Innovating Leadership:
Co-Creating Our Future

Resources:

Neil’s book, Own the AI Revolution, is available on Amazon at https://amzn.to/4kx4af3, and on Audible at https://amzn.to/4lgS0rK.

Learn more about Neil on his website at https://www.neilsahota.com/. He shares his latest observations on AI in the workplace on his Substack channel at neilsahota@substack.com.

Our host Maureen Metcalf posts a newsletter every week on LinkedIn. You can subscribe here.

The book Maureen and Neil referenced which they wrote together is Innovative Leadership & Followership in the Age of AI. You’ll find details about it at https://bit.ly/LeaderInAI, or check out the Kindle version at https://amzn.to/44buVz8. The audiobook version is now available at  https://amzn.to/4dTCleZ.

Neil’s other binge-worthy episodes about AI are:

– The 10 Leadership Skills You Need to Safely Adopt AI

– Unleashing the Power of Human-AI Collaboration

– How to Own the AI Revolution

Guest(s):

Guest(s) Bio:

Neil Sahota (萨冠军) is an IBM Master Inventor, United Nations (UN) Artificial Intelligence (AI) Advisor, author of the best-seller Own the AI Revolution and sought-after speaker. With 20+ years of business experience, he works to inspire clients and business partners to foster innovation and develop next generation products/solutions powered by AI.

Transcript
(auto-generated)

squadcaster-04bj_1_05-28-2025_121713: [00:00:00] This is Innovating Leadership Co-Creating Our Future. I’m your host, Maureen Metcalf, the founder and CEO of the Innovative Leadership Institute, where we help leaders be future ready. Helping us on that mission to today is returning guest Neil Sahota. Neil is an IBM Master Inventor, United Nations AI advisor.

Author of the book Own the AI Revolution and professor at uc Irvine. This multi-part series will ta take you through the five reasons organizations fail at AI and recommendations to ensure you are not one of those failures. So Neil, welcome.

neil-sahota_1_05-28-2025_091728: Hey, Maureen. Great to be back.

squadcaster-04bj_1_05-28-2025_121713: It is great to have you back. So let’s jump in. This idea of reasons [00:01:00] for failure is kind of intriguing. Um, you’ve identified five Fatal Mistakes Enterprises make when implementing ai. Can you briefly outline those mistakes and explain why four of them are people-centric?

neil-sahota_1_05-28-2025_091728: Yeah, sure. It’s a, it is a interesting thing in. Uh, I see a lot of companies today that they’re like, Hey, we try to do something with ai. We’re doing the common use cases in our industry, we failed spectacularly. Yet our competitors who are not as smart as us it done.

squadcaster-04bj_1_05-28-2025_121713: Hmm.

neil-sahota_1_05-28-2025_091728: so they’re really wondering why, why is that happening? And. It, you alluded to, it’s more a, a people challenge and requires more leadership. And so let me share the five and as the listeners are tuning in and see which one you can guess is not a people one, but the, the number one, number [00:02:00] one most common failure point is they treat AI as software and it’s not really software.

It’s a high energy intern. So number two. They’re just looking at so every day there’s about 12 new AI tools they get released. Some are hype, some are legit, they’re trying to figure out what are the right tool or a couple of tools to try and use. Again, it doesn’t quite work that way. Number three is more around actually expectations, what people believe AI can or cannot do. Number four. Relying on the smart technologist to tell you the pain points of the business and tell you where to use ai. And number five, actually, and this is the one I think a lot of people will be surprised by, is only looking at automation rather than also thinking about innovation. rather than [00:03:00] just what you already have and automating everything, including the flaws, for a better way of doing the work.

squadcaster-04bj_1_05-28-2025_121713: And the last one has been an issue for decades. The, there was an article when I first started working about don’t pave the cow paths.

neil-sahota_1_05-28-2025_091728: for, for for sure, Maureen, right? We’re better or worse, we tend to wire organizations to look for continuous improvement. And I’m not saying there’s anything wrong with that, you cannot continually improve. And somehow achieve the light bulb. So if we

squadcaster-04bj_1_05-28-2025_121713: Okay.

neil-sahota_1_05-28-2025_091728: new capabilities, all these things, we should be looking for better ways of just doing our work.

squadcaster-04bj_1_05-28-2025_121713: I love that and I do see how that’s a people issue.

Getting humans to reconceptualize what we do and how [00:04:00] we do it is can be tough.

neil-sahota_1_05-28-2025_091728: Well, we’re, we’re basically taking the approach to, we’re trying to fix something that’s not necessarily broken. Right. And for a lot of people it’s like, why would you waste resources and time trying to do that? And it’s like you could actually maybe create a lot more value that way. I mean, we don’t use leeches to treat ourselves for health issues anymore.

Right.

squadcaster-04bj_1_05-28-2025_121713: Uh, my doctor does not, I’m not sure about others.

So how are those people problems? Because they did sound, some of them sounded technology centric.

neil-sahota_1_05-28-2025_091728: Uh, it does, but let’s, let’s start with the most common one first, then let’s talk about that. You know, this perception that AI is software and, and it’s really not. Right? It’s a high energy intern. So, and I, and I get, we’re used to computers like running a program. So a program is just a set of [00:05:00] instructions.

