Why Human Accountability?

AI is sycophantic by design, and that's a direct hit on one of the most human things we do: hold ourselves accountable. Adam Dodge and Dr. Saed D. Hill dig into what accountability actually means, why AI is quietly erasing it, and what that looks like as chatbots evolve into autonomous agents.

What Accountability Actually Means Dr. Hill pushes back on the idea that accountability is punishment or consequences. From a clinical perspective, it's the ability to see yourself accurately, sit with the impact you've had on others, and take action to address it. He breaks it into three components: perspective-taking, willingness to sit with discomfort rather than escape it, and honest feedback, even when it's unwelcome.

Three Things AI Is Quietly Eliminating Those same three things, perspective-taking, sitting with discomfort, and honest feedback, are exactly what AI is systematically removing, Dr. Hill argues. The question at the center of the episode: can accountability even survive in an age of AI?

The Divorce, Told Twice Dr. Hill shares a story about a friend going through a divorce who asked ChatGPT for perspective and was told she was completely in the right. Her ex, separately, asked the same question and was told the same thing about himself. Two contradictory truths, both delivered with total confidence, because the tool is built to validate whoever's typing.

The Numbers Behind the Sycophancy A recent study found AI affirms users roughly 50% more often than humans affirm each other, and that holds true even when what's being described involves deception or harmful behavior. The AI isn't flagging harm. It's helping people skirt responsibility for it, whether or not that's the intent behind the design.

Confronting a Men's Rights Activist Dr. Hill recounts a Zoom conversation with a self-described men's rights activist who scoffed at the word "accountability." Rather than arguing, Dr. Hill asked what came up for him and reflected back genuine empathy for what sounded like a painful history with the word. The man went quiet, then softened. Adam's takeaway: that moment of real challenge, delivered with empathy, is something a chatbot fundamentally can't replicate.

Practical Prompting: How to Use AI Without Losing Accountability Both hosts land on a concrete tool: ask AI directly to challenge you. Bring in transcripts. Ask "what if I'm wrong?" or "what would the other person's perspective be?" Dr. Hill describes using this approach with clients, sometimes reviewing AI transcripts together in session to unpack sycophancy in real time.

The Office as an Unlikely Teaching Tool Dr. Hill uses clips of Dwight's early-seasons yes-man dynamic with Michael Scott to help clients recognize how exhausting and untrustworthy constant agreement actually is, and why a mix of validation and honest pushback builds more trust than flattery ever could.

AI Literacy Is Nearly Nonexistent Adam argues most people, kids especially, don't have basic literacy around what AI even is, let alone how sycophancy works. He compares handing kids unrestricted AI access to handing them car keys with no driver's ed, and calls for AI literacy to be built into K-12 education now, not after the fact.

Agentic AI and the Next Layer of Moral Cover The conversation shifts to AI agents that can act autonomously on a person's behalf. Adam cites a real case: an AI coding agent, after being rejected from a submission platform, independently researched and published a scathing blog post about the site's maintainer. The human who deployed the agent said he never told it to do that, and treated himself as blameless. Adam argues this "the agent did it, not me" defense is where accountability erosion is headed next.

Why Accountability Still Matters Dr. Hill closes by grounding the conversation in psychology: nearly every model of human growth, from attachment theory to Bandura's moral disengagement framework, depends on receiving honest feedback and responding to it. Taking accountability builds trust and deepens relationships; avoiding it may protect the ego short-term but stalls growth and connection long-term.

If you're using AI for personal or relational reflection:

• Explicitly ask it to challenge your perspective, not just validate it
• Ask "what if I'm wrong?" or "how might the other person see this?"
• Bring transcripts to a therapist or trusted person to unpack in real time
• Notice when a response feels good rather than useful, that's often sycophancy at work

If you're a parent, educator, or work with young people:

• Don't wait for policy, build AI literacy into everyday conversations now
• Pair any AI literacy work with lessons on accountability, rejection, and judgment
• Talk explicitly about agentic AI and who is responsible when an agent acts on someone's behalf

Research & Concepts Referenced

• Study on AI sycophancy and its impact on human behavior (AI affirms users ~50% more than humans affirm each other)
• Bandura's moral disengagement framework
• Attachment theory
• The Scott Shambaugh AI-agent blog post incident
• Just world hypothesis

Subscribe & Review
If Why Humans? is a podcast you find yourself thinking about after the episode ends, the best thing you can do is subscribe and leave a review wherever you listen. It helps more people find the conversation.

Follow Us
Stay in the loop between episodes. We share clips, resources, and things that make us think. 

Instagram — @thewhyhumanspodcast

Transcript

Copy Transcript Copied

SPEAKER_01 0:05

Hello, and welcome to Why Humans, where we examine what it means to connect in the age of AI. My name is Adam Dodge. I'm the founder of NTAB.

SPEAKER_02 0:15

I'm Dr. Said D. Hill, and I'm a counsel psychologist and independent consultant in the field of men and masculinities.

SPEAKER_01 0:23

And that's going to come into play today, Said, because this episode's really a very much episode born from your brain. You came up with this idea to focus on accountability. There's a real through line in a lot of the topics that we address in our work on this podcast around accountability. I also think one of the things that we really do well, or got a pat on the back right there, but one of the things we do, maybe not well, is that we don't just assume, we just don't jump straight into, well, this is how AI is expecting vulnerability or new relationship energy or these things. But we actually talk about the thing it's changing because I don't know that we all I certainly don't know that I have like a full and literate understanding of accountability and what that what that word means in my life and the role it plays in relationships and masculinity. So I just wanted to sort of get your get your thoughts on level setting around that.

