Episode 3: Shadow AI is Already On Your Team

Brian Milner (00:22) Welcome in everyone. We're back here. We're on episode three now, if you can believe it or not. We're at episode three of the People Over Prompts Podcast. This is the podcast where we talk about the future of teamwork and AI and how that all melds together. I'm your host here, as always, Brian Milner. And today I have another good friend of mine that I'm I'm happy to introduce to this audience if you haven't. cross path with him yet. His name is Mr. Lance Dacy. Lance, thanks for coming in. Lance (00:56) Absolutely, Brian. Wouldn't miss it for the world. I'm glad that you're ⁓ you're coming back on with these new topics. It'll be so much fun to to talk about these things. I mean we go way back on this stuff. It'll be nice to dig up some of that dirt. Brian Milner (01:07) Yeah, absolutely. I mean it's it's funny, it's like we've been having these conversations anyway, and I I think that's kind of very reflective of our topic today, and also reflective of kind of how society is at large with these topics, is that the they're happening, the conversations are there. They may be happening in hallway conversations and they may be happening kind of on the down low a little bit, but they're still happening. And so that that's kind of why we wanted to have Lance to come in. ⁓ by the way, Lance is a certified Scrum trainer like myself. ⁓ Lance has its own business called Big Agile and teaches classes just like I do. ⁓ and I'll put links to his ⁓ stuff in our show notes so you can find him afterwards. But we wanted to talk about what what I'm calling shadow AI here, or I I guess I'm I'm not calling it, others are calling it this as well, but really about this whole idea that shadow AI is already a part of your team. And I'll kick it off because Lance, we were talking about these stats. There's a couple of surveys that that came out that really are are eye-opening. ⁓ there's one from PagerDuty that was published this year that shows two-thirds of office professionals have used AI tools at work, even though. They believed it was not permitted by company policy. Two-thirds of the workforce are using AI tools even though they believe it's not allowed by their company. What do you make of that stat? Lance (02:45) Well, I I believe it. I mean, you know, just to go back to what you were saying on the introduction, you know, we've been solving problems for years, right? Let's just call it that. we typically solve things with the ⁓ the agile mindset that a lot of the work that we do, people don't know what they're doing. It's never been done before and look where we are right now. So it's like we just continue to have these common problems. What I love about us though is we don't necessarily have to focus on tools, let the teams kind of decide the tools for that. But I find it's ⁓ entertaining when I go into organizations and they're like, yeah, our leadership doesn't let us use AI. Well, I believe that stat sixty sixty-six or whatever percent are using AI despite believing it's not permitted. And and I think what's other another interesting point to that raises to almost 75% at 1,500 and plus employee firms. So when you have over 1,500 employees, that's even worse, obviously, because it's gonna scale with that. But I believe it. what about you, Brian? You believe it? Have you seen it? Yeah. Brian Milner (03:47) Yeah, I I have seen it and and I believe it as well. I mean, I I think it's sort of ⁓ I think there's a parallel to draw here. And I think the parallel is in the ⁓ education world. Because in the education world, there was sort of at least initially, there was this this feeling of ⁓ AI bad. And so we we kind of need to keep it at arm's length and that's not Lance (04:14) Or not trustworthy, right? Maybe not bad, but it's not like we just didn't have enough data to trust it either. That was the other problem. Brian Milner (04:20) Right. And the fear, obviously, from educators was, my gosh, my college students are now going to just have AI write their paper wholesale and they're not going to learn anything from my class. ⁓ and and so there there kind of was this initial complete backtrack against it to say, if you use it, you're you're you're not being ethical and you shouldn't use it at all. And I've seen that shift. I don't know if you've seen this as well, but I've seen seen that shift in in kind of the educational world where it's now there's an acceptable use policy. And it's not that, hey, don't use it at all, but it's within these confines. It's got this very kind of ⁓ defined set of circumstances to say, look, it's good to help you plan, it's good to help you research, it's good to help you organize your thoughts. Don't tell it to write the thing for you. That's gonna be obvious for one. And for two, you you know, you're not gonna demonstrate your knowledge in any way. So that's gonna kind of defeat the purpose of what it is we're doing. So the boundary there is don't use it for that, but do use it, use it in other ways because it's it's advantageous to you to it for the workforce to be able to use this. Lance (05:21) Well. The other thing is we don't have to get off on this tangent either. But if there is an instructor or a PhD professor out there that's saying, my gosh, I can't ⁓ my students aren't gonna learn anything, they're not learning anything by writing a paper and doing busy work anyway. So guess what we're doing? We're taking all that grunt work and moving