From PepsiCo to AI Expert: Lessons for FP&A Pros to Lead with Mindset over Tech with Tariq Munir
In this episode of Future Finance, hosts Paul Barnhurst and Glenn Hopper welcome digital transformation expert Tariq Munir to explore how AI and data are reshaping finance leadership. The conversation centers on the core message of Tariq’s upcoming book, Reimagine Finance: A Guide to Leading in the Age of Data, AI and Digital, and how finance professionals can drive transformation without needing deep technical expertise. The episode emphasizes mindset shifts, the human role in AI adoption, and the practical use of digital twins and reasoning models in finance.
Tariq Munir is a keynote speaker, writer, and digital transformation consultant with over two decades of experience working with Fortune 500 companies. Formerly with PepsiCo, Tariq recently transitioned to full-time entrepreneurship. His work focuses on empowering CFOs and finance teams to lead enterprise-wide digital transformations. He’s also the author of the forthcoming book Reimagine Finance, scheduled for global release with Wiley.
In this episode, you will discover:
Why mindset, not technical expertise, is the key to digital transformation in finance.
How AI is changing leadership expectations and essential skillsets.
What digital twins are and how they can optimize finance processes.
The challenges of applying reasoning models in financial AI use cases.
Why critical thinking and human judgment remain vital in an AI-driven world.
Tariq shared his inspiring transition from corporate executive to digital transformation leader, making this episode a compelling listen for anyone aiming for future-proof finance in the AI era. His reflections on leadership, his vision for human-centered AI adoption, and practical strategies like digital twins offer essential insights for finance professionals seeking to embrace innovation, drive change, and lead with purpose in an increasingly data-driven world.
Follow Tariq:
LinkedIn - https://www.linkedin.com/in/tariq-munir
Website - https://www.tariqmunir.me/
Join hosts Glenn and Paul as they unravel the complexities of AI in finance:
Follow Glenn:
LinkedIn: https://www.linkedin.com/in/gbhopperiii
Follow Paul:
LinkedIn - https://www.linkedin.com/in/thefpandaguy
Follow QFlow.AI:
Website - https://bit.ly/4fYK9vY
Future Finance is sponsored by QFlow.ai, the strategic finance platform solving the toughest part of planning and analysis: B2B revenue. Align sales, marketing, and finance, speed up decision-making, and lock in accountability with QFlow.ai.
Stay tuned for a deeper understanding of how AI is shaping the future of finance and what it means for businesses and individuals alike.
In Today’s Episode:
[02:23] - Writing "REIMAGINE FINANCE"
[03:45] - Key Takeaways For Finance Leaders
[06:02] - Career Shift And Current Update
[09:56] - AI And The New Breed Of Leaders
[14:57] - Balancing AI And Human-Centricity
[20:21] - Jobs And Technological Change
[28:42] - Challenges With AI Reasoning Models
[36:00] - Fun Questions & Episode Wrap-Up
Full Show Transcript:
[00:01:51] Host 2: Glenn Hopper: Welcome to Future Finance. I'm Glenn Hopper. Along with my esteemed colleague, Mr. Paul Barnhurst and today our guest is Tariq Munir. Tariq is an international keynote speaker, writer, trainer, and digital transformation expert. He bridges the gap between technological innovation and human centric solutions. With over two decades of experience working with fortune 500 companies to rate brings a unique perspective on how organizations can embrace AI and digital transformation while keeping humans at the heart of change.
[00:02:23] Host 1: Paul Barnhurst: You know, Tariq, you recently published a book titled Reimagine Finance: A Guide to Leading in the Age of Data, AI and Digital. What was it like writing the book? Let's start there.
[00:02:35] Guest: Tariq Munir: So I think Glenn can probably attest to that. Writing a book is challenging, but it is an incredibly rewarding experience. Like, I would do that again. Anytime you learn so much, you get so much out of your research. You come in getting out of your comfort zone, talking to people, understanding different perspectives. So it's a rewarding and highly energizing experience. I spent pretty much two years interviewing CFOs, critically evaluating my own experience in the digital transformation space. Reading journals, reading books. Debating. Arguing with people. I had the opportunity to talk to some of the brightest minds working in the digital space. So all I can say is that this is one of the most rewarding experience I would have got, and a humbling two as well, because sometimes when I believe that I know a lot, when I talk to people and I go out there, I realize that, you know, there is a lot that I still don't know. So I need to learn, learn more, and stay humble.