So you have a hard coded path, exception paths, all these things. You don’t really program ai. You teach it, right? You give it something we call ground truth rules on how to make decisions. give it data, you give it to human teachers. So that’s how it learns and becomes competent. But the problem is everyone wants to buy a box, right? I need help my taxes, I’m gonna buy a software package that’s good at tax preparation, right? If I need something that’s gonna help me manage, you know, paid time off and associated employee benefits, there’s software packages to do that. everyone’s looking like, where’s the AI software package that’s gonna do X for me? And it’s like, that’s actually not how AI works. It’s not software. You don’t program it again like you teach it. So they wanna get the most out of ai. You have to think of it as a [00:06:00] high energy intern, that AI is ready to do whatever tasks you give it, no matter how boring, tedious, administrative, grunt work and do without complaint and do 24 7. But like any human intern, you have to teach it how to do it first. you gotta teach it business on your data, your processes, all these things because we all operate a little differently. So you’re not really tell running a program for the ai, you’re telling it how to do something for your company.

squadcaster-04bj_1_05-28-2025_121713: So I am curious, I did a presentation last night on AI and leadership, so the, the book we wrote and one of the questions was. In turn means they don’t yet know everything. They still need to learn. How do you make sure, and this was specifically on the [00:07:00] topic of hallucination, how do you make sure they get it right?

And I’m curious to see how you answer, and then I’ll say what I said, ed.

neil-sahota_1_05-28-2025_091728: Yeah, sure. I mean, it, it’s, it’s a great thing and that’s why I use the word intern. It’s not gonna be perfect.

squadcaster-04bj_1_05-28-2025_121713: Mm-hmm.

neil-sahota_1_05-28-2025_091728: But, excuse me, however, but for the AI and these hallucinations we have to understand is that it doesn’t know what it doesn’t know, and we create something. It doesn’t actually understand the, the content and the context of it. so there are things that we actually take for granted. we’ve, we’ve probably all heard the story and this, it still happens today, surprisingly. get, uh, a lawyer. Goes to like chat GPT and says, gimme three cases that support my case strategy that I can use in court. Right. And you get these things and they don’t check them. Right? Well, the thing with AI is that it doesn’t understand [00:08:00] intuitively that when you ask for, a supporting case as a precedent, it has to be true. Right. It doesn’t actually know that. So when you ask it with a prompt saying, give me three cases that support my case strategy, it’s gonna create three perfect cases as precedent for you. Right? Anything AI does should be considered a draft, right? Much. We review an interns work. It’s the same thing we should be doing with ai.

squadcaster-04bj_1_05-28-2025_121713: You know, as you say that, one of the things that strikes me is we think we’re training it, it is also training us to be more precise in our inquiry. So give me three accurate. Well, well tested cases, and then it’s a management issue. It’s a leadership issue. I wouldn’t have my junior associate or my intern go prepare my [00:09:00] legal case and not review it and stand up in court that that just seems like bad, bad practice.

neil-sahota_1_05-28-2025_091728: Uh, and I think that’s part of the challenge that we still, we’re still think of AI as a computer, a software. So it’s like if it gives me an output, it must be accurate, right? It must be true. ’cause it’s following some path. It’s doing a calculation and it’s like, again, it’s not software, it’s an intern in terms are gonna make mistakes, right?

AI is not perfect. Our data is flawed and it has human teachers and we’re not perfect. Right? The goal is that the person using the machine more accurate than just the person or just the machine. I.

squadcaster-04bj_1_05-28-2025_121713: So it is the human machine hybrid.

neil-sahota_1_05-28-2025_091728: Yeah, we call that hybrid intelligence. We talk about what’s the future of work, where are we [00:10:00] going in terms of like leadership and management. It’s towards hybrid intelligence. we know that AI is gonna be better at us than at millions of variables and scenarios and math and all these things, but there’s a lot of things that people are much better than ai, right?

First of a kind situations, creative thinking. We’ve seen this, it’s the person using AI is far more effective. That’s why when people ask me, is AI gonna take my job? I’m like, no. But the person using AI will take your job, right? I, I, I like it to this, and I, I find it, unfortunately, it’s becoming a dying example, but chess, if you’re a human being, you play against an ai, player, your odds of winning are one out of a million. A human teamed up with an ai, chess player just playing against an AI player wins nine outta 10.

squadcaster-04bj_1_05-28-2025_121713: Interesting. A, and we will see the AI [00:11:00] continue to learn. So at some point they may be smarter than us, either at specific tasks or agents at bigger tasks, I assume.

neil-sahota_1_05-28-2025_091728: Well, there’s some things AI is, is better than us that we, we find surprising. I mean, that’s failure point number three. There. We, we’ll, we’ll get to that. But the, the honest truth is we, uh, AI won’t be able to do things unless we were able to teach it. And like one of the things we have not been able to figure out is how to teach creative thinking, right?