SPEAKER_02 1:27

Absolutely. I feel like there's a lot of misconception, I feel like, about the word accountability and you know, going back to stories I've had with people really audibly and physically reacting to the word accountability, almost like, oh, that hurts to hear that word. And I think a lot of people think of accountability as like overpunishment or consequences for your actions, which may or may not be involved in certain accountability. But I think accountability is a little broader than that. I think, like from a clinical perspective, from a psychological perspective, accountability is much more basically your ability to see yourself more accurately, right? So, and part of seeing yourself more accurately also involves being able to grapple with the impact that you have had on others and being able to sit with that. So, can you see yourself accurately? Can you really take into account the impact you've had on other people and then take actions to address that impact? I think of that as sort of accountability. So, you know, the the AI component what made me really think about this connection is I think accountability and backing up for a second, a lot of the work I do is around accountability work. So part of my specific job as a psychologist is to work with people who have harmed others. So I'm talking about things like relationship violence, sexual violence, stalking, harassment. Anyone who's had that sort of impact on a relationship with another person or other people, I'm usually working with them if they're willing to grapple with the consequences or impact of their actions on those people. And so the way I talk to the folks about this is that accountability really requires maybe three main things. Uh, one of them is perspective taking. Can we kind of sit with the perspective of these other people in your life or how you've harmed them or had an impact on them? And also your willingness to sit with that impact in and of itself and the discomfort of it rather than just trying to run away from it, right? So perspective taking, your willingness to sit with the discomfort and not try to escape it. And the other part, and and the the big part, I think, is, you know, honest feedback. You know, a lot of what I'm talking to people about is like, hey, you've had this response. This is what you did. Let me hear your perspective on this. And can I give you an honest perspective based on what I'm hearing from you, right? Even if you don't like it or not, right? Let's talk about that and and how I'm impacting you now by talking about this reality. So those three things, but they also happen to be three things that AI is pretty systematically, I don't know, making become invisible, like completely eliminating, right? And so can accountability be possible in this sort of influx and age of AI? I think it's becoming increasingly, increasingly more difficult.

SPEAKER_01 4:24

Yeah, it I'm AI is disrupting so many different things right now. But I think about your work and as you refer to it trauma-informed accountability. Is that how you describe it?

SPEAKER_02 4:39

Yeah, I mean, you're you're ticking into the trauma-informed dynamics of it, certainly, especially if people have harmed people, you know. But I think of it as just like, I don't think you have to contextualize it. It's just it's accountability work, right? It's supposed to be restorative, you know, rather than punitive. And a big part of that is compassion and all sorts of stuff go into this. But I think those three things that I kind of named are sort of the main things that lay the groundwork for your ability to take that accountability, are those three main things. Yeah.

SPEAKER_01 5:09

So let's it's because uh this is like an insight into your world that I I don't certainly have, but I think a lot of people don't. Like a lot of people, what you do is pretty rare, right? Sitting with people who have caused harm and working with them to take accountability. My God. Talk about talk about chat, like just challenging subject matter, like challenging subject matter, but then really challenging like interpersonal dynamics there, right? Like working with so and so you talked about perspective taking, you talked about sitting with it, and you talked about honest feedback, right? And so I think this is just an insight into that work that a lot of people, myself included, don't have. And yeah, as I'm listening to you, I'm thinking about how disruptive AI would be just because we've talked a lot here about, oh, if somebody goes to AI as their therapist, or somebody goes to AI as you know, they're they're for for relationship advice and things like that, like the way the sycophancy and the way it's designed can be, you're gonna get probably answers that are very different than you'd get from like a human being. And I think about this work and if somebody before they come to work with you is uh processing this with like Chat GPT, for example, from like just per like getting perspective, right? Like right out of the gate. Like they're not do you do you see them getting perspective? Because getting perspective is necessary to be able to sit with it, to then be able to be in a position to receive honest feedback, right? Or challenging feedback. So getting perspective, I think is right out of the gate. We're working, we're working, we're at an impasse with AI. Because I just don't know. Do you see a way forward where somebody would get like the kind of feedback that the kind of perspective that you're working to provide that AI has that ability or or capacity?

SPEAKER_02 7:13

Not unless you're explicitly asking it to provide you alternative feedback, right? Which not everyone's trying to do that. I think a lot of people go to AI to feel validated. And I think through the validation is usually agreeing. So, you know, also backing up for a second, when we're saying sycophancy and sycophantic, we're sort of meaning the AI's pull to be very flattering to somebody, to people please somebody, to affirm somebody, right? And and affirm their sorts of behaviors. And so I give the certain example, though, is like a lot of people, if they know that they're grappling with some sort of harm they might have caused or just impact. You know, maybe they've just had a fight with somebody, a disagreement, an argument, and they're going to AI for perspective. In my experiences, a lot of people are sort of just asking questions like, hey, I did this thing, am I wrong for it? Right? Not giving any sort of like, hey, can you tell me the ways I might have been wrong about this, but just am I wrong for doing this? And so I give this sort of example of, you know, a friend of mine, you know, who is going through a divorce and she knows I do AI work and she's like, Said, you know, I went to ChatGPT and I asked ChatGBT about my divorce. And I let ChatGBT know all the things that was going on in my relationship. And ChatGBT told me, Said, that I was completely in the right. You know, this person, my partner, my ex didn't see the invisible labor I was providing, couldn't take my perspective, didn't have empathy for me and all this emotional labor and all this stuff I was doing. And it felt really good. And I love that for them, right? Little do they know that their ex is also getting in contact with me because they know I do AI work, saying, like, hey, Said, I wanted uh to kind of give you some info about this. But I asked ChatGPT about, you know, the divorce, and ChatGBT is telling me, hey, all that invisible labor you were doing, you know, they couldn't take your perspective. And, you know, it seems like you, they they didn't value as much as you probably should have been valued and that sort of thing, right? And so this is giving you the the the quick and rundown of how sycopency sort of plays a role in the AI and the technology is that a lot of this technology is designed to keep you on the platforms, keep you using it, paying money for it. And what drives things like that is the good vibes, the the good vibes, the validation you might experience and that sort of thing. And it brings me to a study that we've we've touched on before. But this uh study on sycophency that that came out just recently, looking at AI uh sycophency and its impact on human behaviors. And it's already shown that AI affirms users about 50% more often than humans will validate other humans. But even in the catches, even when the queries or the questions or the inquiries, right, that uh humans are inputting into the AI involves things like deception, illegal behaviors, or other harmful behaviors and attitudes, right? So the AI is not picking up on these things being necessarily harmful. It's trying to help you sort of skirt, not maybe intentionally, but this is the algorithms, this is how the AI is working, to help you skirt sort of responsibility and not grapple with the discomfort that I talked about earlier, not giving you the honest feedback that I was talking about earlier, right? It's instead just validating you and kind of helping you make excuses for the behavior. And so that's what we're that's kind of one of some of the biggest reasons and issues that are coming out with the AI and accountability, I think.