[00:03:37] Host 1: Paul Barnhurst: I love it. So it sounds very rewarding, but it can also be humbling at the same time. Love that. What do you hope finance leaders take away from your book as they read it? What's the message you hope they get? What do you want them to kind of come away with after they finish?
[00:03:54] Guest: Tariq Munir: So very simple message. Paul. And that is it is all about how finance leaders can elevate themselves from that back office support we are known for, to the digital change agents in the organizations who are the navigators and catalysts of the business growth. At the end of the day, when we go for transformation, when we go for digital transformation, the ultimate goal is that eventually we want to redesign the business model, we want to reinvent business operations, or we want to find new avenues of growth. And finance should be the one leading that. Finance should be the one who are the catalysts of that. And that is the entire premise of the book. What I really want finance leaders to take away is that you don't need to be a technology expert to lead digital transformation. All you need is a right mindset and a redesigned business model, which acts as a foundation of your digital transformation journey. And then of course, I had. Based on my experience and research, I offer a playbook for, uh, for CFOs and finance leaders to orchestrate such transformation. And above all, again, it is all about shifting their mindset, which effectively changes everything. How can we use technology to solve some of the biggest business problems, and also how finance captures and delivers value to the broader enterprise? So I do hope that after reading the book, a lot of finance professionals would be able to at least move in the right direction of leading the enterprise wide digital transformation.
[00:05:27] Host 1: Paul Barnhurst: I really like that, Glenn. I can tell you're itching to say something. This is your wheelhouse.
[00:05:32] Host 2: Glenn Hopper: I'm itching to go off script because, Tariq, since we last spoke, you've made a pretty big change. And I didn't even realize this when we were putting together the questions for the show. But I'd love to hear about it. Because when we last spoke, you were full time at Pepsi, and you were also doing all this thought leadership. You were working on the book. You had a pretty big following on just your guidance around this. I'd love to hear about the change that you made. What kind of drove that shift? And then, Paul, I'm getting a bonus question because I'm going to ask my planned question right after this. So I'd love to know what it was that drove you to make this change and how it's going and all that. If you could give us an update on one kind of where you are right now.
[00:06:11] Guest: Tariq Munir: Well, of course, Glenn, of course. Thanks for the question. And when we last spoke, I was still at PepsiCo, of course, and I was doing my full time job and also trying to, you know, juggle between these different thought leadership stuff and what I would call my passion and purpose. So effectively, I've been thinking about this for quite a while. It's not something that, you know, one fine morning I woke up and I'm like, okay, that's it. I found my purpose. You know, I was reading, um, I think, Hidden Potential from Adam Grant. And he once said that you can find it's not just that you find your purpose. You can build your passion, or you can build your purpose over a period of time as well, right? So you can find something, you can find a passion for you and you build yourself. And then eventually it can create. It can become your purpose also. And it's something you feel fulfilled with. And this is a similar thing that happened to me from a digital transformation point of view from AI and all those technology advancements. So I have been upskilling myself. I have been trying to, uh, you know, getting into that while I was in the corporate as well to get those kinds of experiences and gigs, which will actually help me lead transformation and bring more experience and index based.
[00:07:24] Guest: Tariq Munir: So I have been thinking about I've been doing that for the last 4 or 5 years at least. So uh, then uh, towards the end of last year, I made up my mind that, uh, it is now time for me to unleash myself into the into this space, because I did believe deep down that this is what I'm very passionate about. I love going out and talking to people who are smarter than I am, like in this podcast. And I end up learning more. I end up growing myself more. My book is coming. First book is coming up. I have built a reasonably good, good, good brand in the market or profile in the in the in this space. So I can now use that, leverage that, go out, take a leap of faith and then fully work on my own gig. Now this is, this is from leaving the corporate job to moving into this, uh, this side of the entrepreneurship. It's an entirely different experience. Right. And it is, of course, there are. First of all, I would say it's not as glamorous as a lot of people would see. Would think it is.