it out of their way so they can't actually learn. You realize this as a as a coach and adult learner, you know, more than just teaching classes, but Just reading something and regurgitating it, you're using like twenty percent of your brain. I think I read somewhere. I can't cite that study. So what do we do in classes? We force them to talk and collaborate about it, teach back, make a visual about it, you know, draw it out, talk it, make an infographic. you're using like 60% of your brain. So I think shame on the the professors out there that are scared that their students are just gonna use AI. That means your your ⁓ assignments fail. Brian Milner (06:03) Yeah. Lance (06:23) You know, your your assignment that you're given. It's kinda kinda like a recursive problem. So maybe that's a good thing 'cause we're highlighting that there hasn't been a whole lot of edge ⁓ innovation in the education industry. But that's probably another topic sometime, Brian. I'm sorry to bring that one up. But I Brian Milner (06:37) Yeah. No no no. I I mean I brought it up and I brought it up as a parallel 'cause I think it's No. Lance (06:40) I didn't mean to get us off. I I love professors. I'd love to be one one day, but I'm gonna tell you, my kids are in in high school and college, ⁓ abysmal instructors out there, and just shame on the system for doing that because people learn differently now. And I I'll just say, you know, that other study you were sharing before we got on the call was or Deloitte, the Gen Z and Millennial Survey. So I'd like to tie that into this ⁓ because that's really what the workforce is looking at these days, and you have to be able to upskill Brian Milner (07:00) Yeah. Lance (07:10) management leadership and all these other things to account for the different mindset that these people have. Look, I'm a Gen X person. We approach life differently than a Gen Z millennial. Why? Because I got them living at my house. I know that. But 74% of them from that ⁓ are using AI and daily work. They're not afraid of it at all. In fact they're they're embracing it. So man, that is something to sit with right now. That's what I think about it. I'm sorry. You're probably sorry you asked that question. Brian Milner (07:38) No, no, no. No, I love that ⁓ that that study as well. That that's one from Deloitte just this year in twenty twenty six. And you know, it does draw that that generational kind of gap there with with Gen Z and and millennials versus the older generations. And, you know, the the there's a tag onto the end of that. You know, seventy four percent are using it to some extent their daily work. And then it also says, and many feel they are adapting faster than their organizations. And I Lance (08:06) Yeah. And maybe learning faster too. The best way to learn is is have to construct things and using AI to do the grunt work, you can now apply that knowledge sooner. I think we missed a boat, or not us. I I just think some people miss the boat when they think we're taking shortcuts in education by writing a paper. Well then maybe the paper you're not learning much anyway. Brian Milner (08:09) Right, right. ⁓ Yeah. I I I I've shared this story before in class. I'm not naming where it's from or anything, because I'm not gonna shame anyone, but ⁓ there is an organization that I have a tie with at some point in my past that I remember hearing tales of the CEO of that company having his secretary print out his emails for him. And that was how he interacted with email was he would get printed copies and he would write his response and give it back to his secretary to to reply to. And that that seems to me a parallel here as well of, you know, there's there's there's always been sort of a generational gap in technology. And that when the new technology comes along I dun I don't know what it is about the human condition that does this, but I think there is a tendency, ⁓ as humans to Rely on what we've learned up to this point and feel like that's the safe zone. The stuff that's now new is is shaky. It's it's you know, thin ice. I don't wanna have to tread out onto that thin ice because this is stuff that other people know much more about. So or or maybe it's that I'm elevated high enough that I don't have to learn this new stuff and I'll just depend on everyone else to learn it and you know, benefit from it. But whatever the case might be. Lance (09:56) Yeah. Or you'd be a lagger, you know, you you embrace those things later. And I think, you know, when we go back, ⁓ let me just tout Gen X, you know, say what you want about these generational gaps, but since I am part of Gen X, I can look back and say, you know, we were kind of the last generation to play outside all day and then had to embrace the internet. Well, the internet was a transformational technology that was novel and you could either embrace it or be scared about it. ⁓ you know, my dad's generation They they had a lot of other kind of transformational work, but they approach that differently. I just think we have the sweet spot generation here where we typically embrace things and be like, well, that's we just adapt, right? We've had to adapt all the all along. And I think the the Gen Z and the millennia are used to having that information. They're not scared of it at all. They're like, how can I use this better? I mean, you saw in that study that 79% of a ⁓ 79% of those Gen, what was it, Gen Z and Millennial. Brian Milner (10:51) Yeah. Lance (10:52) They're using AI to