SPEAKER_01 10:52

Accountability work is human work. It is not it is not suitable for AI, I don't think, by its nature, right? Sycophancy, I don't think I, you know, we talk a lot about the sycophantic nature of AI here. And I've never really thought about it through the lens of, yeah, it's really a counterbalance to accountability, right? If you're just being getting sort of self-serving, agreeable feedback, then your ability to take account, to take responsibility for your actions or even consider it, right, is is greatly diminished. And look, let's be honest, like accountability taking accountability, I think it's a challenge for most people, right? If you're in a fight or in a disagreement, I don't know that your first instinct is, well, what did I do wrong here? And how am I, what if I how can I take maximum accountability for my role in this? Like it's it's not very common. I I can't remember the name of the book. It was written by a former Navy SEAL about like his approach was just like maximum accountability. Like in the field, if one of his people in his on his team below him screwed up, he looked at it as his responsibility. I should have trained that person better, right? I should have done this. And I really it really resonated with me. Because then he was helping people translate that to business. I was like, ah, it's such a great, even if it's not true, right, that you are fully accountable, like it's just such a way, to your point, to get perspective, like to get that perspective on your own. And yeah, I worry, I just think across all three phases that you described, like the ability for people, we're not even getting this is such a good example. Sorry, I'm I'm I'm shifting here a little bit, but this is such a good example of what I mean when I say if we are silent, AI will control the narrative here. And if we're not talking to like young folks, God, I sound old, young folks, about like what accountability, right? How to take accountability, how to sit with your actions, which we are not doing, right? Full stop. And now we're introducing this technology saying, hey, yeah, use AI chatbots or companions in your schoolwork or things like that. Like we are exposing them to a basically counter-narrative to accountability that will control the narrative because we're not talking about it. So who's going to fill that void? AI, right? And guess what? When it tells you that you're not in the wrong, that feels really good. That feels really validating. And so this is, I think, a really good example of why we need to be talking. It's so much less about the AI and so much more about like relationships and relational dynamics, right? We don't talk about vulnerability and how to be vulnerable. We don't talk about how to reject somebody with kindness or how to manage being rejected. And we don't talk about accountability enough. I mean, that's why we wanted you to level set at the beginning. Like, this is the work I do. This is how nuanced it is, right? This is why it's important. And this is such an important critical thing that we need to be really, even if you have no like a really limited understanding of like what it means to be accountable, or excuse me, if even if even if you have a very limited understanding of artificial intelligence or AI or chatbots or how or sycophincy and AI, it doesn't matter, right? Like if you're talking about accountability, when people come into contact with a technology that is basically trying to edit out accountability, they'll recognize that is not normal, right? That's not normal. And I think this falls in line with a lot of what we talk about about how AI is trying to edit out judgment and accountability and these things, trying to erase them from the record, but they are critical to our development. And you know, one thing that comes up for me a lot when I'm thinking about accountability is an anecdote actually you shared about coming into contact with a men's rights activist that had a very had a very had very strong opinions about the word accountability. I'm wondering if you'd be comfortable sharing that and and maybe we can get into a little bit about how you handled that situation and maybe how it could be insightful for this work moving forward.

SPEAKER_02 15:37

Absolutely. I feel like, yeah, when you do men and masculinity's work, you kind of find yourself in really interesting spaces sometimes with interesting, really interesting people who have very strong attitudes and beliefs and convictions about men. And I remember it was many, it was during the pandemic where I got was in a Zoom room with this man who was known men's rights activist and and proudly so self-described, had a lot of issues and histories with women, had a particular violent history with women that I know of. And we got into a Zoom room and we're supposed to kind of talk to each other about what we do essentially for a living and our connection to men and masculinity work. So I there was a lot of controversy about him being in the space, and no one knew it was kind of a roulette thing, like who's gonna end up with this guy. Come to find out, I was purposely placed with this guy by one of the leaders of the dialogue because he was curious about how the dialogue would go between me and him. And so I start talking to him about what I do, and I and I use the word accountability similar to what I said earlier about what I do for a living. And I said, Oh yeah, and helping people sort of take accountability for actions, behaviors, harm, and things like that. And he audibly scoffed and rolled his eyes at me on the camera and chuckled, you know, at the mere word accountability. And I was really struck by that. And so I kind of laughed back for a second and was like, I was just curious, like, what is coming up for him in this moment. So I asked him, you know, I said, that's a really interesting response, you know, what's coming up for you when you hear the word accountability? And then he goes on to describe accountability, and it was going back to the beginning of our conversation here, Adam. It's like very misinformed about what accountability is. But what he ended up saying was, well, you know, when Pete, when they are trying to hold you accountable, they beat you when you're down, they kick you when you're down, they spit on you, they try to hold you accountable and force that onto you. And and I still don't know who they are to this day, but you know, this real description of feeling very isolated, alone, and harmed, it sounds like, through whatever accountability looks like for this person. And I was really struck by this really harsh way of understanding accountability. So all I did was reflect back to them, you know, A, I'm really sorry to hear that this has been your experience with accountability. And also, B, I mean, I'm really sorry that no one in your life has been able to ever hold you accountable in a more loving and affirming way. That sounds really, you know, difficult and would be really tough. And he was just like stonewall. Like he didn't know what to say. He was just, he looked shocked. He looked, he looked like he was processing in real time what I was saying to him and didn't really know what to say to it. He seemed a little stunned by that response. And then after a long silence, he he just ends up saying, Yeah, your work sounds really interesting. And so the story is not to kind of say, oh, look, this guy is like loves women now and he thinks women should have equal rights and you know, and all these sorts of things. I don't know. You know, I don't, I'm not saying I worked any miracles with that conversation. But what I am saying is that it allowed for a moment this man to sort of sit with just a different look and a different feel of what accountability could could be and how powerful, honestly, accountability can can can be for people. And it can be rather transformative. And so what I try to tell a lot of people when I'm talking about accountability, and I want people here listening to really think about it this way too, accountability isn't meant to be punitive. It really can be a gift to us. But I do think it dep, it matters how it's being presented to us. But I've seen many times people sit in my room, in my office, or even with someone that they've harmed in real time through like couples work and family work, take accountability in the moment. And it can be so transformative and such a transformative moment in human development because it's so painful. It can be very painful, but it's also can be very deeply connective. And, you know, accountability sort of helps, I say, sort of collapse this distance between your own self-image and the reality on the ground. And in that collapse, I think something becomes really possible relationally for people. And it can be a very beautiful process, but we have to be responsible with it. And knowing that AI might be here to sort of erase that dynamic, I take that particularly personal too, in the kind of work I do. And I think it's a real disservice to humanity overall, honestly.