The AI is never gonna be good creative thinking ’cause we don’t know how to actually teach that. Our, our ourselves. Sometimes we’re not really the best creative thinkers either. So if we can’t teach ourselves how are we gonna teach the machine?

squadcaster-04bj_1_05-28-2025_121713: I wonder if at some point the machine figures out how to teach the machine, but that’s a maybe different topic.

neil-sahota_1_05-28-2025_091728: But you get, you’re getting the whole thing about now AI interacting with AI in the near future. I mean, you’re gonna have an AI assistant and some [00:12:00]AI sales bot’s gonna try and convince your AI assistant to buy something.

squadcaster-04bj_1_05-28-2025_121713: So what you’re telling me is my password vault should be my head, not, not some separate AI system.

neil-sahota_1_05-28-2025_091728: Might be a good idea.

squadcaster-04bj_1_05-28-2025_121713: At least my bank. Okay. So what was number two?

neil-sahota_1_05-28-2025_091728: So number two is this whole idea of like tools, right? We hear about all these companies, all these, they’re all doing ai. Some of it, like I said, is hype. Some of it is not really ai. Someone that doesn’t really quite work the way it, it’s promised and someone is legitimate, right? And on average, there’s about 12 new tools that come out every day. So I can’t even through this. The thing though is again, it’s, you can, I’m a big believer of not reinventing the wheel, [00:13:00] right. But, uh, it’s not a question of just mix and matching tools together to get you what you need, right? Again, because you have to teach the ai, there’s some level of customization. So, you know, back, back in the day, you know, like we’re doing some work with HR Block, they were looking at things like, could AI help us, you know, with tax preparation. Kinda looking at the data, about 70% of tax returns are fairly the same. So it’s use case for ai. You can teach it about, uh, the tax code, how to read all these things, right? But it’s not like, okay, I just pluck this tool over here and I pluck this tool over here and I have what I need. Right? We use several different kind of APIs and tools, but you still have to spend time teaching it. What are some of these concepts mean? Give it the various tax codes. Doing that instruction, doing teaching it the way h and r blocked prepares taxes and how they communicate with the clients. So all those things [00:14:00] had to be factored in and I mean, they, they built it and you know, it was wor, it’s works for the most part, pretty well.

When the AI prepares the tax return, I mean, we will ask for stuff that might be missing, we’ll prepare it. But then a human accountant actually reviews the work. Make everything looks okay and because that human review, you know, we actually noticed something early on that kind of made our head spin a bit and that AI tax returns on average were having, the tax refund was about 203, 203 $6 higher. And it was like, why, why is that happening? Is it making a mistake somewhere? So we’re looking through it and we noticed it’s going like, person qualifies for this $25 tax credit over here because, you know, the municipality has this, $50 tax rebate. They’re all legitimate deductions, right?

It’s just we, kind of minute and stuff. I. You know, the typical human account didn’t realize that, you [00:15:00] know, have you ever tried to read the tax code? You know, you got federal, state, municipal, you know, if you have insomnia, I recommend it. But because there members every land, that tax code, every time it changes, it knows what people are actually eligible for. So we can actually catch and aggregate a lot of these small things, which is why those people were getting a slightly higher tax refund.

squadcaster-04bj_1_05-28-2025_121713: So to your point, there are things that already does better. I.

neil-sahota_1_05-28-2025_091728: I mean, it’s catching that level of details. Iic memory, it’s gonna remember everything in the, the tax code, the tax laws. There’s human that can do.

squadcaster-04bj_1_05-28-2025_121713: So one of the things I take away from this then is 12 new programs a day or a minute, um, seems like a minute. They’re always coming out and so ha having someone who is an expert in AI and probably the industry and what’s coming out related to my [00:16:00] industry would be very helpful rather than me trying to keep up on it.

neil-sahota_1_05-28-2025_091728: A hundred, a hundred percent. But I get asked the questions a lot like, look, we’re vetting the vendors out. What are some things we should be looking for? I always, I always say, there’s two things you can do, especially as a non-technical person if you’re looking at a vendor or a tool. Number one, is there anybody on their team that has domain experience?

squadcaster-04bj_1_05-28-2025_121713: Mm.

neil-sahota_1_05-28-2025_091728: If someone’s trying to sell you a AI marketing tool, is anybody on on their core team there actually do marketing, Is that actually their background? ’cause if it’s not, how, how do they know? Right? If, give you an example of this. meta has been burned by this twice. First with Instagram, then with Facebook. But you know, they, they rolled out their new AI kind of ad tools. There were, you know, we had some real estate agents trying to market properties and the AI intentionally excluded certain groups, right? like, well, either they couldn’t afford [00:17:00] it or whatever. Some something in there, you know, triggered AI to do that, and so they got busted for redlining.

squadcaster-04bj_1_05-28-2025_121713: Hmm.

neil-sahota_1_05-28-2025_091728: There’s Fair Housing Act. You cannot exclude people, any group of people from residential housing advertising. And so second time they get busted. They’re like, oh my God, they’re huge fines. And they go, their engineers. I’m like, how come you guys didn’t account for redlining? And the engineers are like, what’s red lighting? know, smart technologists, but they don’t understand the regulatory environment of the housing. So it never entered into their mind to incorporate that as part of the teaching?

squadcaster-04bj_1_05-28-2025_121713: So they, they incorporated things. They understood like the software would weed out bullies, but not.

neil-sahota_1_05-28-2025_091728: Yeah, that’s, that’s the problem. So the number one thing I tell people is check, they actually have the right domain expertise on the team.

squadcaster-04bj_1_05-28-2025_121713: Yeah.

