society. And now it's kind of mid to low. There's plenty of societies that are lower, but we're kind of in that mid to low category by many measures. And so, idealistically, I just think, yes, we need to find ways to extend trust to people. And that should be the starting point. I was actually trying to envision a scenario where the emergence of AI and its embeddedness, so to speak, in society may actually be a net positive in terms of trust. Is it possible, at least at the societal level, that the emergence of AI may make these institutions more valuable again? I would like to believe that. I'm optimistic. I think this is an inflection point in history. And I would like to see a lot of prominent people step up to create that environment of trust in the public sphere. And so, yes, I think it's certainly an opportunity. At one thing that you see with the AI, you know, the various LLMs, they all rely on a corpus of information. And ensuring that you've got the right body of information as the basis for that AI output is critical. And I think that's just one of the areas in which you may see more trust. I would just give one example. This summer, I've had the opportunity to work with a company called Josh Berson. They do human resource consulting. They're completely in the HR world, but they've created this pretty cool AI tool for all of their users that is completely grounded in the expert research they've been doing over the last few decades. And that becomes more reliable as an AI tool, so to speak. I think there's a place for institutions to ensure that the AI that we see is more reliable on kind of the AI side. But aside from the AI side, there's an opportunity for leaders of those organizations to engender trust with society. So, independent of AI, right? Yeah. And I would love to see it because the trend's not positive right now. We've kind of talked about that. The trajectory is down for nearly all institutions. Yeah. And I guess part of that thought experiment that I was sharing earlier was that it's a good moment right now because there's a need for trustworthy institutions because I think increasingly, it will be more at the loss of understanding and distinguishing between what is real and what is not. And so having trustworthy, reliable institutions that can in a way put a good housekeeping seal of approval, let's say, on either AI products, but also just on the information itself, right? That the information is credible and trustworthy. Then I think that that is important. That's actually just as we wrap this up. I want to kind of get to the last part of this. So, you know, talking about solutions. So what do we do about this, right? We have discussed how, I guess, AI could both threaten interpersonal trust and societal trust. But there's also this open door that we just discussed, the possibility of maybe making some of these institutions relevant again. What role can and should regulation play in addressing these challenges? And by institutions, I don't mean just public institutions, but also private. Yeah. So I think I want to take this in two directions. First, I want to focus more on companies and the workplace because that's my world. And so I think there's a place for creating policy and norms within companies. Secondly, I think it's important to frame the way AI affects our relationships as more than simply a threat. I undoubtedly think there's a threat of AI reducing trust. But there's a lot of people experimenting with AI in ways that can build trust between people. So I'm going to try to put those two together. We've heard from just hundreds and hundreds of professionals that they're okay with AI in the workplace in certain ways. And so we've seen them experiment with tools in which they're helping AI is actually being a guide in terms of teen dynamics and pointing out which people are dominating conversations and which people need to be heard from more. And so those tools can actually help people function better in the workplace. But the key is, are you doing it in a way where there's some consensus around how the tools are used? And so do you have, are you building those norms and policies inside the organization from bottom up with from the middle top down in all directions to ensure that people have a voice in how those tools are used? And the more you see a voice in those tools and a collaborative approach to developing how we use the tools, you're going to have people on board and you're just going to have way more trust in the workplace because it's guided by people's actual hopes and desires. So to that question of policy, I would respond with a regulation, I would respond developing policy and norms in the workplace to the degree that employees are involved consistently in developing those, you're going to have a culture where you can have AI and you can have trust at the same time. As we close, if you could offer our listeners one principle or mindset to carry forward as AI becomes increasingly part of our lives, what would it be? Stay informed and document your own AI use and how it affects your life. There's always been a proponent of people journaling for many purposes, develop their own leadership and communication abilities. And I think the benefit of journaling is you're really carefully documenting the impact of various things in your lives and in this case technology on your relationships. And it creates an awareness and an intentionality that will be especially critical in the AI world. So stay informed about your own use and critically evaluate it constantly. Wonderful. Peter, thank you so much for taking time to be with us today. And of course, thank you to our listeners as well. Join us again next time for another discussion at the Intersection of Law Society and Politics. Until then, stay human.