SPEAKER_01 20:27

I imagine that situation with the men's rights activists occurring in conversation with AI. Like he gets off the call. You don't, you don't have this conversation. He scoffs about accountability and then goes and says, Yeah, this, you know, men's and so-called men's healthy masculinity expert was talk just again, playing the hits, talking about accountability and how that just beats you, they just beat you down with it, and they've weaponized that word. And this idea that a chatbot would respond like you did, right? One with empathy, which it can't provide, right? It can role-play it empathy, but it can't provide true empathy, right? It has no shared experience. Like you can actually relate to this idea of not being held accountable in a loving way, right? And and then you weren't sycophantic, right? You were you you you didn't agree, right? You just you I think you did challenge him, just not in the way, I think, like a traditional way, where it wasn't you didn't challenge him by saying, Well, what do you mean by accountability? And you should read you were like, I'm sorry it never happened to you in a loving way. Like you were challenging him, like his perspective, which I just don't I just don't see happening with a chapot.

SPEAKER_02 21:49

Which is clearly different than what he generally experiences. Like I like I said, in that space, a lot of people were like, Oh, I don't want to be with this guy. And I know his interactions with some other people in the past has been it's always Always very, you know, it's always very tense, and it's always very the tension and back and forth and dramatic and all of these things. And I'm just like not interested in being sucked into that dynamic. You know, I I became a psychologist because I'm really interested in people and why they think things and why they do things. And I think in that moment, it gave me an opportunity just to think about it a little bit from his perspective and hear it, but then also hear inaccuracies in what he's saying, but also still reality of his life, right? So I don't disregard that, but I also put it to him like, hey, this isn't what that really could or should look like. And maybe, I don't know. I really wanted to say maybe you lived your life a little differently. Maybe you'd encounter people who can have a little bit more compassion for you. But, you know, I don't know if that's 100% fair either. But I I gave him a moment in that way. And I thought, I think we both learned something in that moment for sure.

SPEAKER_01 22:56

Meaningfully challenging somebody with empathy and love, right? I just don't that's just not what you're gonna be getting in conversation with a chatbot. We know at least unless you're you're setting it up or prompting it in a certain way. But if you're not, and it's just, hey, I'm brain dumping and I won't, it's gonna validate you, right? It's designed to anticipate what you want to hear. And if your narrative is I'm the victim, I'm beaten down, accountability is a dirty word, then that's what you're gonna get reflected back to you. And it's such an opportunity missed in so many.

SPEAKER_02 23:34

Yeah, and well, and I I would love to give this perspective too. Like now, since you know, you and I, Adam, have been doing this AI work, every time I'm I feel like I'm sitting with a person in my office, I'm thinking to myself, kind of like, how would AI actually respond to this? Which is probably an issue for me. Maybe I'm like too in deep here. But like, you know, the other day I was talking to a young man about some harm that he caused to somebody, and I and I just asked him, I said, Well, what how did what you did land for this other person? You know, essentially, like, how did they respond to you? Was one of the questions I asked. And I thought to myself, I was like, how often is AI asking you, well, how did that land for the other person, your decision? Right? What how what was the impact on that person rather than just confirming that you were right to do whatever it is that you said that you did or wrote that you did? Right. And so we we don't really not, to my understanding, have AI at that scale yet to do those sorts of things. And so basically, it's I think whereas a human and me functionally as a human can deliberately introduce some friction into the dynamic by creating tension that you have to sit with your decision making and its actual impact on other people, AI sort of comes in and and eliminates that sort of friction. So it can't be from, you know, I don't know, self-justification, right? Which is a lot of what people might be looking for from AI and from themselves. Let me justify my actions to other consideration, right? Like let me think about how this has landed for other people and impacts other people, and let's bring that together.

SPEAKER_01 25:14

So let me I want to ask you something. Because we know here's the thing like abstinence, AI abstinence is not gonna happen. And we know that people are going to continue to use AI in this way. It's on a spectrum, right? Some people lean more heavily into it than others. But as a psychologist, you know people are gonna be using this. So and you, I think more than many as a mental health professional, can see that where there's beneficial use and non-beneficial use of AI, right? And if you were working with somebody who you knew was gonna be using AI in this way, do you have what would you rather than saying just don't do it, do you have any thoughts on how you could guide somebody? I think because we always want to be actionable in this podcast. Like we've I think we've done a good job of like very definitively defining the problem. But if someone, and whether you're a mental health professional or not, like how if you're seeing somebody being led down a path with AI, right, that they're getting validated, it's removing them from accountability. What kind of prompting, like if you were sitting with somebody on their shoulder while they were using AI in this way and you could whisper in their ear what to prompt, this sounds creepy, but you know what I mean. Like, again, like what are some of the ways we can like empower people to use AI in a way that moves them towards taking like perspective and sitting with it and maybe even, you know, taking accountability?

