The Future of AI in Business Mastering Effective Interaction & Resource Planning with Conor Grennan

Welcome to the Future Finance show where we talk about how the robots got treasury wrong and Spacely sprockets and how CogW's COGS capitalized on their mistake.

The strange thing about generative AI is that it's not really a learning curve kind of thing. The hard part, I think prompt engineering to me is borderline myth. More clear. Your prompts will help, but everything that you do in a prompt is just how you would instruct a new colleague.

You've gone all in like few people on generative AI. So what happened in that transition that you saw this and decided just to go all in on?

I don't think this thing takes jobs. I think it takes tasks. If you are in a job where you have a very limited number of tasks, then that job is at risk.

But you know the good talent is going to be okay. If you do a really good job at your work, they're not going to just fire you because, oh hey, we have generative AI really all about efficiency.

If you just had everybody trained on it, you'd see, oh, they're not amazing. The tool is amazing. So it really throws off talent evaluation. There's a 43% improvement in work by lower skilled workers and a 17% by higher skilled workers. It's going to take your job if you are using this because.

Future finance is brought to you by qflow AI, the strategic finance platform solving the toughest part of finance Planning and analysis. B2B revenue align cells. Marketing and finance seamlessly speed up decision making and lock in accountability with qflow AI.

So one thing I hear over and over from finance folks who want to use generative AI but haven't yet is that they haven't gotten comfortable with data privacy and security with these cloud based tools. And you know, it makes sense because when you're dealing with sensitive financial information, the idea of uploading that data into the cloud can set off all kinds of alarm bells, no matter how secure the platform claims to be. But here's the problem. The most powerful AI models live in the cloud. And that's where you find the models with billions of parameters trained on huge amounts of data. That's where ChatGPT, Gemini, Claude and all those models are. And you know, I've talked a lot before about how these providers offer premium accounts where your data isn't used for training the models. But for some businesses, especially those with strict data policies, even that's not enough. They want to keep their data on their own servers, period. So I've been thinking a lot and Playing around a bit with Meta's Llama. So Llama offers a couple of smaller AI models that businesses can run entirely on premise, no cloud involved. It's a pretty unique approach in the AI world and these models are getting better and better. So when we talk about Llama, we're talking about kind of a family of models. The three most recent ones are Llama 3.18B, Llama 3.170B and then the large model 405B. And the numbers after the version represent the number of parameters in each model and that's essentially the level of complexity and sophistication of the model. So llama 8b and 70b are the small locally runnable models that I was talking about. And, and they're designed for everyday business tasks, optimized for speed and for small data footprints. But there's no getting around the fact, you know, they're just not going to be as powerful as these massive models. So llama 405b on the other hand, is a, is a monster. It needs serious hardware and runs in the cloud and only in the, in the Meta environment. But it goes toe to toe with the industry leaders in terms of raw capability. So it's a spectrum. The smaller models give you that data security and low latency, but, but you trade off some of that high end performance and it's really all about finding that right balance for your business needs. So with these smaller local models, they do have some limitations. They can't match the raw power of big cloud based models and complex tasks like generating detailed financial forecasts from these massive data sets, they may not work as well with the smaller models, but for a lot of everyday financial work, think document analysis, basic number crunching, these local models can get the job done and they do it without any of your data ever leaving your control. So it's a trade off on one side. You've got the heavyweight cloud models that you know you've got to come to terms with what your data policy is going to be and what you choose to put or not put in those models. And then you've got the kind of, well, the rock solid data privacy of running AI in house. You just don't have the high end power and in finance and for a lot of businesses that privacy is going to be a top priority. So I think the good news right now is that these smaller local models are getting better all the time. Every time, whether it's the Llama models or Even the other LLMs out there are creating smaller models like ChatGPT 4O mini is one that still runs in the cloud, but we're seeing better and better performance on these smaller models and we're going to see them. These are the worst models we're going to get. They're going to keep getting more powerful and as they do, people will be able to keep that local control and have access to better models. So, you know, if you're in finance and you've been looking at AI but worrying about the data side of things, keep an eye on what's happening with these smaller, locally run models. They might just end up being the solution you've been looking for, in a way to tap into the power of AI without compromising on data privacy. Of course, it's all about finding the right tool for your specific needs. But with options like LLAMA in play, businesses that put data privacy first have more choices than ever. And in the world of AI and finance, that's a pretty big deal.