[00:08:29] Host 1: Paul Barnhurst: We're all laughing because we can relate. You know, you're talking about.
[00:08:34] Guest: Tariq Munir: It is not what people think that is okay. You know, I mean, it's like four hours a day, you work and then the rest of the day you are just, you know, relaxing. That does not happen. I woke up in the morning. I'm working probably more, but I'm feeling a lot purposeful and a lot fulfilled from a point of view that whatever I'm doing right now is something that I had a vision and I have a plan towards, or just simply something I enjoy doing a lot. So it is going really great. I'm enjoying it. I'm, uh, my first book is, as I mentioned, is coming up, uh, it's releasing in September. I was, uh, lucky enough to get, uh, get a deal with Wiley. So I will have a, like, a global distribution and that kind of stuff. So that is really good. And, uh. Yeah, I'm, I'm, I'm, I'm working on some really exciting stuff on building digital roadmaps, helping CFOs specifically to build digital roadmaps and, and really deliver, uh, and make some change in, in what I want this space to look like in probably 5 to 10 years.
[00:09:38] Host 2: Glenn Hopper: And that's I mean, I guess that's what you saw is this is the time and sort of what I'm saying around generative AI bringing so much more attention to the work we've been doing for years. And I know you recently spoke on the topic of AI and the new breed of leaders, and you have an opportunity now to go help influence and guide them. So how do you see that new breed of leaders? How do you see AI changing the way people lead going forward, as it becomes more and more just ingrained in what we do?
[00:10:07] Guest: Tariq Munir: You're absolutely right, Glenn. And one of the missions I have is to how do we change our mindset? How do we change our future leaders, including myself as well? Right. I'm not excluding myself. So to be ready, not just ready for the future, but really not just survive, but thrive in that future space. I do believe that there are a couple of skills that any future leader would like. I mean, finance specific, of course we are talking about. But even non finance this is as applicable as non finance people as well. First of all, none of the skills that I'm going to talk about include anything technical, right? So, so with everything happening around us and you rightly pointed out generative AI and now there is this AI space, you name it. I mean, the world is changing. Everything we are working on is evolving at a, at a like it's become a cliche now that AI is revolutionizing finance at a rapid and at a rapid pace. Now, what happens is that whenever these kinds of changes are happening, it's exhausting as well. You come up with one technology, you make a change in a process, then suddenly something else they change. Business environment change. We have geopolitical issues. We are having a talk on politics earlier.
[00:11:24] Guest: Tariq Munir: So all of that is changing the entire environment. So we are always in a state of flux change. That is where I do believe that the future or leaders need to be effective change managers. That is the first most important skill any leader needs to master, I would say. Secondly, with AI or other technologies deciding on behalf of us or creating an information overload for us, critical thinking becomes the single most important skill in the future. Or as of today, because of the very verbosity and the very volume of information that is coming out of generative AI. Again, as we talked, I think, in our last podcast, as well as learn that it's not deterministic. It is open to interpretation. It can vary between two different outputs. So how do you critically evaluate that? That is one of the most important. And we do not have skills and we do not become a victim of what is called machine output that we create an overreliance or or an automation bias. We can call that a yes if it is created by machine or if it is created by AI. It must be good. It must be great because, you know, I mean, it's trained on billions and billions of data.
[00:12:39] Guest: Tariq Munir: So that's very important. And the third most important skill I would say is around having that courage to be data driven. Now with all this information that is coming out, with all the technologies that are helping us now to make better decisions, data sets at the core of it, and unfortunately or fortunately, what happens is that we are we have started to lose. Our leaders have started to lose their narratives. We were used to doing things in a certain way, gut feelings, things and okay, you know, I know the business, but data might tell a very different story. That courage to be able to rely on data and then drop your own assumptions in the light, what data is telling you. And I'm not saying do not talk about your gut feel, of course, that what makes us human instincts makes us human. But when data tells a different story and we are thinking differently, that is the time we start challenging ourselves, think critically and be data driven. And lastly, how is it going to pan out? What technology is going to do so there are no deterministic black and white views. There are more gray areas than ever.