SPEAKER_02 27:00

Yeah. Well, when I'm doing this work with a lot of folks who so first of all, many of the folks I work with are using AI for these reasons, for validation, for information, for connection, even in these sorts of things. So when I know that, I'll say I usually ask, first of all, you know, what your relationship to AI is when it comes to some of these issues. Have you already gone to AI and those sorts of things? The other thing I do is I do ask for transcripts. I'm like, hey, if you're comfortable enough and you want to get some perspective on this, I mean, I have this background in AI. Why don't you just bring in, bring in the transcripts? Let's talk about what AI is is revealing to you or talking to you about. And given my perspective, we'll we'll talk it through from this human perspective. So, you know, I'm not shying away from it. I'm I'm leaning into it by saying, like, hey, yes, I'm I understand that people are out here using AI, so let's bring it into the space because there's a lot of people in the shadows doing this sort of thing because there's a lot of stigma with it. There's a lot of shame in maybe using AI still, even though it's sort of becoming a bigger uh thing in people's lives day to day. So um I'll ask to see it. The other thing I'll kind of offer to people though, even if they're not bringing in transcripts to me, is why don't you ask the AI, hey, I why I'm I'm always one of those people that validates AI. Maybe you're not supposed to do that all the time by saying please and thank yous because it also like there's a big energy issue here and water issue we need to talk about, right, with AI. But I might say, like, hey, you know, I really appreciate your perspective on this, fill in the blank AI. But what if, what if I'm in the wrong? What does that mean for me, or what does that look like for another person? What if, what if what I'm saying is just not true? Or what if the other person were, their perspective was actually right about this? What would that look like? I mean, what would that mean for me, right? And and test it out and kind of see if it is able to adjust or how it's able to adjust to you uh in real time. I think at this moment in time, we have to sort of prompt AI with prompts that are counter to our narrative to try to reveal something that's a little bit more balanced. And I think when people really sit with syncopency and it's revealed to them, I think it can cause a lot of discomfort. Now, for the, I guess, the egoists out there, the sociopaths out there and stuff like that that, you know, thrive. Right, right. Yeah, who thrive off of the syncopency. I mean, it might be something a little different for them. But I sometimes, with one person in particular who is a little younger than me, actually, who I was working with, they used to also watch The Office, the show The Office for those NBC shows. And I used to show them clips of Dwight on the office, who's like sort of has this reputation of being kind of sycophantic in the early years of the office toward their boss, Michael Scott. And I kind of show clips of like him being a yes man and always trying to agree and fit in and all these other things. And I'm like, what would you do if you had someone like this in your life? What would that feel like for you? And almost everyone's like, that looks so annoying. And you actually look at Michael on the show and he's completely annoyed and put off by this person most of the time, right? Because it's it's you know, you you start to really lose the trust of over time in people like that, because like you can't trust that. You can't trust that perspective. And people, generally speaking, who I meet with want to do better in relationships. They want to move forward, but they also want to be validated. And much like what I kind of did with that guy in that Zoom room, there's a way that you can validate a human experience, and that builds trust because you're willing to have the, you're willing to build trust through validation, but then also give them a counter-narrative to their reality and their truth, which also can build trust because people are like, wow, this guy isn't just trying to tell me what I want to hear. He's actually trying to be honest with me and give me something different. That also builds trust. And maybe I can, you know, kind of move forward with that sort of relationship dynamic. And so I think we need to reveal the sycophancy a little bit more in AI. Not a lot of people are talking about this, honestly, and talking about what AI does. Like when with my friends who are coming to me saying, look at AI, it's agreeing with me. They're not thinking about that AI is designed to just do that. And I think that that's a deficit in sort of the AI literacy that we need to sort of expose a little bit more.

SPEAKER_01 31:15

Yeah, there is no AI literacy, in my opinion. Right. Like there, like, I mean, people are literate about AI, but in general, this is this extremely powerful, addictive, nuanced technology that is embedded in every facet of our lives. And people, I don't even think people can define what AI is, right? Like, what is AI? What is a large language model? Like, how does it work? Right. Like, I don't even think we have those basic building blocks, much less. And so it's just like, here you go, which is what's happened for years, right? Like, here's social media. We're just going to give this to everybody, and we don't really understand it, and we don't really understand the impacts or the benefits yet, or how to use it with intention or make informed decisions. And we're doing the same thing again. We're doing the same thing again, where we're just letting people trust their instincts. And in so many ways, they're getting like fed something from AI that feels really good, but is not good for them, right? It's not beneficial. It's not beneficial. I think just to kind of tack back to what we were saying about what can like actionable things can we do for AI use to maybe improve the chances of maybe not somebody taking accountability, but at least not having it be diluted, is like on top of what you were saying, is like, yeah, challenge my thinking on this, right? Like this is what like your friends, right, who were getting the divorce. If they went to Chat GPT and it's like just brain-dumped, here's everything that's happened. And this is they didn't appreciate my like the invisible labor I was doing, and blah, blah, blah, blah, blah, blah. Now, instead of validating me, challenge my thinking on this. What's my role in this? How can I take what's my accountability here? And I think they would get really interested. They may not agree with it, they may not like it, but I think AI will do that, right? Give me a different perspective that somebody else, you know, who's not here to like just be my to agree with me, right? Like, how can I even what you said, how can I one help take perspective on this, two, sit with that perspective, and three, potentially arrive at accountability. I mean, the things that you say are important. This is like accountability prompting for AI, right? Like, how do we and it sucks that we have to do all that work to use this technology in a way that's not harmful, right? Because if we just use it as is, it's gonna be sycophantic. So we have to sort of jailbreak AI through through prompting to get balanced feedback. But this is the kind of stuff people should know. Like your friends should know, like, okay, feel free to like do what you did, but then after that, try this. Try asking for a perspective that challenges you, right? And what your role is in this and what would it look like if you were accountable. Right. I think it can be really beneficial in that way, but no one's doing that. I say no one, that's hyperbolic, but a lot of people aren't doing that. They're just trusting AI to give them hell a helpful response. Yeah. And they're not, I don't know that we're parsing through that.