In this week's episode of you know your job is safe. For now, I'm not going to talk about the silly or ridiculous things that AI has said, how it hallucinates, how many of the answers can be less than ideal and make us feel very safe in our roles. What I want to talk about is what we can do to make sure we're safe in our roles. Because yes, AI makes mistakes. Yes, AI is not there. We haven't hit AG Artificial General Intelligence as Chat GPT laid out. There are five different levels and right now we're on level one, the conversational AI. We have a long ways to go to get there. In the meantime, what can we do as finance professionals to be prepared? First and foremost, find out how you can implement technology and AI into your workflows, whether that be using it to help write letters. An example I know of somebody early on that used AI to help write a collection letter and they collected over a hundred thousand from a customer. I know Christian Martinez has shown many ways you can use AI and Python to enhance your analysis. Seeing people use AI with vba I've used it in my work. Oh, I need a quick script. Do this or do that. I ask AI to write that. VBA have it help you write that formula, analyze data. Another great one from Jim o' Neal in a prior episode is he talked about how it can be the contrarian ask it for all the risks and all the reasons you shouldn't do something, all the counter reasons. So the more specific you can get with AI, the better the answers you can get from it. But the reality is you need to learn how to start using it so you could be more effective. Many people do not impact that. AI will take away a lot of jobs in finance, but one thing it could definitely do is companies will start looking for a productivity tax, almost an efficiency. Like, hey, we're not going to give you this many employees because we expect you to be 10 or 15% more efficient. You may delay hiring. You may have to justify every headcount and say, why can't AI do it? The better you are at using AI, the more efficient you are, the easier it is to maintain your role because you're being very productive. So this week, the focus is on being better with AI, not talking about the shortcomings of AI.

Welcome to Future Finance. This week, Paul and I welcome Conor Grennan. Connor is the Chief AI Architect at NYU Stern School of Business and the founder and CEO of AI Mindset, where he's helping leaders across industries harness the power of AI. Connor's also a New York Times and number one international bestselling author. His influence extends far beyond the classroom. He's been recognized by the Huffington Post as a game changer and honored with the Unsung Hero of Compassion Award by the Dalai Lama. Connor holds an MBA from NYU Stern, a degree from the University of Virginia, and lives in Connecticut with his wife and two children. Connor, welcome to the show.

Thanks for having me. It's good to see you guys again.

Yeah. I feel like we would need, like, two hours to talk to you, to talk about everything you've done, but we're going to get laser focused here and talk about.

Right. Yeah. Glenn, I could hear you talk for hours. I've watched all of your courses. Highly recommend. They're awesome.

Yeah. And we're going to get into your course. Yeah.

So what I'm going to do so we get this done in time is Glenn can't ask any questions. I'll cover this podcast.

Yeah. I do tend to ramble sometimes. Paul and I have a thing about that. I think I make our podcasts twice as long as they need to be.

No worries.

But he makes them good.

He.

He's great.

So I guess let's just. Let's dive right in. So the first time that I saw you speak, I. It was such a. I think we were doing an event together. Connor. It was, I think, for Chief Executive group, maybe your approach to AI has been. I quote you all the time and whenever anybody. Well, what I love you always lead with that. You're not a technologist, you know, you're here at the forefront of generative AI and training people how to use it, but you make it very clear that you're not a technologist. So, I mean, I guess tell me a little bit about your background and what happened that caused you to just. I mean, you've gone all in, like few people on generative AI. So what happened in that transition that you saw this and decided just to go all in on it?

Yeah, it was a strange transition. So my, my background is in, you know, academia, I'm a writer by background, things like that. I had spent the last 10 years as dean of students, dean of MBA students for NYU Stern. So that you can imagine kind of like that profile, right? And what happened was essentially chatgpt came out. I saw it. My wife is in the AI at McKinsey, but just sort of doing responsible AI, and she saw it. She's like, you got to try this thing out. I tried it out and immediately saw, wow, this is going to change everything. So I just, I was asking my boss, the dean, hey, is there. What do we have going on this? And I started attending lectures by, you know, AI people, machine learning people, things like that, which we have obviously at Stern. And I kept realizing over and over again as I started to learn more about machine learning and everything that I didn't need, that there was nothing to learn. There was. It didn't help me with ChatGPT and other large language models, and that was really, really curious to me. So I started to think, well, I think this is just, we just have to sort of learn it. So I started to teach it at nyu. And then another strange kind of thing happened, which is I started to do this with a lot of companies and help them out with this. But the other strange thing that happened was I had a framework and a model. It's like, hey, easy use cases to advanced use cases, etc. And I started seeing something really strange, which was the fact that this wasn't transforming how people worked. They would take the use cases I gave them in healthcare or whatever and keep on using those healthcare use cases. Even after I left, I'm like, but I don't know anything about healthcare. And I realized, okay, this is actually something really, really different from any kind of typical digital transformation. And so for me, I think it's actually really helpful that I don't have a tech background just accidentally because, you know, I'm relating to all these folks who also don't understand it. And I think sometimes if you have too much tech, you don't know what the hard part is, but I'm like, no, the hard part is this. So let's. It's almost like learning basketball from an 8th grade coach instead of Michael Jordan. I'm like, you know what you really need to do? So, yeah, it's a strange background, but. But it's also a strange technology. And that's how I got started.

I love it. That's. That's a great way to get started.

Yeah, I know. We were talking about the switch that Connor made from Dean of Students to Chief AI Architect, and you kind of addressed that there. So. Paul, I don't want to. I didn't want to steal your. Your thunder on that question.

Yeah, no, I. I'd be curious. You. You addressed that a little bit, but what's it like being the chief AI architect at. @ Stern? What does that involve? Obviously, you went from Dean of Students because you really got involved in this AI to being the chief AI Architect. So what exactly does that mean for the students? What does that role involve?