[00:13:48] Host 2: Glenn Hopper: Absolutely you do. When you were talking through that and thinking about the change that we have to make in leadership, I heard someone the other day and I apologize. I cannot remember who it was. Make the comparison that generative AI right now is like having a tiger cub. Like it's really, really cute. Where we are with AI right now, but it's a cub and you can control it. But what are we thinking? How are we going to plan as this gets more and more powerful when it could end up? Obviously not a doomsday or around it, but it is like thinking about how we are in management and you're still not making moves to understand and implement AI. I mean, then you're just ignoring the tiger cub and then it's going to sneak up behind you if you I don't know. I'm maybe going way too far with that metaphor, so I'll stop it there. And we do have other questions, Paul, now that I've monopolized the first half of the show.
[00:14:39] Host 1: Paul Barnhurst: Come on, Glenn, just admit you're a doomsday prepper. It's okay. You've talked a lot, and it's very clear that, you know, you see, AI is incredibly important. But, you know, just as I'd say even more important is that human element, right? That we've been talking about the mindset. You've mentioned that several times, being willing to change the mindset. And so with your emphasis on maintaining the human touch, you know, making sure that human element is there is you're doing AI projects and transformations and bringing in AI. How do you balance the two? How do you keep that balance between, hey, look, we need to use AI. Here's what the data is saying. We need to get to that, you know, level where we can trust it in certain situations, know when to challenge it, and make sure we have that human in the loop and you get that right balance. It feels tricky to me. And so I'm just curious how how you manage that or how do you think about that.
[00:15:35] Guest: Tariq Munir: No. Yeah. Great question Paul. And it is challenging. It is not. There is no single formula or a hack that we can say, okay, you know, if we do this, we will bring human centricity. It will depend upon each and every scenario. And it might differ. But there are certain principles that I followed. I would say maybe two principles if I really categorize them. Number one is that first of all, we need to bring a little bit of sensitivity to this topic around AI replacing humans. When we approach AI with this mindset that I will replace humans, I will replace the people. We are putting people in a box, and we assume that people or humans are some kind of entity that can just be replaced. One for one. Was that one AI or two AI replacing, replacing, or a group of a genetic Replacing a human. That is, I would say, if as a minimum it is the desensitization of this narrative. Secondly, So we need to stop the victim mindset, that victim mentality. And we need to start owning AI at the end of the day. The basic principle of any AI responsibility, or any AI system or technology system today is that humans are responsible for the output. At the end of the day, we are creating AI and we are using it for our advantage or some other person's disadvantage. Right now, we need to make sure that we are then utilizing AI to our benefit while keeping human at its core.
[00:17:11] Guest: Tariq Munir: Anything we do in the organization should be able to make, uh, make people live better, solve some of the problems, create access to information which was not available or otherwise. Even people who will have issues upskilling or who will have problem upskilling or reskilling themselves. They will. There will be an impact. There are no two views about that. But there is no. Again, there is no deterministic black and white view. We have to look at these things on a spectrum. Then the third one would be to be able to create a safe space for humans, for people to experiment. Now, any AI system in the world, or any digital transformation in the world at its core requires innovation, requires experimentation. So we should be able to bring that psychological safety within our teams for them to be able to bring out new ideas and experiment. Not mindless experimentation, but of course, experiment within some, within some guardrails and within some set rules and principles. Kate Isaac is a senior faculty at MIT. She wrote a research paper on nimble leadership. So this is what I mean, if anyone is interested. I mean, do check that out. So this is how you keep the human touch. Again, it's not a simple matter of one plus one or doing a simple equation or something. It's a lot more complicated.
[00:18:37] Host 1: Paul Barnhurst: Ever feel like you go to market teams and finance speak different languages? This misalignment is a breeding ground for failure in pairing 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 of B2B revenue. Q flow quickly integrates key data from your go-to-market stack and 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 breeze, and closes the planning loop. Create air-tight alignment, improve decision latency, and ensure accountability across the team.