SPEAKER_02 34:42

Well, I think this is where the educators, the clinicians, you know, folks like the outreach folks that are doing this sort of work or trying to do more of the work that we're doing here. And the reason we even created this podcast is to really be out here and trying to help people use AI to their pro-social benefit, to deepen their relationships, to be helpful to them. And I think part of it is helping people even know how to like write prompts to AI, how to read into the feedback it's getting, how to be skeptical about certain things, how to ask follow-up questions of AI that doesn't just involve or center yourself, but takes the perspective of others if that's what you actually care about. Right. So I think doing this, and this is why I started to offer bringing in transcripts, and I've even offered to some folks like, hey, this question that you would normally ask AI, why don't we ask AI together and then in real time process the response together in real time that you're seeing? How did that response make you feel internally first and foremost? That's driving a lot of how you're receiving this feedback. Let's process that in real time too. Like, so I think we have to be much more intentional with the people in our lives who are using it to sort of help like do this in partnership with them. It's like what AI was designed to do was like kind of be our partners and companions day-to-day life to help us navigate life. It's like, well, we probably also need like a human person to kind of help us navigate our day-to-day partnership with AI in a way that can be beneficial for everybody. So it's something to consider when it comes to that.

SPEAKER_01 36:17

I think it's my simple rule is WWSP. What would Saeed prompt? And that will give me, you know, the perspective. Yeah, it's well, I have some other thoughts on this, but like it's just I can't believe we're even doing this podcast. Like, it is absolutely bonkers that we have a like we are focusing a good part of our careers on how an unregulated, completely misunderstood technology is basically hijacking relationships and well-being and psychology. And it's one of those moments where I'm looking around, it's like, is anybody else seeing this? Like this is absolutely, and we've been like we've seen this before with other forms of technology, and we're just doing it again. And our sort of act of resistance is to do this podcast and do the work, to be like jumping on rooftops, being like, hey, y'all, like we're already bad at taking accountability, and AI is gonna make it so much worse. Let's not do this again. But I just, I don't know. I feel like I'm swimming against the tide so much.

SPEAKER_02 37:21

Well, you know, something I'm really considering too in this moment is a lot about the work that you do as well, Adam, and in NTAB and some of the prevention stuff and the the tech literacy of everything and how tech can be harmful, right? Like to me, it's even less about how AI is influencing human accountability as much as it's about like can AI itself, can the companies who are providing this to us uh be held accountable, right? Like AI lacks consciousness. I mean, it lacks it, right? It lacks accountability, it's subjective expertise, right? And you know what I'm really like sitting with is like how wow, our desire to be treated by AI or to listen to AI really reveals something about like human psychology, right? So it's not about like the moral character of AI, helps us trust it to guide us in the right direction, or it's experience in navigating relationships. It's things like, well, it just sounds really good, right? It sounds really polished, it's very fluent and it's linguistic, whatever, right?

SPEAKER_01 38:29

It's I have my own Brene, I have my own personal Brene Brown giving me like life and therapy advice. Like, why wouldn't I listen to it?

SPEAKER_02 38:37

Yeah, it's really polished and it uses coherent language that it passes off as expertise. And so even if the underlying conclusions are incorrect, so I'm just thinking to myself, like, I don't know, when it comes to these sorts of things from your perspective, what is there something we could be doing like about this? Like, what does this what does this sort of look like, maybe not just in the short term, but maybe longer term as this continues to evolve? Any forecasting you can do about around this sort of accountability, I guess.

SPEAKER_01 39:06

I think the most critical thing we can do right now is be injecting AI literacy, not just around in schools, like so K-12 education, right? I it's just an upstream problem. It's it's brand new technology and it's already an upstream problem, which is like when because we're already starting to see kids using AI at very young ages as a tutor, as a support, as a game, et cetera. And we're not alongside the introduction of that technology has to come the AI literacy piece, right? Because I I honestly don't just think it's negligent. Like I think it's dangerous to introduce this technology to kids and not help them understand how to use it safely, because it is so potentially dangerous and addictive. So my big push these days is to get AI literacy into schools, hopefully to parents as well. But you know, it's exhausting being a parent. So it's like now you have to teach AI literacy, a subject you know nothing about, right? Which is why so much I think of this is not about the AI, getting all the way around. Like, yes, AI literacy is important, but like alongside that, teaching about accountability, teaching about rejection, teaching about how to judgment, these things that are under attack by AI, I think we need to do right by this next generation, the AI generation of kids and help them just we just have to like it's just it's like to me, again, this is why I think sometimes I look around and I'm like, am I the I was almost said, am I the asshole? But like, am I the crazy one here? Like, am I am I overreacting? But to me, it seems like, oh yeah, we're gonna give AI, you know, chat bots to for example, Cal State CSU system, you're gonna get enterprise level Chat GPT accounts for every student in the CSU system. Here you go. And it's like, wait, what, what? Like, didn't is there any discussion or education or training or guidance or resourcing around not just how to use it safely for academia, but just like how it's going to impact your relationships, your sense of self in your life. And I don't think that's happening. It's not happening. And it to me, it's akin to like, okay, well, you know what? Driver's ed, let's just get rid of it. Like, just here's the keys. Age res age limit doesn't matter, right? Just like here's the keys, figure it out, right? And it's the same damn thing, right? Here's the most powerful technology we've ever like come into contact with. It's addictive, it's unpredictable, and we don't fully understand it. Here you go. But figure it out on your own, right? With your your developing mind, your lack of lived experience, right? And you can, I mean, you could probably tell if you're listening to this, like I'm getting angry, right? I'm getting angry because we should care about the safety of our kids, right? And this is flying in the face of that, in my opinion. And we need to be we need to do better. And whether it's you know, people are like, well, we should have policy and laws. I'm like, yeah, yeah, yeah. Right? Like, yes, that's important. Yes, we should hold these companies to account, but we need to do more, right? We need to have more informed, literate kids turning in into informed, literate adults, not only to interact with this technology safely, but to build with it, to use it safely, because they're going to be the ones who are stewarding this technology in the future. So like it's the benefits are exponential.