neil-sahota_1_05-28-2025_091728: there’s no way they could have built an effective tool. Second thing I always say is, [00:18:00] and look, the, they’re not gonna reveal everything, but they should be able to tell a few things, is what was their training strategy. They should be able to share a few things about how they trained the AI and some general, I get it. Don’t reveal trade secrets, but general information about the training sets for the AI system because you teach it with biased data. Well, flawed ground truth. AI is gonna make mistakes, right.

squadcaster-04bj_1_05-28-2025_121713: Well, and to your point, it may take a while for those mistakes to surface, and at that point you’ve done harm.

neil-sahota_1_05-28-2025_091728: Yeah, and a, and a great example of this go back in time a bit, but Amazon, you know, most people probably have heard the story, but Amazon built an AI recruiting tool. I think they launched it in 20 17, 20 18, and it turned out that it was sexist, right? They took 10 years of. Hiring history and resumes.

People they hired didn’t hire success. [00:19:00] Right. But the thing they didn’t really account for, and you know, and again, it’s maybe a little unconscious thing, but this is again, where leadership becomes very important, is if you look at Amazon’s hiring history, they predominantly hired men a variety of reasons.

We don’t want you to get into here. Right. But the AI basically learned that, you know, the difference between some of these candidates? And it’s not that, you know, there’s the genders on those resumes, but what it learned was that when men speak and write, they use certain types of words and women use different kinds of words. And so there’s a difference in the language that, you know, I don’t think people kind of picked up on least consciously, but the AI did. That’s how I was able to filter their resumes. As a result, if you’re using words that women tended to use on resumes, it would deprioritize.

squadcaster-04bj_1_05-28-2025_121713: Interesting,

neil-sahota_1_05-28-2025_091728: creating a sexist bias.

squadcaster-04bj_1_05-28-2025_121713: interesting. So I [00:20:00] should use an AI to edit all of my writing.

neil-sahota_1_05-28-2025_091728: But we’re getting to the point now where I think everyone’s generating their stuff with ai. So it’s kind of hard, hard to tell now, right?

squadcaster-04bj_1_05-28-2025_121713: So it was only a short term bias.

neil-sahota_1_05-28-2025_091728: I, I, I don’t, I don’t know about that. Right? Because I, you’re seeing more and more tools out there trying to screen out generated its,

squadcaster-04bj_1_05-28-2025_121713: Mm-hmm.

neil-sahota_1_05-28-2025_091728: Resumes and things that maybe there’s a different kind of bias going on now.

squadcaster-04bj_1_05-28-2025_121713: Interesting. So what’s number four?

neil-sahota_1_05-28-2025_091728: Uh, well the tools is number two.

squadcaster-04bj_1_05-28-2025_121713: Okay.

neil-sahota_1_05-28-2025_091728: So number three is actually around what we think AI can or cannot do.

squadcaster-04bj_1_05-28-2025_121713: Okay.

neil-sahota_1_05-28-2025_091728: So, you know, we’ve, we’ve come in with this attitude that. You know, we’re used to getting to the second generation of computers, do something faster, cheaper, less error software programs. There’s, there’s some things computers can’t do.

Like, uh, AI cannot understand emotions. Right. And we’ve actually [00:21:00] shown and benchmark that AI actually can understand emotions. It doesn’t feel them, but we’ve been able to teach them what these emotions mean for people and the science. what we’ve actually found is that AI is far better. Detecting the emotional state of a person that another human being is right.

The AI is a laser-like focus on you. It’s not thinking about what to say back. It’s not wondering what the kids are up to or what was that sound over there. Right. And while the best human can look at maybe five to seven points

squadcaster-04bj_1_05-28-2025_121713: Mm-hmm.

neil-sahota_1_05-28-2025_091728: real time AI is looking at hundreds of thousands of points.

It’s looking at your word choice, the inflection of your voice. It’s processing a lot more data to figure out. Emotional state. As you start switching, maybe you start becoming angry. It’s already picked up on that and changing how it interacts with you.

squadcaster-04bj_1_05-28-2025_121713: So what. I, I attended the International Leadership Association Conference last year, and I walked in with the bias [00:22:00] that computers can’t manage humans. And what it said is for basic interactions I. Completely supporting what you just said, that it is actually better at interacting with humans. And it doesn’t forget if you tell it to go say hello in the morning, it says hello to everybody.

It doesn’t, it’s not tired, it hasn’t missed its first cup of coffee. It, it can read you consistently and shows up.

neil-sahota_1_05-28-2025_091728: Yeah. And that, that, I mean, that’s the beauty of ai, right? It, it’s not gonna get tired. But again, you have to teach to do these things. The funny thing though is that if you’ve taught at some of these things we call like psychographics and your own linguistics, learn the specific patterns of people.