find learning and development opportunities, ⁓ use it for career advice. They're not afraid of it. But organizations, can you blame them for saying, well, hang on, because if you owned an organization and had a whole bunch of people using AI that you didn't understand yet, and you land yourself in the courtroom, well, that's not good either. So we have to find, you know, the delicate balance here is really what this is all about is being cautious, but yet allowing for experimentation. Does that sound familiar? Allowing for experimentation. That's psychological safety and culture. We teach that all the time. Brian Milner (11:21) Yeah, right? Yeah. But but I on the other side I I still I I do sympathize with the leaders and the governance type areas of the organization because this is a brave new frontier and it's and it's it's changing. I mean, we were talking about this before we we Lance (11:33) Yes, absolutely. Brian Milner (11:43) press record. It's changing daily. And you know, like I I I'll I'll learn something today that I didn't know yesterday and I'll think, my gosh, I've got to go change my class because this is entirely new information and now the paradigm has shifted once again. And and so when if I'm a governance person in an organization and I'm trying to keep us safe, that's my primary job. ⁓ so how do I ⁓ Lance (12:07) That's your primary job. Brian Milner (12:13) If I'm if I'm not as well informed about this, then I'm gonna be like those professors early on this that would say, Hey, I'm gonna slam the door shut on this because there's potential for misuse and maybe I I can't differentiate enough to say where it's crossing the line into potential harm. So th the default position is always going to be move that border back further. so that now we have a buffer of safety and we may not be able to do everything that we ⁓ that could benefit us, but at least we'll be safe because we know from this point forward it's it's gonna be fine. Lance (12:53) And I think about even when we when we coach organizations on flow and removing friction and waste in the organization with processes, you have bureaucracy and processes for a reason. And I tell everybody all the time, if you were a leader, you would have them too, right? So the whole point here ⁓ is I do want to defend the leaders who ban this stuff. You know, they're the ones who banned it first. ⁓ because they're not necessarily wrong about the risk. There is certainly some risks out there. ⁓ And and ha you know, the technology is changing every day. Even you and I aren't experts on this. We're out here talking about it's going to change tomorrow. But when a third of your people are, you know, pasting financial documents into a a public tool, ignoring that would be malpractice, in my opinion. So ⁓ you know, sh kudos to the leaders, that's why they're leaders, is they have to look at the entire entirety of the situation, which I think a lot of us on the ground that are just working day to day kind of see that as it's like mom and dad telling me I can't do anything. It's like, no, they're in that position because they are respectful of the organization and the risks that are associated. But the the ban doesn't remove the behavior. That's the flaw in the thinking. It it removes visibility into it, just like I tell my wife, you when we're disciplining our kids. You know, when we tell them they can't do something, they're just going to go do it somewhere else. I'd rather them do it at my house, so I can see. So this the data is still leaving. You kept the risk, but you lost the controls. And so I I want to encourage leaders, you know, be careful how you phrase it. You I I certainly empathize just like you were saying, and I want to defend the leaders out there that that their first tendency, I mean, what else would you do? It's like, well, I don't know enough about this. Let's just ban it. Okay, great. But the problem is they're still going to do it. Brian Milner (14:35) Right. Lance (14:40) And you just lost control and visibility into it. It's almost like that saying, keep your enemies close. You know. Keep you know, that's the whole point of all this is embrace it. It's uncomfortable. But the more you're in it, you create, you know, the psychological safety. ⁓ you know, it's it's just that's the the soft side of leadership that I think a lot of us forget anyway. Brian Milner (15:02) Yeah, and and you're you're I mean you're bringing up Gen Z stuff and Gen X stuff and ⁓ here here's a little lesson for people who didn't live through this time, because I'm Gen X as well, ⁓ that I think is really applicable to what we're going through here, right? ⁓ back in the day, young'uns, ⁓ back in the day, there was a time when music was transitioning from being done on CDs and physical media to digital media where we could download it and use it wherever we wanted to. And during that time it was sort of the Wild West because things like Napster came out where you could download stuff and left and right. And there was all this concern about piracy and everything else. That was the bad part of it. The good part was how easy it was to get music now, right? And so there's a parallel I'm drawing here because back during that time, that was the music company's kind of approach was slam the walls down on this. It's no good. We're not going to do it this way because it's too easy to to ⁓ be abused. And what happened was it proliferated on the black market kind of areas because it was so easy