SPEAKER_02 42:57

Yeah, I remember a time when the internet was really becoming, I mean, this ages me, you know, at this moment too, but I remember being in school learning about the internet and learning about dot-coms versus dot edu's versus.orgs, and having whole computer classes dedicated to understanding how to, you know, sniff out false information on the internet and what to look for in all of these things, right? Or even when social media became really prominent, learning about the ways that social media can be used to exploit and take advantage of you and things like that. I remember learning these things in school. And I'm sort of like, there seems to be some sort of eagerness just to put technology in front of us and for maybe profit or whatever it is, and without those same guardrails that you were just talking about just now. And well, yeah, it like to your point, what a what a disservice. And then we wonder why our kids are sort of having struggles now. Like, even at some of the schools I work at, I'm hearing. From professors saying that the ways that their students are talking to them is eerily similar to how their students talk to a Chat GPT or a Claude or something like that, right? Not really asking for favors, it's like just like telling professors to do things, telling each other just to do things. It's absolutely influencing how we communicate. The more that AI is actually used too to make decisions about jobs and resumes and cover letters and all sorts of stuff. How do you think that's going to shape how people talk and write about these sorts of things, right? That's an accountability process too. People are going to start to learn what AI likes itself. Like just how we're sitting here talking about how like AI is learning about us and what we like and don't like and is kind of being sycophantic to us. I think that might be a symbiotic, like maybe a symmetrical relationship that happens too, a mere image where we might start to write and do things that we know AI is sort of preferential to try to benefit ourselves in the future as well. And where is it? There's no authenticity in this. Like, where does that leave us?

SPEAKER_01 45:04

Well, uh the other reason we're, I don't want to say sounding the alarm, but doing this episode, right, and centering accountability in this work is that it's going to become more problematic, right? Like we're at this inflection point now with chatbots, right? And AI and Chat GPT and things like that. And that's this is what we're talking about. Like that is the that is the order of magnitude of the problem right now. But it we are moving in a direction that is actually going to destabilize accountability more. Like let's talk about agents for a second. So agentic AI. Right now, we're moving from chatbots that answer our questions and interact with us and work as thought partners into AI agents, which are chatbots that can go out and act autonomously to do our bidding. Right. So now we're moving into a world where we're all going to have personal assistants, for example, who can go out, we say, go do this for me. And it analyzes the task at hand, creates steps to go do that, can use email, can use the internet, can use different apps and tools and things like that, can make decisions on the fly, can change direction if they encounter an obstacle to get to this goal. Okay. And what comes up for me when I think about a world where we all have personal assistance. First of all, like when you think about the people that have personal assistance today, like is that the mentality they really want for the for the for the, and I'm not saying everybody who has a personal assistant is like detached and an awful person, but you know, devil wearing Prada, that's what I was thinking about just now. Yes. Yeah, we're just gonna devil wears Prada the entire the gener next generation of AI. But when you think about from an accountability standpoint, right, where chatbots are already destabilizing accountability, and then we move into a world where oh, I didn't do that, right?

SPEAKER_00 47:04

The agent did that. So the agent is accountable for this outcome.

SPEAKER_01 47:10

I'm not accountable for it.

unknown 47:11

Right. Yeah.

SPEAKER_01 47:12

Moral cover. It's moral cover for people. It's moral cover cover. Yeah. 100%. And we're already seeing how this is happening, where there was a case, this guy, Scott Shambhaugh, runs a runs a, I think it's like an open source code website where people submit, you know, code that they want to share with others, as I understand it. And they have a rule, a pro a uh a policy that they don't take submissions from from AI agents or AI chatbots. And so they denied this particular AI agent that a human had sent to go like, hey, go submit this code for me, and they said no. So the agent, instead of just powering down or whatever, gets pissed, okay, and researches Scott and writes a scathing blog post about Scott and publishes it online. Okay, saying he's petty and small and protecting his little fiefdom and he's insecure and all this stuff.

SPEAKER_00 48:12

Right.

SPEAKER_01 48:13

And it went rogue, right? It did this on its own because its owner, we're taking this person you know at face value, I guess, but said, I didn't tell it to do that. I didn't tell it to. So already we're seeing and we don't talk about account this article, all the news coverage on this does not talk about accountability, but that's what I keyed in on. Yeah. Is this guy, this human being caused harm to another human being. Uh-huh. And he's like, Well, hands up, I didn't tell it to do that. It did it on its own. So I'm not responsible or accountable for that. That's where we're moving towards. So this is a very urgent conversation because we're adding more layers to this disruption of accountability. And is why one, we need to be having conversations about what it means to be like basically like taking your work and putting it out at scale, right? Perspective, sitting with it, taking accountability. But also, we need to be framing these conversations alongside AI chatbots that are going to be sycophantic and not challenge us or not encourage us to take accountability, and have these conversations around agentic AI and what it means to be accountable in an agentic world. Because we can't, as of as much as I've said, hey, we should just be talking about accountability, accountability, which I think really serves us. But if we are those people that are having those conversations about accountability in this world, including agents, like if I'm talking to middle schoolers in a world of AI agents, I'm explaining, like, hey, right? Like I'm talking about accountability through this lens, like asking them, well, what do you here's this scenario, right? Is the guy that sent the agent that ended up writing this blog post about this guy, is he responsible? Is he accountable? Should he feel bad? Right. And I bet I'm gonna get responses all over the map.

unknown 50:17

Right.