So learn like this. When I greet this person in the morning, a high energetic high, this person likes, you know, when you crack a joke as part of the good [00:23:00] morning, and so it’ll actually custom to create a deeper level of engagement with each of those people, right?

squadcaster-04bj_1_05-28-2025_121713: Which, depending on the sensitivity of the human manager, that isn’t necessarily a given.

neil-sahota_1_05-28-2025_091728: Yeah. Right. I mean, we all have tough days or we’re in a hurry. Get that right. The, the flip side here is also more, more fascinating in that people I always say like, well, aren’t people kind of a little coy or they’re a little skeptical, or, you know, maybe they’re uncomfortable. Talking with an ai, right. AI manager, what we’ve actually found is no, that people are actually far more comfortable talking to the AI than their human being. They’ll actually tell the AI things. They would not tell like their boss, their doctor, their lawyer. feel like they’re not being judged.

squadcaster-04bj_1_05-28-2025_121713: Uh, there’s an HBR article how people are re really using Gen [00:24:00] AI in 2025, and one of the top uses is therapy and companionship. So I assume therapy is the, I’m not being judged. I, I’m not sure I understand what companionship means in reference in this article, but certainly. Like I would with a therapist and or a boss, I can share things like I’m not feeling well today, and my human boss might tell me to stop drinking so I’m not hungover in the morning where my AI boss might have a little kinder response.

neil-sahota_1_05-28-2025_091728: Ironically, probably more empathetic from the AI boss.

squadcaster-04bj_1_05-28-2025_121713: Yeah.

neil-sahota_1_05-28-2025_091728: to say it that way. Well, look, we, uh, you know, the Los Angeles PD is using, uh, starting to use ai. to do [00:25:00] minor police reports like your bicycle stolen. Right? And they’re, they’re doing this because it frees up officer time for more, more serious offenses.

But they’ve also found is that when people are filing a, a police report, they’re a little more honest with the information that they’re sharing, Like, I think there was a case where guy’s car was stolen out of their driveway or something like that. They’re saying like this, and the AI’s just like, did you, were those car doors locked?

And the guy’s like, uh, you know, I, I think I forgot to lock the car doors. Whereas if it was a human officer, they would probably never attest. Oh yeah, of course. I, my car never do such a stupid thing, you know?

squadcaster-04bj_1_05-28-2025_121713: So really interesting use cases then.

neil-sahota_1_05-28-2025_091728: Yeah, and you know, it, the companionship is an interesting one. ’cause you know, there, there’s a lot of work that’s been done that space. I mean, before COVID loneliness [00:26:00] was the biggest illness in the world, I think about 40% of people suffered from moderate to severe loneliness. The question became is, you know, someone that is that lonely, do you create a safe space for them? They can at least feel connected. So not substitute out human relationships, but an environment where they at least feel connected, that they feel like they can take some of those next steps address the loneliness and maybe build those human relationships. And found that AI was a good fit, right? These empathetic AI bots are now robots that are out there. They, it was a lot easier for these people to establish that kind of communication, feel some level of connection to something, because again, they didn’t feel like the AI is judging them taking pity on them or any of those things. It’s just like, it’s, it’s literally just listening to me, you It’s acknowledging the things that I say.[00:27:00]

squadcaster-04bj_1_05-28-2025_121713: Which is something that most of us want, whether or not we’re lonely. We appreciate the, the empathy and the understanding.

neil-sahota_1_05-28-2025_091728: Yeah, I mean it’s, it’s interesting thing you think it, you know, in the workplace, sometimes people just want to be heard, right? It’s not necessarily that they wanna push their own idea or some of these things, but they wanna feel like that I am, I’m being acknowledged and I feel like I’m actually contributing in some capacity.

Right? And you know, was it brainstorming? It’s 90% of the time, it’s either the first or second idea that gets picked because people anchor to that. you think about that we don’t really create that level of environment where people really feel like they’re probably being heard.

squadcaster-04bj_1_05-28-2025_121713: Yeah, I’m just, as you’re talking, I’m thinking often we would like that at home too. Come home and have someone. Ask how your day was and sit and listen intently rather than[00:28:00]

thinking about making dinner, running out the door to get the kids. All the stuff that we do as we’re saying, hi honey, welcome home. Let’s go. I,

neil-sahota_1_05-28-2025_091728: Yep. Right. That, that’s why we talk about AI companion. It’s, always there, right? If you

squadcaster-04bj_1_05-28-2025_121713: mm-hmm.

neil-sahota_1_05-28-2025_091728: show, uh, is it, uh, black Mirror? I think that episode, white Christmas with John Ham, they literally take your brain and grams and imprint that to create your AI assistance. So the AI knows you as well as you know yourself. So

squadcaster-04bj_1_05-28-2025_121713: Interesting.

neil-sahota_1_05-28-2025_091728: your needs and all these things. And so that’s why it became the ultimate companion for you. what you probably want for breakfast, whether you had a bad, you had a bad day at work and you know, play your favorite song. You know, that’s kind of the holy grail for a lot of people. means then you talk about empathetic ai, you know, uh, you go. That’s what we’re a lot of people really want.

squadcaster-04bj_1_05-28-2025_121713: So [00:29:00] what’s the next of the five failures?

neil-sahota_1_05-28-2025_091728: So number four is try to rely on the Spark technologist to tell us what to do with ai. So probably gonna freak some people out with this for a variety of reasons, but I’ll ask you the question, Maureen. you let Elon Musk perform open heart surgery on you?

squadcaster-04bj_1_05-28-2025_121713: Not with his chainsaw.

neil-sahota_1_05-28-2025_091728: Uh, how about Mark? How about Mark Zuckerberg? Would you let him manage your retirement portfolio?

squadcaster-04bj_1_05-28-2025_121713: I, I am sure Mark is brilliant at a lot of things, but not everything.

neil-sahota_1_05-28-2025_091728: Yeah, and that’s, that’s, that’s what I typically hear from people is like, no, and why not is like they don’t know the, they don’t know. doesn’t know how to do open heart surgery. Zuckerberg doesn’t know how to manage retirement portfolio. That’s not their strengths, right? They haven’t been trained in those things, and so it’s [00:30:00] like, then why are you relying on your smart technologists to solve the problems of your doctors, your lawyers, your marketers, your financial analysts, you know, customer service reps, they, they don’t know that space, right?