[00:19:46] Host 1: Paul Barnhurst: It rarely is just a really simple thing, but like I said, well, there's a balance there. I think it's important to try to remove the idea of, you know, we're victims, like, right. We've created AI, we're all using AI. We have to learn to adjust to the impacts it has because. Right. We'd all be naive to say nobody's jobs are going to be impacted by AI. Of course it is. Some people in a positive way, some in a negative. Some will lose it, some will gain new jobs. You know, with the new economy, things will come up. We'd be lying if we said otherwise. I always laugh at people like, oh yeah, I raise jobs safer. Everybody's going to lose their job. And I did lose your job during the internet or the computer or steam engine. And was everybody's job safe in any big technological advance? No.
[00:20:31] Host 1: Paul Barnhurst: So I really like the human touch. And I love what you said about the leader, the nimble leader, not the command and control. Right. Because we're in an age where AI empowers people. If they use it, you can be much more efficient, much more effective. You can experiment and do things that sometimes would have taken forever. You know, for example, I had a deck that I wanted to do in a certain style, and I took all the images and created new images in ChatGPT over the weekend and put together that deck. You know, if I had paid someone to do all the images in that style and have it done, I would have spent thousands of dollars a year ago and probably waited three weeks. Yeah, I did two decks over the weekend with just ChatGPT and a $20 a month account. You know that that and there was a lot of innovation and things I could do within that. That's just a totally different world than we used to live in, which gets to being nimble and adjusting to change that you've emphasized and mindset is just embracing it. So I want to shift gears just a little bit. I know you and I talked about this when we've chatted, and I know you talk about it, you know, and the work you do. Digital twins. I know you're a big believer in the use of digital twins to help us optimize our processes. So can you start first by just telling our audience? For those who may not be familiar with that term, what is a digital twin? Let's start there.
[00:21:49] Host 2: Glenn Hopper: Before you answer, Terry, I want to warn you. Paul saw the 1996 movie multiplicity with, uh, like Michael Keaton and Andy McDowell and Eugene Levy were in it. He's ready for you to tell him that he's going to be able to clone himself so that he can wake up later than 4 a.m. every day. What he's hoping for here.
[00:22:10] Host 1: Paul Barnhurst: Actually, what I always thought of for twins was the movie twins with Arnold Schwarzenegger and Danny DeVito. And I'm Danny DeVito. The leftover sperm. If you remember that show. That's really what I think of, not multiplicity, but I think they both came out around the same time. Yeah.
[00:22:26] Host 2: Glenn Hopper: I think maybe.
[00:22:28] Guest: Tariq Munir: Again. I mean, it's not, it has nothing to do with us cloning ourselves or, you know, our digital avatars inside. So. Yeah, sorry about that, Paul, but I keep.
[00:22:38] Host 1: Paul Barnhurst: Popping up before.
[00:22:39] Guest: Tariq Munir: It's okay. Now, this digital twin concept has been there for, like, decades. I mean, it's not. It's not a new concept. It's effectively a digital replica of a physical product or a system. So most of the time, we're digital twins working at the moment is around building digital twins of the entire production plants where you can then simulate and then based on real data, you can, uh, you can create different scenarios. What has happened in the recent past is because of this advancement in AI, and since we are creating so much data, suddenly digital twins have become much more powerful than they were, and they have become easier to build. Also, with IoT, Internet of Things and all those sensors out there, we are able to capture a lot of data, simulate that data in an environment to replicate the real world environment of, say, for example, a wind turbine system. And then similar to that, in fact, countries like Singapore, Singapore was the first country to actually have a digital twin at a country level. So they were able to then use that to, to, uh, do their urban planning environment. They were facing environment issues, environment, uh, planning and so on and so forth. So in finance, I wouldn't say building a replica of a country, which it will be a bit too much, but what has happened is that recently we have seen a lot of advancement. Again, the technology, these technologies have been there for a while, but with AI coming up and with all the data access and computing power, we have seen a lot of advancement in process mining, task mining and data mining.