to get. And I think the lesson that That I've heard a lot of people talk about from that is it it wasn't that people weren't willing to pay for it because as soon as Apple Music came out and it was 99 cents to get a song, people would just go buy the song. Because it wasn't that it was so expensive people can afford it, it was that it was so hard to get. And right. Lance (16:34) Yeah, and where does it go? You know, cloud computing back then is like, I don't want my stuff in the cloud. Now just take it take all of me as now. I just everything's in the cloud. Brian Milner (16:39) So so it finds the path Right. It finds the path of least resistance. It's like a, you know, ri water stream trying to find the river. It's gonna find the path of least resistance. And I think the same thing is happening today with AI, in that if if we try to be too restrictive of it in our companies, it doesn't mean it's not still there and happening. It's just not happening where you want it to, and it's gonna find the path of least resistance. The answer is Lance (16:49) Yeah. Yep. Brian Milner (17:07) Give it a path. Give it a way that it can actually work and and put in the controls you need, but but make it easier to do it the way that you want it to be done than the other way where people have to jump through too many hoops to do it the you know, the the the off-campus way, you know. Lance (17:28) Do you remember too? So you bring up the music stuff, which we could talk about that forever, but you remember going to concerts, no cameras, you couldn't have cameras out. Boy, that was like you would get shamed if you had your camera out. What is it now? Please post it online so everybody can see it and hashtag it. And so it just commoditized in another stream because it some concerts ⁓ allowed it. They said, Yes, please post it because that's more marketing for us. And so to me Brian Milner (17:35) Yeah. Everyone's doing it, yeah. Lance (17:55) I'm filing all that under as I try to help organizations with this as well. you know, psychological safety. Psychological safety is not ⁓ you know, because it it gets misread constantly. It's not being soft on anything. It is not comfort, it's not lowering the bar. It's a shared permission for, let's say, candor, right? Especially when the data's uncomfortable. You know what just like parenting. I'm gonna have to have these hard conversations with my kid, but they're they're Brian Milner (18:16) Yeah. Lance (18:24) success is my first primary thing. I think leaders are like that as well. We want success for the organization. ⁓ you want AI use in the open so you can hold it to a higher review standard. Like I was just telling you the last few quarters or last quarter, I've actually been talking about what are some tools we can use with teams to do that. ⁓ you can't review what you can't see. You remember that old management cliche you can't manage what you can't measure. Safety is what makes the higher standard enforceable. So I I've I try to encourage leaders. Yes, you have to worry about the compliance of finance and all that kind of stuff, but ⁓ maybe just giving them a sandbox to try it out and see how things work, you will be able to see how they're gonna use it and you can put better I don't wanna say controls around it. Standards. Let's use that word is is probably a better thing. Brian Milner (19:11) Yeah. Well and and I'll talk about this sometimes in my classes because I I I for all the talk that we we talk about in the classes about using AI and how to how it can help you do your job and everything else, I always wanna throw the caution word out there because you know, I tell everyone the last thing I wanna have happen is someone come back to me and say, Hey, I got fired. because I was using AI that I learned in your class and now I used it and I was found out I wasn't allowed to and I got fired. And there are real reasons why companies don't want this to happen. I it's not being overly cautious. you know, my best the my favorite example of this is that depending on the kind of of account you have with certain AI providers, LLMs, that they can actually reuse whatever you put into the prompt as part of training data. now, not not all accounts have that, but some Some of the accounts do. And just I mean, just imagine that you're, you know, you're on the marketing team and you have ⁓ a marketing survey that that cost your company ten thousand dollars, you know, to to put together to survey this this set of people. You've got the results of it. You want the AI to help you make sense of the results. So you give it the raw data from your marketing report. And if you're on the wrong kind of account with the wrong AI, well now that's part of the training data. So six months from now? Right. Lance (20:38) Well your competitor just just ⁓ succeeded from that. You know, that's I tell you, I tell people all the time, if you start understanding, I know LLM's, you know, large language models, I guess for is for I use that that term, ⁓ is fundamentally what these things are. And I know they've progressed significantly from when we first developed them, but if you can think of them as simply just an auto-complete tool, the world's best auto-complete tool, then when you go to talk to it. Brian Milner (20:50) Yeah. Lance (21:06) you will be a lot less dangerous to your organization or yourself because you will think about how to structure your your questions and prompting to it ⁓ to narrow down those result sets because