Yeah, it's a good question. Right, because we're really in uncharted territory here to a certain extent. You know, usually in an academic setting, there is a process by which you learn things and there's a learning curve to it, et cetera. E. The strange thing about generative AI is that it's not really a learning curve kind of thing. You know, it's not like, you know, if you're learning Excel or something like that, which I know you guys are very familiar with. You know, you take a long time to learn it, but then you're. Then you just go, then your life is easier. Then you're, you know, doing financial modeling and everything else. Same with. If you were learning French or something, it takes a long time to learn, but then, you know, then you get to speak it, and then you get the reward. This is much more akin to something like, you know, exercise or something like that, where it's nothing you have to learn, you just have to change your behavior. Behavior. And so it's not really like, okay, here's the curriculum by which everybody has to do this. It's much more like, well, who do we want to be as people? How do we change our culture? Like, you know, it's. It's that big change management question, which is much harder. So being in uncharted territory, I'm sort of trying to carve this out with obviously, you know, help from people much smarter than me. But. But my mandate really at NYU is to essentially level up the organization and by that I mean we want our students, our MBA students to go out into the workforce really being fluent in generative AI. We want our faculty to understand how to integrate this into their curriculum. Because as I don't have to tell you guys, you know, the proxies for grading have all sort of been set on fire, right? I mean, like around school settings. So they have to change how they work. Now it's hard for faculty who have been very successful for two decades doing this the way they do it. How all of a sudden do they change how they teach? Not exactly overnight, but pretty close, and then also leveling up the administration. So Stern is functioning as a really well oiled machine. That's a little more similar to the work I do with companies because it's, you know, running it, running a business. The faculty and the student are very, very different. They're all very different things. But I have to approach them in all different ways. You know.

Yeah, it is fascinating, especially the part where you said, you know, teachers and just the change rate because students are using it and there's a question of where's the right way to use it versus when you want them not to use it. For learning from the exercise and just managing and balancing all that, I imagine there's a lot of different opinions that you're on the subject that you're trying.

To manage the hard part. I think I've been trying to come up with an analogy for this for the long, for the longest time, and I think I finally did, which was when we say people know it and some people don't know it, there's not a, there's not a metric. There's no way of really knowing whether people know it or not. So I, the, the analogy I sometimes use to go back to Excel is if I'm talking to 500 people at a keynote or something like that, and if I say, hey, who here uses Excel or knows how to use Excel? Most of the hands would go up. And then, you know, because people are using it for, you know, grocery lists or personal budgeting. But then if I said, okay, so who in here has used it to create sophisticated financial models in order to, you know, price out a billion dollar company, you know, that kind of like an investment banker type use case, like five hands would go up or something like that, that's like generative AI has, everybody has tried it, but who's using it 50 times a day, et cetera. And there's not really a good way of parsing that out. So that makes it tricky. So even with students who feel like, oh, I know how to use it, I'm like, okay, so. And not, not to even, not to test people, but I'm like, how many times do you use it? Are you paying for a model? Have you heard of or youth Claude? Have you, you know, and like, things like that. Those are some of the ways you can start to sense like, how are people actually using it?

100 agree. There's all these different levels. What, what does it even mean to really use it, like you said?

Yeah.

Because are you using it to check an email?

Which is actually. You just threw me a softball, Paul. For my, for my next question here. And I've been this, I want, I'm gonna.

That was deliberate. I know what's coming. I, I, I know this is your ax to grind.

I'm gonna get the video from this. And every time someone mentions this to me, I'm just gonna play the video of Connor here. So one thing that drives me crazy. So Connor, you and I both teach a lot of AI courses and every course I teach, someone will raise their hand or their digital hand and say, are you going to send us the prompts? I think we share this. It drives me insane whenever I hear someone talk about being a really good prompt engineer. And that's not to belittle what, you know, if you're. There are areas.

Technical prompt engineering.

Yes. Yes. The way most people mean it is like saying, I'm really good at googling, you know, so. And I always refer back to you on this when I say the word prompt engineer to you. Where does that take you?