[00:24:18] Guest: Tariq Munir: Now, as a very basic for a process, effectively think of your any end to end process and imagine you have a digital replica of that process where you can actually play around with the process and think about how do I change certain scenarios or certain inputs or variables, and then what will be the impact on the process. So for example, we use a lot of examples of integrated business planning. So it goes through the entire process of your portfolio planning. Then you are determining your demand. Then you are doing your supply planning. And then based on that you are running your machines and everything. So instead of doing all of that in real time, in actual scenarios, you build up, build a small replica of that, and then do all those changes on, on, on there and in scenarios there. So what happens effectively is that you process mine. Process mining only works on digital systems, but in finance, a lot of stuff happens on our computers, on our Excel sheets. For that, there is a technology called task mining. So you can combine the two task mining and process mining to effectively create a digital replica of your say for example, your FP&A process.
[00:25:32] Host 1: Paul Barnhurst: And be scared that for some of my processes they would have been big red everywhere. Inefficient. Inefficient, inefficient.
[00:25:40] Guest: Tariq Munir: What really makes this digital twin is because, as I said, that the recent advancements in AI, which has helped us to build those AI engines and which most of these comes out of the box in different tools as well these days, where you can then now actually link either a real time data or a near real time data pipeline into those digital twins, and then your ability to simulate that different scenarios, that what makes that model alive. And that's what makes that model a digital twin. You can think about, you know, changing, for example, to our earlier demand forecasting. Kind of a scenario you can think of changing a product packaging now that will have an impact on your raw material sourcing, that will have an impact on your demand, on your marketing plans, on your ESG matrices, your financial matrices, all of that. You have a digital model. You just change a few variables and then based on that, similar different scenarios, how will it impact that? I have worked on one digital Twin where we did that in order to cache processes. And, you know, literally it's like it's mind blowing how quickly it can bring the work that we do of process optimization just itself from weeks to literally days. It actually challenges a lot of conventional Six Sigma. And again, I mean, I have nothing against that. I mean, it's a great way of process optimization, but it just condenses that a lot. It just condenses it to a few hours from like a workshop, which we would do in five days. We can just do that in pretty much a couple of hours.
[00:27:14] Host 2: Glenn Hopper: Thinking about where we are with technology and the ability to do that even a couple of years ago, how much more complex. And I know you've been in this space for a while, but just everything is moving so quickly right now. And I, Paul, I'm not going to do my rant about agents again. I am trying to get Marc Benioff. I'm not trying to get on his radar. So he starts a Twitter fight with me or something about agents. But thinking about digital twins and thinking about the way that people are trying to use generative AI just off the shelf, you know, ChatGPT or Claude, whatever tool they're using to solve problems. And you talked about an AI kind of what the leader will be like in this new sort of AI driven world, but we all sort of have to shift for now. And maybe the technology gets there soon enough that we don't. We end up not having to shift, but thinking about our role as we interact with AI are when I do these training or if I'm building not an agent Paul, but an agent workflow.
[00:28:14] Host 1: Paul Barnhurst: Say this is a soap box. He very much wants everybody to know there's a difference between the two. It's Glenn's passion right now and he's been on it for a while. So just humor.
[00:28:25] Host 2: Glenn Hopper: Because every time I hear an agent, I want to say, is it really tough?
[00:28:30] Host 1: Paul Barnhurst: Or.
[00:28:30] Host 2: Glenn Hopper: Is it a chat bot that has some neat wrapper on it? You know, so and I've also knocked prompt engineering because people who normally say I'm a prompt engineer, it's like saying I'm good at googling. It's. However, I'm getting Paul ultimately to my question here. That's a real if you're designing these workflows and if you're trying to figure out how to get complex outputs out of the AI you mentioned, I know you're as is up on the reasoning models as anyone else, but you recently posted about how AI struggles with complex reasoning tasks, and that's a big part of why agents aren't working right now. They can't. They have to. Things have to be simplified for them. And I'd love to hear your tape because I struggle with this in my day job. One that pays me more than podcasting. But it's really hard. You have to break it down into the smallest tasks that you can and work through them sequentially, even though they're getting better. But tell me about what your experience has been, why you think the models, even the reasoning models, are still having an issue with that, and where you're seeing improvements and kind of where you see us using AI in finance, because if you're in marketing and sales and you're writing copy and it gets a little bit confused, it's an easy correction. But if you're trying to do some kind of complex forecast or do a multi-step analysis, it's a mess. So that's a big part of where we are and where we're going. And I'm wondering what you're seeing and how you're recommending to address that.