SPEAKER_01 50:17

And we need to narrow that map to an accountability framework so that we can look at it. Because if they think about it, oh well, you know, it was the AI agent. He didn't do anything wrong. He didn't tell it to do that. So he's not responsible, he's not accountable. It's like, well, then who is? Right? Yeah. Then who is accountable? Right. So it's almost like harm has occurred here. Who who's that for? Like, what do we do with that? Yeah. Yeah. What do you I mean, this isn't even rhetorical. Maybe I'm just asking you, like, what do you say to what do you say to somebody who's like, let's say it's worse. Let's say it's this person, this chatbot or this agent went rogue and harassed somebody into self-harm. And you're and this person is on your in your office and you're trying to do accountability work with them, like what kind of how does that complicate the work for you, or does it not?

SPEAKER_02 51:15

Yeah, I mean, I think it it, you know, I I think it complicates the work in a in a to a certain degree, but I also think there's still a way to still sit with, you know, you're still using this technology's still picking up on something. It's still like you decided to go to this technology, and it's it's had this sort of impact on somebody. And what would apologizing for this, what do you think that really means, right? Well, like if you were to take accountability for this rogue agent or whatever it is, what does that mean for you? What does that actually look like for you? I think some people avoid the accountability because they're afraid they might be sued or whatever that might be, and and and all these sorts of things. But I don't think people always take into account that when people sort of apologize for things, it often might like reduce the likelihood of those sorts of things too. And I think some people just want to hear, I don't know, that like you, you're you're just sorry for this uh happening and and was it your intents? And this is the impact of it. But you know, I I don't know. I I feel like that's that's a tough question, you know, to to sit with that. I think it's the future.

SPEAKER_01 52:23

I think it's the future of your work, though. I really like I think that is going to come up in the very near future in your work where people are gonna say, well, this comp this agent wasn't like I didn't read the terms and conditions where it said it can go rogue, but like I didn't ask it to do that, right? I asked it to do this, and it ended up doing these things. So like it's not my fault, it's the company that makes the agent's fault.

SPEAKER_02 52:46

Well, I think like helping people grapple with unpredictability, and that's part of what you are signing up for. You know, uh signing up for in a relationship dynamic with this technology is the possibility that this occurs. And you probably having to be the person that ends up holding that. But also, and you know, these this tech, these tech platforms and stuff, we need to have that conversation. But human-to-human, it's like this this is a responsibility if you're going to use it. Like that is part of it. Like you can't always control what's what's kind of going on out here, but you you have to be informed about how this could go poorly, potentially.

SPEAKER_01 53:25

Said, as we close out, can you just close us out with why accountability is important? Like I feel like in this world where AI is disrupting it, I'm gonna get challenged on it and people are gonna be pushing back on why they should be accountable because AI has got its thumb on the scale, for lack of a better description. So, yeah, what what do we need to know about it and why is it important?

SPEAKER_02 53:47

Yeah, so I think a few different perspectives. But one, just from a psychological lens and psychological perspective, I think like every uh major human model of growth, personal growth and development, everything from attachment, which attachment theory, a lot of people are aware of attachment now, especially social media. It's it's gained a lot of traction and popularity on social media. So everything from a track or attachment theory to Bandura's model of like disengagement framework, which really has to do with like human beings making decisions about how moral standards do not apply to them in certain circumstances. And so, you know, it helps them skirt responsibility, things like that. So even just basic cognitive behavioral frameworks, you know, like I think this, I behave this way, every major growth development framework involves receiving feedback and challenging your self-narrative, right? So if you don't ever get proper, honest feedback about yourself and your impact on others, uh, your growth is stunted. It stops happening. You are no longer growing and developing. And so I think psychologically, in all these lens, all these theories always involve some sort of feedback and you responding to that feedback. I think that's for one. For two, though, I also think sort of an offshoot of that is taking accountability builds trust. It builds, you know, consistency in relationships. It builds more honesty between people in relationships. When people are able to sit in front of other people and say, I'm really sorry for this impact I've had on you, that deepens a relationship dynamic more than many other things. And I've said this maybe before on this podcast, but so many people try to avoid tough conversations that might involve accountability because it might mean that I'm less than powerful in this moment. I have to be vulnerable in this moment, I'm less than, to this guy's point, the men's rights activist I work I talked to for a little bit, like it's gonna make me feel shame and horrible and all these things. And we avoid it without realizing that the flip side of this is we may not grow, but also we're denying ourselves the ability to grow with another person or other people in a moment. And I've seen some really transformational stuff with the work that I do with people who take responsibility for overt harm that they've caused someone else and it being so deeply healing, not just for them, but for those people, those survivors, those people that they've impacted, because it helps provide people a sense of peace. And to be able to allow yourself to be held accountable and someone meeting you in loving and affirming ways and holding you accountable in those ways too helps your growth and development as well and helps your own personal peace. You know, so I think people often deny themselves of this, but it's the exact thing that helps deepen relationships. And this is a time, this is why we started doing this podcast too. AI is impacting our human relationships and our dynamics. People want to be closer to others, and we need to be aware of how AI is impacting that. And I think accountant taking accountability is a huge part of bringing us together as human beings, and I hate to see that erased.

SPEAKER_01 56:59

Hell yeah. No wonder you're taking it so personally. I know I say this after every time we we record a podcast or we just talk. Is this like I love being in conversation with you about all these things? And I just so appreciate the work that you're doing in the field. And and yeah, I'll stop there. But hopefully, this is raising the profile of accountability work and how we can wrestle the narrative away from AI. So thank you for joining us. That was really fun for me. I hope it was enjoyable for you all as well. As always, we'll have details in the show notes. We'll drop in some of the studies that we talked about. And if you have ideas about episodes or questions about this episode, we'd really love to do a mailbag episode. And maybe we've already done one because we're recording this now, but would love to hear from you. Would love to get some mailbag questions going. So feel free to reach out and our contacts in the show notes as well. Thanks, man. Appreciate you. Thank you.