Yeah, it sends me to sort of like barfing in a bucket, right? I mean, because it's so frustrating. Like, I'm with you. Right? Because I think this is where people started off and I have a theory around this. Glenn, I want to run by you too. This is what I talk to people about now as well. Is that the reason that we want prompts and people want to learn prompting and all that kind of stuff is that our brain does certain things really well, right? It does like pattern prediction, automation, things like that. And your brain wants use cases, it wants clear patterns. It wants to know, like it is not comfortable with something where you just say, hey, use this in any way that you want to use it. People are like, okay, but what are the prompts? That's why, you know, ChatGPT started having the conversation starters and things like that, right? There's no reason for those conversation starters. They're so dumb, right? Like, it's like, hey, plan a trip to Peru. Nobody's going to do that. Or, hey, help me do a recipe for, you know, chicken masala. Nobody needs to do that. But it's for the brain. It's so that people can see. Like, see, you can just type in this, or you can just type in this. Prompt engineering to me is borderline myth. It's. I know that, yes, more clear prompts will help, but. But everything that you do in a prompt is just how you would instruct a new colleague. Give them context, give them examples, tell them what the outshoot output should be. Give them sort of like a sense of like, hey, this is what's good, or not. But the more context you give, the better. It's not about framing something, but people think that it is, understandably, because that's the history of technology. You need to sort of like, well, tell me how to use this. You can't just say, well, talk to it like a human. That's not helpful. And so the thing about prompting for me is that people want, like, this list of prompt libraries, which even back in the day, Glenn, you and I have known each other for a while now. Like, when people would send, like, oh, well, here's a thousand prompts, you and I would barf over that, right? Because it's like, think about, like, for a second, like a YouTube video that shows you how to clean your pool filter or something like that, right? And that has 9,000 views. All 9,000 people needed to clean their pool filter in that moment. Nobody's just watching that thing just to watch it. So, like, when you send out prompt libraries, who needs that exact prompt in that? It's like, oh, here's my favorite prompt for a sales call. Like, who needs that at that moment? You'd have to be. It's. It drives me bonkers. So the most important thing is talk to it like a human. Prompting is not a thing. However, that's hard because your brain is seeing this thing and your brain has a hard time. In the same way, if you talked to, you know, trying to talk to a baby like a college professor, right? Your brain has a hard time with it. That's the problem. It's a behavioral thing. It's not, give me more prompts.

Hopefully I didn't steal this from you. And if I did, I'm going to have to please steal away crediting you for it. So what I always say when I'm doing these, and people will ask and I'll send them. You know, when I do A training video. It's like, well, here's the prompts I used. But you're not always going to get the same exact response I did. So if you're just treating this like a binary, like copy, paste, copy paste my prompts, you're not going to be going down the same road that I am. So. So the approach that I tell them is it's just like you said, treat generative AI. And right now, I mean, with. After Strawberry and you know, when GPT5 or Opus or whatever they're calling the. Their next Orion, I guess, is whatever open AI is calling the next model after that comes out. This may not be true, but for right now, I say treat generative AI. The chatbots, you're using like a very, very bright but very, very green intern. So you have to give them their context. You have to set them up, but don't just throw, you know, so when you're prompt engineering, what you're doing is just the way you would, like you said, with a new colleague, interact with it this way. And if you, you know, if you get frustrated because there's things I ask you to do all the time, and I'm kind of pushing the limits at it, as. As I'm sure you are. And then if you're patient, you back up and you come at it a different way, you can usually get there. Like, it's. There's very few things that I've just completely stumped it on, but it can't go. But if I were just trying to use canned, you know, prompt libraries, then I wouldn't. I wouldn't get there.

No.

100%. Well, I was gonna say your whole AI mindset, that's the perfect name for it because you have. It's a shift in mindset. This is software. Yes. But it's not binary. It's. You're not flipping switches and turning knobs. You're interacting with it like a person. So AI mindset is just the perfect explanation for what you do.

Yeah, thanks. And the way you described is exactly right. You did not steal that from me, by the way. I've heard that before around interns and everything else. Yeah. So I. So I have this course, and one of the things in the course is this digital course. And one of the things I was doing was coming up with, like, a lot of different prompting frameworks that I've just used kind of instinctively, but putting names on them. And one of them I call Bezos in the corner, which is like, imagine if you had just Jeff bezos sitting in the corner, and you could at any point turn to him, be like, hey, I'm trying to do this marketing plan or something like that. The problem is that, like, you know, first of all, you have to sort of get into the headspace where you're actually taught. Even though it looks like Google, essentially looks like a search bar, you have to pretend that you're talking to something like Jeff Bezos. But more importantly, you know, if you really did have Jeff Bezos there, you wouldn't say, hey, give me five ideas for marketing. And then Jeff Bezos gives you five ideas and you're like, okay, get out of here, Jeff Bezos. No, like, you'd ask him, follow up, do you know what I mean? Like, and that's what's so different about this, is that it's a conversation going back and forth with this machine. And our Google brains, I like to call it, like, are more based on command response. Like, you sort of give something and then you get the response and then you walk away. That's what we've been doing since the history of the Internet. And so to get in that different frame of mind is that's a behavioral change. It's really hard. But to your point, like, it's not like, hey, here's the right prompt to use. You just have to get into this mindset of being iterative with it. And then prompts almost go out the window.