[00:29:57] Guest: Tariq Munir: You are spot on, Glenn. I actually had a very, very recent experience where we were evaluating certain workflows around how can we bring it into that space? And, um. I mean, evaluating from a point of view of of of of streamlining and, and and digitizing, but of course, uh, you know, looking at technologies like authentic AI and, and what can we do? And it comes down to the point where I always advise the CFOs, and I always advise my clients as well, that, you know, go to the lowest level. Yes. The way we we work with ChatGPT these days and a lot these days, it seems like, you know, there's a lot of information coming out. It's very verbose. It has it is they are fluent. They will create output like in seconds and it will just or you. But when you are talking about exactly a process, for example, of your treasury, for instance, for your payroll, for example, we are talking about a very well-defined process in which you need to define every single click that you are trying to do, because if you will not define to to that granularity then then usage Nikkei or whatever workflows, you will not be able to to really use AI to your advantage. And in fact, in many of the scenarios, you will notice that a lot of these workflows does not even need authentic AI. But what happened that recently Apple published a study called Illusion of Thinking so, which effectively really revealed that the researchers found that the reasoning model literally breaks down as the complexity of the task increases.
[00:31:34] Guest: Tariq Munir: At the core of these models, it's all probabilities, right? They are, at the end of the day, probabilistic models trained on huge, huge, huge scores and scores of data. So whatever you do, whatever question you ask, they will create an output based on some probabilistic models or some neurons fighting, right? It is just like, you know, when my three year old, you will have a question about everything, right? Because it is how their, their, their brain works, right? They will just ask questions. So similar to that, I will just give you answers whether it lacks reasons or not. That is for us to decide. Now the problem here is that these models, unless you are very specific and you try to just box them in, they overwhelm us with so much information. And again, I'm not against models again. Again, of course I am. And that this is what I do and I do believe in those. But at this stage what happens is that they overwhelm us with information. But what is important here is to understand is that the verbosity of a model does not mean a lot that it is logical, or it is actually thinking it is a good thing for us humans to bring our own unique perspective and critical thinking on top of an AI model or on top of the reasoning model, so that we are not over relying on on information from from here.
[00:33:08] Guest: Tariq Munir: If anything, this makes it even more important for humans to have that ability to question, validate the output, and not just rely on it at face value. If Paul will ask a question, this question that you have asked me right now, if Paul will ask it to a ChatGPT and, uh, the same question in the same words Glenn would ask her. I would ask it to give three different answers, and three will be equally compelling, probably equally verbal and, uh, very and sound. I mean, depending upon what you ask it to do. Very intellectual as well. But that does not mean it's actually thinking also. So it's important to understand these differences for finance leaders or for any business leaders as well. What is the technology that we are using? What it can and cannot do. What are the limitations of those technologies? What is important for us humans? Our ability to question. So that is what we need to build. To be able to question the output, to be able to question what is happening in this space of AI. And then, you know, do a bit of an effort to go out and understand these differences, which is very important.
[00:34:18] Host 2: Glenn Hopper: Two thoughts on that. Then I know we've got to get to our fun section. Paul. We got it. So first off, I don't know if you've seen it yet. It just came out right after the apple paper did. But there are some pretty interesting critiques of the apple paper. And a lot of people are saying that, um, Apple released that paper as sort of a justification for how far behind they are on AI. But look at the counterpoints to that, because there are some interesting nitpicks around it. Uh, the other one, as you were talking, you know, Yann LeCun and actually met us having a tough time with AI as well. They're getting left behind, and they just are paying an exorbitant amount of money for a scale, which is not even an AI company. They're a data labeling company. They're paying like 15 billion for 49% anyway. It's a ridiculous amount to pay because they're struggling to catch up and there's a whole thing around that. But, Yann LeCun is brilliant , you know, he heads AI for him, and he, um, he says that, you know, he frequently says generative AI is no smarter than a cat. It's basically very good at mimicry and all that. So as you were talking, it was making me think about what he said. And it is, you know, I feel like we just blew past the Turing test and, uh, nobody even mentioned it. But so the mimicry part is very good, and that's, that's that human element. You can get good information out of them, but you have to discern when it's giving you just anodyne, washed out, just cliche stuff back to you. That sounds good. It's just but it's ultimately word salad. So I know we're I know we're running along, Paul. And we gotta get this thing back on track here. So maybe we should jump in with the fun part of the show here is that.