if you just say give me an answer to this it's going to go do a probability score on every word in the internet. Do you want that or do you want to say, hey, this is the marketing results, only use this data and and yeah, don't let it be training data for other people as well. So you still have, you know, that's what leaders are worried about. I would be too, you would be too. Brian Milner (21:36) Yeah. Lance (21:36) ⁓ we're we have to it's already hard enough to be competitive in the industry than having somebody go, well here's all of our production database download, just help me find a common theme for these records. It's like holy macro, man, you just now that's you're in the general counsel's office, right? Brian Milner (21:53) Yeah. Not to mention the, you know, privacy and you know, customer data and all that kind of stuff. I mean, there's there's a million ways it could be misused and you could find yourself in trouble. I I I think the the the thing to that you were drawing on earlier is a is one of the most important points in this, and that's that you know, i it all out bands are gonna create secrecy. And that secrecy ⁓ is a source of risk. And ⁓ you know, like it's like they always say with like government things and everything that, you know, sunlight's the best disinfectant, like transparency is the best approach to this because we we ⁓ you can't manage what you can't see, as you said. And if we have all this stuff happening in secrecy behind the you know, the this shadow kind of veil. And we don't really know how our employees are using it, then you may have this leak happening where you have proprietary information that's going out that you don't want to have going out. So that's why you need to have this alternative path that you can give them to say, no, no, no, don't do it that way. Do it this way. This way you can still get the same results, but this way you're safe. And this way our company is safe. Lance (23:07) Yeah. I had a a client I was working with just I think it was about a month ago, and it spawned me creating ⁓ you know, I do a weekly email to people who want to subscribe to it, and I called it the ⁓ AI accountability charter, basically. It's like, hey, we don't we don't know how to use this stuff, so I tell you what, let's get everybody together. I mean, all you need is about a 45 minutes, not a meeting, by the way, it's a working session. I just had to do another post on that because I get tired of everybody saying Brian Milner (23:24) Yeah. Lance (23:35) Well, we're in too many meetings. We can't get real work done. You can't get real work done unless you have agreement and conveyance and collaborative, you know, the the none of us are smarter than all of us together. It's like it's not a meeting unless you've let it become a meeting. This should be a working session. And ⁓ I basically published this one page ⁓ version ⁓ where we get the team and we say, okay, number one, AI assisted work, no extra review required. This is low risk work, normal review still applies. Let's let's write all those things down and get agreements to it. And then you have the the second level of that, ⁓ which is AI assisted human review required. So a named human reviews the AI output before it w proceeds. and then we have another one called the reviewer of record, like code check-ins. You know, these are a lot of times people like just writing code and checking it in. No, no, no. You still need to put your name. on the pull request, because you're the one saying, I agree that this code was well written and it belongs in our enterprise system. You're writing something for mom and dad, then that's fine. But if you're writing stuff for enterprise stuff, I'm like, well, we need a reviewer of record. Who's the named human accountable? Not so if we get in trouble, we point fingers, we can course correct it. So what are gonna be the things that are required? ⁓ the different categories and who is the reviewer of record for those? You know, if you do a check-in, it'll be you, but Brian Milner (24:35) Right. Lance (25:04) then you have quality bar for AI assisted work. What must be true before this is done? Does that sound familiar? You know, we talk a lot about acceptance criteria and definition of done. So I was like, we need a quality bar for that stuff. And then the last one was ⁓ event tracking. So, you know, how do we log and review AI related bugs, findings, rework, start making a log of that stuff so AI can go search it again, not because it's busy work. But we can start recursively feeding our own model and learn from it. That's to me how a leader could embrace this and say, okay, look, we want to use this, but let's sit down and think of our work. You know, what requires human review, what doesn't? You know what happens? We're too busy to do that. You remember that old I remember you and I we used to teach a lot together. We had a, I believe it was you and I, ⁓ the the Lego guy that was pushing a cart on square wheels. Brian Milner (25:47) Yeah. Lance (25:57) And the guy behind them with a round would Hey, stop real quick. Let's replace these wheels. And they're like, We can't stop, we're too busy. And that's what happens is the organization we're so delivery focused, we forget that if we built a better process around this stuff, we will go faster. So my whole thing was use governance to speed up stuff. You know, it's like that sounds counterintuitive, but that's what we have to do. This is all new. It's all