The technologists. Are good at creating the tools, right? They’re gonna keep building better hammers and saws for you. the use, right? How you actually use these tools comes from domain experts.

squadcaster-04bj_1_05-28-2025_121713: And so this ties back to your earlier comment about redlining and Amazon and or Amazon.

neil-sahota_1_05-28-2025_091728: yeah. Amazon with a AI recruiter. Yeah, that’s. That’s the key thing that we tend to forget. We think it’s a it thing, so technologists will be able to tell us what to do, but it’s like we’re, we’re, we’re hiring an intern, we’re hiring high energy intern. We’re not buying a software package

squadcaster-04bj_1_05-28-2025_121713: And, and this ties to our joint book, [00:31:00] talking about the 10 things leaders need to do. One is have domain expertise.

neil-sahota_1_05-28-2025_091728: a hundred percent. You can’t, you can’t lead something you don’t really fully understand.

squadcaster-04bj_1_05-28-2025_121713: Uh, yeah. The lawyer who asked the AI to go come up with three. Precedent winning cases needed to have legal experience. You wouldn’t want me doing that.

neil-sahota_1_05-28-2025_091728: No, no. If offense’s to you, Maureen, but yeah. You,

squadcaster-04bj_1_05-28-2025_121713: You don’t.

neil-sahota_1_05-28-2025_091728: in play here, right? Because you, you’re teaching the AI. That’s what we tend to forget. A lot of people treat AI as like an IT project, right? The technologists just don’t know the on the ground problems of a lawyer or a marketer or these folks, or the regulatory environment or the challenges working with users or patients or clients. That’s why you look at like legal tech. There was a ton of money [00:32:00] pumped into AI and legal. And 90% of it got wiped out because they were investing in technologists, not lawyers or paralegals. You look at the, the ones that were successful, like you know, a company called Legal Nation. It was started by three lawyers, no technical knowledge, right?

But they understood the space, right? They’re like, you know, one of the tedious things to do read a complaint and generate the corresponding court documents, right? It takes a few hours of an associate lawyer’s time. Could AI do something like that, right? And so they looked the problem, they learned a little bit about how to use the tools.

They said, this looks like it’s feasible. Then partner up with smart technologist people to actually help build and train the AI system very successful with it today. Right? And know a lot of people ask, what about those associate lawyers that used to do that work? They didn’t get fired. They’re actually doing more important work now.

Now they can actually do things like jury selection. They can do rainmaking like [00:33:00] business development. You know, they’re so, they’re doing things that add more value to the firm, but they only got to this point because, we as lawyers understand the challenges here and the, the monotony of this type of work.

squadcaster-04bj_1_05-28-2025_121713: So I wonder having been involved in ERP implementation, so enterprise software, typically enterprise wide program. So we have the technologists, but we also had. Subject matter experts in every area. I need to understand order to cash, which means I need financial people engaged so that the machine diverts the money to the right place.

Very simplistic, but so it sounds like the same principles or similar to implementing effective solutions, not software. Would come into play. This is a business solution. I [00:34:00] need to map my business processes, understand how the processes and the people touch the the system, and build a solution that integrates people, technology, and frankly culture into the integration.

neil-sahota_1_05-28-2025_091728: Percent. Hundred percent Maureen. That’s the thing. Look, there’s a lot of smart technologists out there and some are probably in, you know, each of our organizations, but that doesn’t mean they have the right answers for us. One of the reasons wanted to start the whole AI forget initiative in the United Nations was, if you just leave it up to technologists and you only put a bunch of smart technologists in the room, you’re gonna think of things like a flying car. Right. not gonna think about like, well, could we leverage some of these things to build solutions to help fight climate change or advance medicine, right? That’s why we, we really need a community. That’s why the best AI [00:35:00] solutions come from domain experts, not technologists, right? Let the technologist build the better hammer for you, but you as a domain expert, a subject matter expert, have to learn to use the hammer, right? technologists don’t think of those things, right? All they

squadcaster-04bj_1_05-28-2025_121713: Brilliant.

neil-sahota_1_05-28-2025_091728: to an outcome. I need a better hammer. Right? It has to be lighter, right? Whatever that they’re gonna do that they’re not thinking about, like how people will use the hammer, which is a whole different set of issues, right?