[00:36:00] Host 1: Paul Barnhurst: Yeah, let's go ahead and head there. In the interest of time, I'm sure. I know it's getting close to the next hour for Tariq as well. We probably all have things. So here's how it works. This is our, uh, 25 personal questions. We fed your profile. Let it look at the internet, the questions. We came up with the bio and said, basically, come up with 25 unique, personal and fun questions that we could potentially ask our guests. So these were all crafted by ChatGPT. You know, the only human in the loop was the prompt, basically. And so I do it two ways. Glenn does a different way. You get two options. You can pick a number between 1 and 25. And I'll read the question based on the number you pick. Or the random number generator can pick a number between 1 and 25. And we read that question. So which one are you picking?
[00:36:50] Guest: Tariq Munir: I'll just pick my pick. Epic. Let's keep the human in the loop.
[00:36:53] Host 1: Paul Barnhurst: What number do you want?
[00:36:55] Guest: Tariq Munir: 18.
[00:36:56] Host 1: Paul Barnhurst: 18. I have no idea which one this is. All right. This is a unique question. I haven't seen this one before. I'll read it exactly as it is, because it's kind of. It shows the word salad a little bit that I can give us. Which international snack keeps you sane during late night workshops?
[00:37:19] Host 2: Glenn Hopper: That's strange.
[00:37:20] Host 1: Paul Barnhurst: One.
[00:37:21] Guest: Tariq Munir: Yeah, but if there is one thing that keeps me up is the coffee. I mean, it keeps me focused. That keeps me. You know, I mean, that makes me who I am.
[00:37:30] Host 1: Paul Barnhurst: All right, so the answer would be coffee on that one. I like it, but I just thought it was really interesting that it said which international snack and it didn't say which kept me which kept you sane.
[00:37:42] Host 2: Glenn Hopper: It was on a section called International Experiences, so it must have been related to travel.
[00:37:47] Host 1: Paul Barnhurst: Yeah. It's not something on your profile that made it, you know, did. It's probabilistic there. But that was a unique one. All right, your turn, Glenn. So you never know what we're going to get with these questions.
[00:37:57] Host 2: Glenn Hopper: And we've used to edit them and fine tune them. And now it's like, no, let's just let the AI do. It's I think so since ChatGPT created this, I have just picked one for us. And I just picked number one. This one's actually pretty good. If I were a superhero sidekick, what would its name be and how would it help leaders save the day? I like that one, Paul.
[00:38:21] Guest: Tariq Munir: It'd be Captain America, right?
[00:38:24] Host 1: Paul Barnhurst: Yeah.
[00:38:25] Guest: Tariq Munir: All principles and everything. Hopefully then. Yeah. I mean, it'll help us save the day. Hopefully it doesn't make Loki.
[00:38:33] Host 2: Glenn Hopper: Actually, Captain America and coffee sounds like about a perfect day.
[00:38:37] Host 1: Paul Barnhurst: I thought you wanted Mission Impossible all day. Glenn. Yeah.
[00:38:42] Host 2: Glenn Hopper: That's right. That was what? We'll have to catch up on that. Yeah.
[00:38:45] Host 1: Paul Barnhurst: All right, well, Tariq, thank you so much for joining us. It's a real pleasure to chat with you. We'll let you go so we don't keep you too much longer. But thanks again. It was, like I said, a real pleasure. Love the work you're doing. Keep it up. Congratulations on starting your own business. I know it can be quite the grind. I mean, it's rewarding as well, but a lot of work. So keep building and congratulations on that and thanks for joining us. We appreciate your insights today.
[00:39:11] Guest: Tariq Munir: Thank you so much Glenn. Thank you so much, Paul. It was so much fun. Thank you for having me on the show.
[00:39:16] Host 2: Glenn Hopper: Thanks, Tariq.
[00:39:18] Host 1: Paul Barnhurst: 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.