new. Brian Milner (26:03) Yeah. Yeah. Yeah, and and I I think I think you're right that there's sort of ⁓ there's multiple layers of this too. I think there's a there's an organizational wide sort of acceptable use policy and it's gotta be it's gotta be flexible enough that and fast enough that when new things pop up that they can be ⁓ incorporated quickly. ⁓ you know, you don't wanna have something where ⁓ hey, there's an you know, I've specified what models of LLMs you're allowed to use. And so, you know, you can use ChatGPT five point five, but ⁓ five point six comes out and it's like, well that's not in the acceptable use policy. How long will it take us to get well six months 'cause it has to go through a review. That's too late, right? I mean by then there's now five point eight. Right. Lance (27:09) You you know how it is in development. You know, you have this new tool come out, it's SQL Server 2007, then 2008, then 2014. Next thing you know, you haven't upgraded at all. And you're never going to upgrade. You got to file new because it's too much work to go backwards. I'm telling you, man, that is such a setting of point that embrace it, keep it in the open, put boundaries around it. And let's let's figure out how we're gonna use it 'cause it is not going away. In fact that those companies who say, Hey, we just don't use it at all, I bet they won't be around for s for very long. That's the tragedy. Brian Milner (27:39) Yeah. I I mean I think there's an organizational responsibility. I think there's a team responsibility. So I think your approach too of having sort of a team agreement around it to say, you know, how are we going to use it as a as a team? Because, you know, let's say our organization says, Hey, you can use AI, but n only one person on the team is you know generating code with AI and everyone else doesn't know that, well now they can feel betrayed and you don't have the safety in the team and y you want to be again transparent about this stuff. So I think you you need to have kind of an openness not only in the organization but in the team so people understand here's how we use it here. Lance (28:17) Yeah. Well, and you were talking about a you know, when you were saying, Hey, I've got some ideas for these topics, ⁓ you had the the five question framework, you know, what AI tools are we allowed to use? You know, tier by the wrist, not the tool. The wrong question is not, you know, which app. It's the right one is what is the output touching or what's the input coming from? And you know, what data must never go into a public AI tool. So, you know, OWASP, the ⁓ the open source, you know, ⁓ security people, ⁓ they have the 34% customer and 31% financial stat. People already do this every day without realizing the exposure. So they're using tools ⁓ with customer data and financial statistics in the organization that probably don't need to be out there in the world. And ⁓ you know, we have to be cognizant as an organizational leader and say these things must never go into a tool, just like you were talking about. and then how do we inspect it every time we turn around? Because if you just look at it once a quarter, ⁓ like you were kinda alluding to, you know, let's wait until the next version comes out, it's gonna be too late. You know, too late for that. Brian Milner (29:27) Yeah. No, I I think those two are really good ones. I think also, you know, trying to be clear about when ⁓ when it's important to disclose that AI is being used, ⁓ and and what what AI generated work requires human review. I think are are two important questions to add on there as well. ⁓ because ⁓ there are there are times when we we need to know this w this part of this was generated by AI. There's also times when we we need to have those clear definitions of well this type of work really must pass other human reviews. And and because there there are things that it's okay for AI to not have a human reviewer. If I wanted to do math for me, great. AI will do the math for me ⁓ and and and I don't have to like manually do it myself. I I can trust AI to do that, but I I don't want to trust it to infer things from my customer data without my oversight and my my ability to come along beside and say, Right. Lance (30:27) Yeah, don't push an email. You know, we can connect to your email and just push one. Absolutely not. Like I'm not even there at that point yet. I don't mind the grammar, you know, and and spell and things like that, but I do not want to send anything to clients, customers, or people that I'm helping. You know, I definitely want to review those. And maybe one day I trust a thing like with the math. Am I gonna really double check its calculus? Probably not. But ⁓ you know, I think that study that That two thirds of people are using it against policy is really kind of the thing that we want to highlight here. And the biggest companies, almost half, would rather you never find out. And that's not a tooling fail failure. That's you know, like we were talking about, it's a trust failure with with psychological safety. And ⁓ trust is a leadership output, unfortunately. So as a leader, that's one of the things that we're responsible for. ⁓ and you can't just say trust. You know, it it's you have to model it, you have to build things out, people have to earn it over time. ⁓ but you know, we just think if if AI is already part of your work, put it in a working agreement, a definition of done, you know, not a 40 page PDF that nobody's gonna