Ever feel like your go to market teams and finance speak different languages? This misalignment is a breeding ground for failure, impairing the predictive power of forecasts and delaying decisions that drive efficient growth. It's not for lack of trying, but getting all the data in one place doesn't mean you've gotten everyone on the same page. Meet qflow AI, the strategic finance platform. Purpose built to solve the toughest part of planning and analysis. B2B Revenue QFO quickly integrates key data from your go to market stack, an accounting platform then handles all the data prep and normalization under the hood. It automatically assembles your go to market stats, makes segmented scenario planning a break freeze, and closes the planning loop. Create airtight alignment, improve decision latency, and ensure accountability across the teams. I love what you said there, the editor, but I'll add two things. It was kind of funny. You say, talk to it like a human. I remember I was typing a prompt and I'll say please Sometimes just because it's habit, like I'm talking to somebody and the students were like, you say please to AI, like, why Are you? And I'm like, just have it. I don't even think about it. Please do this. And they were just like, you're stupid. And it was kind of funny. And the second is, you know, kids are able to make that mind shift so much easier. My daughter will sit and go back and forth with it for hours if I'll let her on things of creating stories for her. And she'll just be laughing and using it to come up with funny scenarios around her favorite characters. And when it gets it wrong, she just, no, this is what I meant. And you can see the conversation in it and, you know, versus the. Well, I gave you the prompt and all the instructions. Why isn't it perfect? So it's like, I have to hack through this. And I'm like, yes, sometimes you have to hack through it. And they were kind of like, well, then it's not useful. No, it's incredibly useful. You just have to be willing to put in the time.

Yeah, this. So I love how you said that. Two things on those two things. So the very first framework I ever came up with was last March, was the first YouTube video I ever put out on this stuff. And I was like, hey, I have a framework. Because, you know, you guys work in frameworks too, so you can't make fun of me. But, like, in business school, it's all frameworks, right? So. So I was like, I need to come up with a framework. So I call it htg. Hi, thanks. Great. And I said, you should say hi to it. You should say thanks to it. And then great is a little different because that's about giving feedback, because you have to give it feedback. But the first two, right to your point, it's just like, why do you say this? I love it when people. I. I once saw this one thing that came out said like, oh, don't say please because it won't give you good enough response. No, I'd rather keep my humanity and get a marginally a response. So I love it when people say hi and thanks and please. And why do I love it from a sort of a purely strategic standpoint, is that it gets you better results, but not because it makes the tool give you better results. It gets you better results because it puts your brain in the right framework, which is, this is a conversation. This is not. I'm going to talk to this. There's a. There's a clip from the movie her, which you probably have seen and remember, which is go back and watch this movie. It is unbelievable. Even now it tells you everything you need to know about where we are right now. But the cool thing is a 13 second clip I played sometimes for people, you know, when I'm doing these, you know, sort of workshops for companies and things like that, which is he's sitting there, he's already fallen in love with this AI, right? So he's having a conversation. He's literally falling in love. So there's no, you know, oh, I'm not. What is this thing? Right? And he's playing a video game, so his brain gets distracted. And she's like, oh, my gosh. Hey, you know, Theo, guess what? An email just came in. And he's like, oh, read email. And she's like, okay, I will read. Like, joking around and what's happening? And he jokes. He's like, oh, my gosh, I'm so sorry, but his brain was distracted. And so, you know, muscle memory pulls him back into this, you know, read email thing. That's how strong muscle memory and neural pathways are, right? And she's like, why are you talking to me like that? I'm not that person. So that's number one on why to say please really quick on the kids thing. My theory on that is that kids are very good at being like, I wonder if this. Can do this. Which is usually terrible if you have kids. Cause it usually means, hey, I wonder if I can ride my dirt bike into the pool. Stuff like that. But in this case, it's incredible, right? Because what it does is it pushes them forward. And that's why I think kids are so good at it.

Yeah, I know. When you said, drive to the pool, I read a few where I'm like, we might need to have a conversation with my daughter. You know, something that you're like that. That's something we should talk about as adults. So come to us and let's have that conversation. But most of the time, it's just. It's fun to read. So I totally get what you're saying.

And I'm going to go. I'm throwing a PSA out here. I always say please and thank you. I'm always very polite because when the robots take over, I want them to remember I was one of the nice guys.

They'll have that tracked.

Don't kill him. Don't send the robot dogs over to Glenn's house.

He's cool, so you're safe when Terminator comes.

So, you know, we talked about kids using AI and they are, you know, they learn languages faster. They're just, they. They can Approach it in a completely different way. Dealing with finance and accounting people, trying to teach them generative AI. It's, it's amazing how much pushback you get. And I'm wondering what your experience has been like and what I'm, what I most want to hear about is because I'm. Sometimes it gets, you know, you can get frustrated with it where people are just, oh, that'll never work. That can never take my job. And I'm saying it's not going to take your job, but look how it can help you. But what I. Yeah, I wonder because you, you give such a, you know, that really, you're really teaching people on this mind shift. So I'm wondering if you can give me any experiences. Tell me about when you're talking to people, whether they're taking one of your courses or at one of your talks where they, you know, maybe they're pessimistic or skeptical at the beginning and then you kind of see this aha moment. What's that? Like when they, you see the light go off.