Because they’re not thinking about other uses misuses. But again, that’s on us as domain experts to also anticipate how these tools might be used.

squadcaster-04bj_1_05-28-2025_121713: And the risks associated with each use or misuse.

neil-sahota_1_05-28-2025_091728: A hundred percent.

squadcaster-04bj_1_05-28-2025_121713: If Elon Musk uses that hammer to crack my head open for brain surgery.

neil-sahota_1_05-28-2025_091728: I, I always say, look, I could build a house with a hammer or I could tear one down. Right?[00:36:00]

squadcaster-04bj_1_05-28-2025_121713: So then what’s the last one I,

neil-sahota_1_05-28-2025_091728: So the last one is that we’re stuck on automation over innovation.

squadcaster-04bj_1_05-28-2025_121713: okay.

neil-sahota_1_05-28-2025_091728: Right. So when people or organizations think about where to use ai, all they’re thinking about is doing something faster, cheaper, or less errors. And I’m not saying there’s anything wrong with that. are only tapping into a handful of capabilities. So you’re only unlocking 20, 30% of the value. And because we’re so focused on continuous improvement, it’s a leadership challenge. So when we automate something, we’re also automating all the drawbacks, all the flaws, all these things that exist with our current of doing the work. So we’re not taking a look at the opportunity like, well, I got some, I got some new tools.

I still have my hammer in saw, but I got this widget thing over here. What could that do for me? Right. We don’t look at is there a better way of actually doing the work [00:37:00] now? Innovation really comes in. What’s that better way of changing the process or the system to unlock more value? And for better or worse, as many as many companies say that we want entrepreneurial thinking, they often either seem risk averse, wanna make the investment, or the most common thing I hear is like you’re trying to fix something that’s not really broken.

squadcaster-04bj_1_05-28-2025_121713: So this is, the examples of this would then be Uber, Airbnb, Netflix, external companies who’ve fixed the not broken thing until there’s a fix that renders the other solutions broken.

neil-sahota_1_05-28-2025_091728: Well, we, we think about disruption, right? We throw that word around a lot. You know, people tend to forget ’cause it’s probably normalized now that Walmart disrupted healthcare. Right. It’s how does a big box retailer do that? Right. And they, [00:38:00] they just learn that, look, we, we have the physical locations, you know, where we tend to build in low density areas, more rural areas, you know, we can bring some of these tools, AI or not, you know, we can provide services like, you know, eye exams. You know, basic physical, some of these other things, and, you know, why do they do that? It’s like, well, while you’re here waiting for your glasses or something, you can buy other stuff. Right. That’s the motivation. But some of these things, you know, couldn’t be replicated. I mean, Amazon, I think had the idea first, but you can’t really take an eye exam using telehealth. Right. And so that, that’s, that’s the thing. I mean, I remember. Back in the day, Airbnb, you know, their original model was you get a beanbag and your host makes you breakfast in the morning. Right.

squadcaster-04bj_1_05-28-2025_121713: Mm-hmm.

neil-sahota_1_05-28-2025_091728: And they just wanted, wanted to build it up enough to show that enough people would use it to sell it to Marriott. Right. And they actually got a [00:39:00] chance to pitch Marriott after they had some traction. And Marriott said, can’t be a hotel chain without owning real estate. Right. It. Well guess who’s the biggest hotel chain in the world today? Right? Airbnb. And they don’t own any real estate. different way of doing the work.

squadcaster-04bj_1_05-28-2025_121713: Uh, and so that illustrates then beautifully that for those of us who are not rethinking the work, uh, and disrupting, we are likely to be disrupted at some point.

neil-sahota_1_05-28-2025_091728: Yeah, and you might get disrupted by either one of those, uh, startup companies, or you more likely get disrupted by somebody in a different industry. Right now, like one of the things that I’m seeing is marketers were one of the first to really embrace AI. For a marketer. The, the best way to try and sell somebody your product is to know them like a best friend.

It was never really feasible. Can’t [00:40:00] know. let alone millions of people, like a best friend. But guess what? AI has enabled that know, again, thanks to the psychology and neurolinguistics, behavioral all. Just for a hundred words, AI can know you as well as a best friend. It understands your hobbies, your opinions, your interests, even your political party, right?

And just scraping that information off social media, your LinkedIn profile, you know, wherever it’s publicly available, and so the marketers can actually know you as a best friend. The interesting thing is you look at some industries like the movie industry, the film industry. They still market $50 million for these movies, marketing budget, couple of big things.

And all they’re doing is kinda like a one size fits all. don’t want to believe that they could target all these people like a best friend, you know? And to put this in perspective, you look at the movie, I think Top Gun Maverick, he did about 600 million [00:41:00] domestic box office. If you strip out. People once saw the movie multiple times.

It worked out to, I think

squadcaster-04bj_1_05-28-2025_121713: All.

neil-sahota_1_05-28-2025_091728: or 32 million people in the United States. See the movie?

squadcaster-04bj_1_05-28-2025_121713: 10%.

neil-sahota_1_05-28-2025_091728: Yeah, 600 million bucks of great things. But when you ask the question, many people you think would’ve watched the movie, you start hearing numbers like it’s probably about 85 or 90 million. it’s like you captured a little over a third. You could have actually done so much better. You know, people fixate on the 600 million they don’t really dig into the numbers. It’s like, you probably left another $600 million on the table. Think about that.

squadcaster-04bj_1_05-28-2025_121713: If not a billion, if.

neil-sahota_1_05-28-2025_091728: Yeah. Complete different perspective. So it’s like there’s a better way of doing marketing that they’re now kind of scratching their heads going, like, how?