open. One page, you know, start out with that and say, Did that help? You know, I'm I'm I'm all for that kind of structure. People like, governance and oversight. Well Brian Milner (31:36) Yeah. Lance (31:48) Use that to speed up your ability to to play with tools that are dangerous but helpful. And that's the only way that I think that we're gonna be able to trust the tools is have people use it responsibly, but demonstrate that and then call out the risk before they start using it. Brian Milner (32:04) Yeah, like a lot of things in life, the you're not stuck with two choices, one being completely everything's open and people can just have the wild west kind of approach to it, and the other, everything's locked down. There is an in between and that's where we're at is kind of finding those boundaries and setting them safely for organizations and then having them be adaptable, flexible enough that when we discover new things, we can change them quickly so that we don't lose our competitive advantage with you know, our customers and our our competition. Yeah. Lance (32:38) ourselves like you know we're we're competing for work out there ourselves and i i sit there and think back to you know the work of it is software which is intangible ⁓ hard things to compare to but i'm not a handyman you know the other day i was out here trying to do some things with ⁓ wood and cutting of baseboards and we have a table saw and i kind of thought of it that way i'm like ⁓ i've not used this table saw before and it could literally cut my whole hand off in in one second without me even knowing it so guess what i did I opened the manual and tried to figure out what all the different ⁓ logos or not logos but icons mean. I watched a YouTube video and I practiced with the little few things without the thing being on to make sure I knew how to wait. Well, that's easy to do with tangible stuff, right? It's hard to do with software, but I still approached it carefully. It's not that I avoided it. I'm like them it took me more time probably to do the first cut than it would have taken 20 manually. But it sped me up later, you know. So it's like that. The ROI of putting it into it at the beginning, it feels slower. It is not a faster way, just like we talk about with Agile. It's not a faster way to work, but it focuses on getting better. And when you get better over time, the power of getting effective and efficient is the competitive advantage. And you start removing waste and highlighting waste and have enough processes and signals that are built in that you can see the waste, you know. But if we just ignore and put our head in the sand. it's all out there. You just you know, it's just it's easier for you 'cause you're not acknowledging it, but your company's gonna die. Brian Milner (34:10) Yeah. Well this has been great, Lance. It's an important conversation and I'm I'm glad you were here to to talk about it with us. ⁓ I'm sure we're gonna have Lance back many times on the show 'cause Lance has a lot to share with us, but I just wanna thank you for ⁓ coming on and and you know engaging in the the topic with us. Lance (34:27) Hey, I always love experimenting, even if I'm wrong, you know, but I I really enjoy ⁓ learning and adapting and being malleable and love anybody to share anything that they can. So let's do this, you know, let let's let's help get this all better together, you know, collective intelligence. Thank you for having me, Brian. Brian Milner (34:44) Big thanks to Lance Daisy for coming on. his company is Big Agile. And if you watch the video you can see why we call him that, Big Agile. Lance has been a friend for a long time and ⁓ has a lot of good insight in this area. I know we're all kind of trying to figure out how does this play out and what does this mean. for our teams and this is just one of those other areas that we're struggling with figuring out where the boundaries are is is how do we handle the safety of this. So ⁓ curious what you guys think about that. If you have any comments, if you have anything you wanna share about this, or if you have some suggestions for future topics or guests that maybe you want me to have on. I want this to be the podcast you want to hear on these topics. So send me an email. You can send it directly to me. It's just Brian at agilityevolved.com. That's my own personal email address. Brian with an I at agilityevolved.com. And that'll come to me. And I I'd love to hear some suggestions from you or feedback, anything that you want to tell me to help us make this show better. The other thing I'd Ask from you is if you can, ⁓ would love for you to tell a friend about the show if you like it. ⁓ we're trying to kind of grow a new audience here. So if this is topics that are interesting to you and you know others that might be interested in it, then please ⁓ we we'd really appreciate the the favor of just recommending us to someone that you know. ⁓ also like and subscribe to us on your podcasting platform of choice. ⁓ especially as a new podcast, it's really useful. ⁓ we we need people to find us, to listen to us, and you know, I'm not I don't have a huge advertising budget or anything. ⁓ it's just me. So I ⁓ I'm not ⁓ I'm not out there trying to you know pay for an audience. I I ⁓ depend on you guys. so if you like and subscribe, that will help us to get more people into the show. And that's about all I got for you for this week. thanks for joining us on another episode of People over prompts podcast.