I think the only way to get there is to demo it. So I, no matter how short a time I have in like a keynote, I've done things where. And Glenn, you and I have spoken on some of the same things before, I think. And it's, you know, even if you only have like, like 25 minutes or something, right? Which is a fairly short amount of time. Even then I always do a demo because if it's. Otherwise you're like explaining why a card trick is so cool instead of showing them the actual card trick. And so I think that I have to show them how fast this is, how much iterates, how much it maps onto what you need. Not you have to sort of like fit yourself into this shape, but it's like talk. What do you need? So I have to show them that you have to. That's number one. Number two, kind of from a, from a higher, you know, kind of 30,000 foot view. You know, I always sort of talk about. In terms of like, I don't think this thing takes jobs. I think it takes tasks. And jobs are not big monolithic things. You know what I mean? Like, it's, it can't take a job because you don't know what's in that job. Like, but it does take tasks away from you or speeds up tasks. And so instead of now, by the way, if you are in a job where you're, you have a very limited number of tasks, then that job is at risk. It just Is like, I mean, I don't want to sort of like throw something like a paralegal under the bus. But if you're, if your only job is to summarize long documents, you know that job is at risk because that task is at risk. And by the way, when I say at risk, when companies hire you, they are hiring you and your brain. The best data lake we have out there is your own brain. And so by itself, ChatGPT and other tools like this, they don't do anything, they just sit there. They need to. I always say, like, if I was going against a market, if you said, hey Connor, 20 minutes, come up with the best marketing plan you can, you can use ChatGPT, it would be amazing. It would be really, really good. I'd be like, oh my gosh, look at this. Like a 5 year old who's brought home the, you know, pottery ashtray to his parents. Look how beautiful this thing is. And then, but then on the other hand, if you give it to a marketer and say 20 minutes, do the same thing, it's going to be way better, clearly. Why? Because they know instantly, very quickly. That's bad, that's good. I need more of this. I need, you know, I mean, like they know quality. Like Glenn, you and I have, have friends, right, who join, you know, kind of mutual friends who are phenomenal at like, you know, AI generated art, things like that. Right. And why is it? Because they're great at that already. They understand already what art looks like and what art takes. So I guess all that is to say that when people are worried about jobs and things like that, it's a totally understandable thing. Well, I will say practically though, is folks, anyone in the sound of my voice, this lane of being like the generative AI person in your company is still open. I can't guarantee it. I guarantee it. If you can say, hey team, here's how. I'm thinking about it, here's how. And I don't teach in use cases, but in this case it's really helpful. Here's the use cases that will speed up our process. I put out a deck help ChatGPT, right? Then you use ChatGPT to write the deck. I put out a deck, let's do like a little lunch and learn kind of thing with the team and get the team manager in there and then get their boss in there, etc. This lane is open. Nobody fires the person who knows generative AI right now. That's what I would tell people. Yeah.

And I had to Say one thing that I, I love that you said there, you know, beyond just the tasks and nobody fires, but, you know, the good talent is gonna, is gonna be okay, right? If you do a really good job at your work, they're not gonna just fire you because, oh, hey, we have generative AI. So I love that because people get so worried about that and they, they get so concerned, like, ah, it's just gonna take my job. And it's really all about efficiency. And so I really, I liked your answer there, There was a lot of, a lot of good information.

I think that's right. And then to build on what Glenn was saying though, like, but it's really going to help you. And so a couple things are happening there at once, right? It's sort of like, you know, pre Excel, you could have the best accountant in the world, but then if somebody else comes along and has Excel, they are going to look a lot or they're at least going to work way more efficiently and, you know, just faster and higher quality because they have Excel. And kind of, to Glenn's point, like, if you have this, it's going to speed you up. And one of the reasons that I say that I always start with senior leadership teams instead of just, I used to just work with teams, now I said, you have to have either a senior leader or I have to work with your senior leadership team first. Because you have to set benchmarks because otherwise it throws off. You guys have seen this, right? It throws off talent evaluation like crazy. Like if you have a 10 person team and two of them are using, two of them were sort of like, you know, on the verge of getting fired, but they're the ones using generative AI all of a sudden, like, those two are amazing, you know, when in fact, if you just had everybody trained on it, you'd see, oh, they're not amazing. The tool is amazing. So it really throws off talent evaluation too. That's why everybody has to be using it.

Great point. And one other thing I want to add that I really, like you said, is I hear from so many people, I don't need to learn that anymore. Like, there's going to be no need to learn Excel functions or why would you want to learn any Python or Dax? And I argue those that at least know the basics, know enough, are going to be more dangerous with the tool because you can validate if it's wrong, you can push the envelope. I think, you know, Glenn's experiences and having kind of learned a lot about AI before he came and Those experiences helped him push the envelope. Wouldn't you agree, Glenn? Would you be able to do as much if you had the degree and the experiences you've had?

No, not at all. And I think, you know, on that note, so one of the first things that I did that just I had to like, get up and walk away was I, early on, this is 3.5 version of chat GPT and I got it to write Python code. For me, it wasn't back then, it wasn't great at writing code. If I didn't know, you know, how to QC it a little bit, it would have, I wouldn't have been able to do it back then. Now though, it's, it's right, it writes way better code than I do. So the automation there. And so even Python's integrated into Excel now. Great. A lot of Excel people get Python, but they're very slow coders while they're doing it. If you can just say, this is what I need, you can, you can prompt it better, not because you have a canned prompt, but because you know what to ask and then get those functions much, much quicker. And then as far as, like, if I have to do something for more than five minutes a day, multiple days, I'm, I'll spend two hours automating it. And if you have this understanding of code or whatever, you know, whatever your domain expertise is, you can figure out how to prompt it and how to check it and know that it's right so that you're not just throwing something over the wall for Jenny to solve for you.