How does that work? You know?

squadcaster-04bj_1_05-28-2025_121713: Well, and this is enabled in relatively small companies. Again, [00:42:00] talked to a CEO group last night. One of the women said when she gets incoming inquiries or when her organization does, they do that scraping and they send back a customized response. And that organization is not as big as the. Top Gun folks.

It, it’s a moderate sized company.

neil-sahota_1_05-28-2025_091728: And that’s what I found. The small to medium sized businesses are more open to the innovation conversation, right? It’s not just a matter of survival, but also thriving these days. ’cause they’re facing a lot of different pressures. So while the big companies automate and kind of eat into some of their traditional business, they need to find different channels to be more effective. I mean, uh, the billable hour is getting killed off in the legal industry. And now even the big law firms are realizing that I have to bid on work. I have to bid fixed price. How do I even do [00:43:00] some of these things? You know, they’re looking at AI tools for cost efficiencies now so they can provide a more market competitive price because no one’s paying the billable hour.

squadcaster-04bj_1_05-28-2025_121713: So we did a project to create a vision playbook for a client. The client had two bids. Uh, and this was partially an experiment, so they got a bid. I was already working with them and we were able to produce it for about a third of the price using ai. So we still spent all of the time socializing the results, revising the thing.

But the amount of time and the amount of cost, they saved well over half. And the other person, highly ethical, principled, would do a great job. So I wanted. Uh, clarify it. It’s not like they were doing bloated or unfair pricing. It. [00:44:00] My assumption is it was apples to apples AI and not ai. And yes, my billable hours were fewer, and that allowed the client to get more value on that project and hopefully spend the rest of those billable hours on something else That was also value add.

neil-sahota_1_05-28-2025_091728: And that’s what we really want, right? We wanna give people more value, add work, more complex, meaningful work. And I think that’s, again, it’s, it’s a leadership thing in that we don’t tend to frame it that way, right? We’re, we’re gonna introduce some of these tools. We’re gonna automate some of these things. Don’t value proposition to your employees, resistance to any change.

squadcaster-04bj_1_05-28-2025_121713: Fear. So we are nearing the end of segment one. I wanna ask you then last question of this segment. What specific [00:45:00] leadership qualities are essential to navigate the challenges of integrating AI successfully?

neil-sahota_1_05-28-2025_091728: Well, other than reading the book we put together that outlines all those items, uh, the, the for leaders, I think one critical step beyond reading the book is to understand that successful implementation of AI actually requires hands-on leadership. You can’t treat it like an IT project. You can’t treat it like this is a box or something.

We’ve done hundreds of times. Is very much a people change. You know, four of the five points of failure are related to people, and so it’s our job as good leaders, innovative leaders, drive that change, to champion it. Show the value propositions, you know, to address those people concerns lead the charge.

Otherwise, the honest truth is you’re gonna be like those other companies that keep talking about all we did was fail, we failed again. Yet again, our competitors who [00:46:00] are not as smart as we are. With ease. But those, those companies, they, they tackle the leadership challenge when it comes to ai.

squadcaster-04bj_1_05-28-2025_121713: And it sounds like tackling the leadership challenge at the outset. Not after we define the solutions, all that stuff, but leaders charter the project, sponsor the project, actively engage in the project. All the way along. Domain expertise, risk management, cost containment, all the things to implement back to your five issues.

Partially, they’re implementing a business solution, not a software program.

neil-sahota_1_05-28-2025_091728: Yep. Right. Think of it as, again, AI is a high energy intern. You’re hiring an army of interns. What are you gonna do with them? Right? Just imagine that they were human. How would you handle that? Similar approach with ai, right? You’re not just gonna hand them, [00:47:00] here’s some requirements, Some.

squadcaster-04bj_1_05-28-2025_121713: Two or three. Yeah, I, I think absolutely. Yeah. Yeah.

Beautiful. So Neil, thank you so much for recording episode one, talking about the five failure points of implementing ai. In the next segment, we’re gonna get into more organizational context. So culture. Organizational [00:48:00] dynamics, ethics, communication, change management so that our listeners have a ro more robust perspective on implementation.

So how do reader or be, how do listeners learn more about you, about AI for good and what you’re doing at the UN and follow your really brilliant work.

neil-sahota_1_05-28-2025_091728: Well, you can come to my website, just my name, neil sahota.com. I’m always sharing out some of the latest information and trends in the industry. Um, if you’re looking for a lot more tactical help for your organization, please visit My Substack, which is just my name, Neil sahota@substack.com. Or you can also follow me on social media, work with the United Nations. We have a whole portal. Just Google, United Nations AI for Good, and you’ll find it quite easily. If you want. Yeah, if you wanna know the latest, greatest and you want some things, uh, best practices, I encourage you [00:49:00] to check out both my website, my.

squadcaster-04bj_1_05-28-2025_121713: Thank you, and until next week to our listeners, thank you for joining us today on Innovative Leadership co-creating our future. Are you interested in elevating the quality of your leadership? The Innovative Leadership Institute offers best in class leadership development, executive advising, and organizational transformation services.

Send me a note at inquiries@innovativeleadership.com. The world’s most successful leaders are created here.

​