This is, this is a great point and it's why, when people are like, oh, is this going to take my job? I'm like, again, it's, it doesn't have to take. Now look, there are some more vulnerable things, call center and things like that, but it's not going to take your job if you are using this because you will stay ahead of everybody else. That there's a statistic that came from the Jagged Frontier study, you know, the one that Ethan Mollock talks about, was involved in last year where it was like, there's a 43% improvement in work by 4 by lower skilled workers and a 17% by higher skilled workers. And I, I did that the early days where I would show that study, the higher skilled workers in the room would be like, well, see, they're going to catch up. I'm like, okay, well that's not how math works. But like, but more importantly, like, this is important that you use your skill set and Bring it in. When like. Like when I'm. When you're running code interpreter or something like that, it's running Python code, like. But I don't know whether I can trust it or not because I don't know Python. So it's one of these things where it's just way easier for a data analyst to be able to then quickly check your work in the same way that me as a writer, I can sort of see, like, oh, this writing is actually really good. Where somebody's not a good writer, I can see, oh, that's not good writing. So, yeah, those are still essential skill sets.

Totally agree. So I want to shift gears here a little bit. And next I want to ask you a little more about your course. You just launched your course for generative AI for professionals. What was that experience? Like, what are you hoping people get out of the course? Maybe just tell us a little bit about that.

Yeah, thank you for asking. Yeah, it's, you know, put it out kind of recently was a. Obviously a labor of love, like, a lot, you know, And I'll bet you guys were into this too. Like, there's never a problem of having enough to say. It's like, what do you cut down? So I cut down to four and a half hours, but don't. It's like 67 lessons, four minutes each. All that kind of stuff. Lots of demos and all that kind of stuff. But what I really wanted to do is just codify this all in one place for people and have it accessible. Because I get hired by companies a lot, but a lot of people don't have access to that, you know, sort of like being in a. In a workshop or something like that. And so I wanted to have something for. For everybody. So it really goes through, you know, this whole prompt, not prompt framework, but this whole, you know, thought sort of framework, this AI mindset framework, which is this Learn, execute, strategize, just a different way of thinking. Rather than, what's the use case for sales, what's the use case for the. Those are valuable if you're in those roles, but not until, you know, you really understand what this thing is. It also forced me really to codify for myself, like, well, what is a large language model? Like, what is machine learning? All that kind of stuff, which I knew basically, but to your point earlier about, like, knowing that stuff is helpful. It helps you to understand why it's hallucinating, not just like, oh, it's pulling from the wrong book in the library. That's not what it's doing. It's statistically thinking this is probably the next word. The San Francisco Bridge moved to Connecticut in 1863. It almost makes sense, right? But it's just wrong. So understanding those things and I take it from a really non technical standpoint and I just have tons of the prompting frameworks and I don't mean prompts, I mean frameworks like this high thanks grade model or this Bezos in the corner model. Like again, ways to think about prompting. And I just, again, it's pretty robust course, but it was fun to put together. Awesome.

That's really cool. I appreciate you sharing that. I did see that on LinkedIn. I thought, hey, I may have to find some time for that here. It looks like it'd be kind of fun to go through. So thank you for sharing that. And then I want to ask you, you've been an author of a book called Little Princess and it's about the lost children of Nepal. So can you talk a little bit about the book and that experience? We just love to kind of learn a little bit about that. Touch on something very different from what we've been discussing so far.

Yeah, that's right. That was my. When I say that I don't have a tech background, my background is in like Nepalese orphans. Right. I mean, like it's so I, you know, I graduated university, I graduated UVA, and then I went abroad for the next 11 years or something like that, working public, international public policy. And then in Europe and then I went around the world and I ended up in this orphanage in Nepal. And just right after the, actually when the war was still going on, I guess the civil war. And yeah, just a lot of crazy stuff happened, you know, like I ended up, you know, finding these kids and then losing them and then going to search the mountains for them. They turned out to all be trafficked kids. Like, then I met my wife, who's American, but I met her out there. It turned into. And I'm a writer. So I was writing all this as I went, and I'm like, this is kind of a crazy story as it's going. And then I wrote a blog and then an agent just sort of like found that blog and she's like, hey, we need to make this a book. And it was kind of very right place, right time. So it, it got kind of big and. But yeah, that was, that's a big part of my life. Like I just took my son to Nepal basically last year again. And Nepal is still a big part of our life for our Family. So it's. It was just a funny thing. I never meant to write the book or anything like that. It just sort of, you know, I'd already written it in the form of blogs, and we basically just took that, made it a book, and, you know, it turned out to be kind of like the story that people wanted at that time. I'm not a great writer, I gotta admit. I'm sort of like, I read it like an 8th grade level, but I've done it enough that I know it's entertaining at least. But I think that was appealing for, like, college students who are like, oh, this guy sounds like he's like, you know, not that great a writer. He kind of sounds like me. Right? So, like, that was helpful too, I think. And it was quite the adventure.

Sounds like some amazing experiences there and where you were really able to get back and help, you know, try to make a difference for those children, which I'm sure you did. I look forward to reading the book. I haven't read it. I learned about it as we're preparing for this interview, and I'm like, I'm gonna have to pick that one up. So I'll definitely give that a read. So thank. Thanks for appreciating that. So we just have a few minutes here left, but we do want to get two more questions in. Well, Glenn and I will each do one. So what we have is we use chat GPT to come up with 25 questions. We each take a little bit of a different approach. So I'll go first here. I have these 25 questions generated to get to know you a little better by chat GPT. And you have two options. You can pick a number between 1 and 25, or I could use the random number generator to pick a number between.

No, I want to keep the human in the loop here. No, I get to pick the number. I'm going with number 12.

You know, you are the first in all episodes to pick the number. Everybody has said random, so I love it. No, no, no.

Too much AI in this one.

All right, so number 12. What is a failure or setback that turned out to be a blessing in disguise?

Oh, wow.

Wow. We'll give you a minute on so many. No, I don't even know which one to choose. I can pick a minute. I think one is trying to sort of, like, get a job when I was probably 28 or 29 and that job falling through. And I'm like, you know what? I'm just. I've saved a lot of money. I'M living in Prague or Brussels at the time or something. I'm like, I'm just going to take my money and I'm forget that job. I'm going to go around the world for a year instead of taking, not taking the job that I didn't get. And that led to me going to Nepal. It just like changed my life kind of going around the world. So. And that's happened over and over again. For me, anything that I haven't gotten has turned out to be the biggest blessing. And so, I mean, and for me, my faith is important to be so like. But I came to faith late in life too. So now I'm looking back, I'm like, oh, that was. Apologies. I don't mean to sort of proselytize on your podcast, but for me, I'm like, oh, God. Was actually steering me in these very different ways that I didn't see. So looking back and I'm like, oh, right, that's what I was doing. So I really feel like everything happens for a reason and they're all good reasons.

Go ahead and cross both sides. That's fine.

Yeah. So it's funny that the human in the loop thing, because no one else has kept the human in the loop. My different approach is so right before the show and today, it was really like you were in the green room. We were about to go on and I looked at Paul and I said I didn't come up with the question. So I went this week to Claude instead of Chat GPT and I used the Opus model and I generated these questions. So normally what I do is I spit out the questions and then I just say, give me one of those questions. But now since you're a human in the loop, I've got them numbered here, so I feel like we should keep that going. So I'm going to have you pick another number between 1 and 25, but we reserve the right. I don't know, I feel like if we land on a bad one because I didn't even QC these things before they came out. If it's a dumb question, I'm gonna pick one. I'm gonna.

Okay, why don't you choose between six and seven? You choose whichever one you wanna. Seven Both these, I'm gonna tell you.

What they are and then I'm gonna. I'm gonna go. I'm gonna just override, but I'll be rapid.

Yeah, yeah, go ahead.

Let's see, it's one is. What do you consider to be your greatest achievement so far? And I don't know. The other one is how do you handle stress and pressure? If you have a good answer for stress and pressure, we'll let you take that one. Because I know you're like high energy and Paul and I are both all over the place too, but you are. I'm the curmudgeon when I'm doing it. You do seem to always stay positive. I'm sure there's. You have your moments, but is there, with everything you've got going on, is there a good answer for that?

Yeah, maybe I have. You're right. I have my moments. I don't have, like this, you know, oh, I go meditate or something like that. I do. I think I. I know myself well though. Do you know what I mean? Like, and, and that's taken, you know, decades obviously. But like, I have recognized in the last kind of 10 years or so that I'm a huge introvert, whereas my wife's an extrovert. So I always kind of thought I was like the big party pooper. So I think for me, when I get like stressed and all that kind of stuff, I know that sometimes it comes. I act like an extrovert all the time, but, like, left alone, I'm not sure I would see humans for weeks at a time, you know, like. So I think just knowing myself and knowing like, that's where I need to be, I just sort of like, I can relax and all that kind of stuff if I'm just like in this little sanctuary of mine by myself. So I think it's not one thing. It's just like knowing how I react.

To things is super helpful, excellent and human in the loop.

I like, know yourself.

You may change the trend now as people listen to this, I say, man, things went pretty well for Connor. We. We're going to have to take power back from the AI Overlords.

I love it.

I think we got a great bunch of great one liners here. The prompting barf comment, the pet peeves we've had in here.

Very professional.

We're gonna have some fun short videos from this one. So thank you so much for joining us, Connor. We've really enjoyed it.

Love seeing you guys. Thanks again for having me.

Thanks for listening to the future finance show and thanks to our sponsor, qflow AI if you enjoyed this episode, please leave a rating and review on your podcast platform of choice and may your robot overlords be with you.