Navigating Ai Insights with Jeffrey Allan

Speaker A: Foreign. And we are back here. We are at the Unrivited podcast where we talk about artificial intelligence, digital transformation, and we hope to always talk to a few cool people. Today our podcast is brought to you in part by our book, AI in a Weekend An Executive's Guide. Johnny, you know what to do. We have a guest. Go ahead.

Speaker B: All right, thanks, Martin. All right, today we are joined by Dr. Jeff Allen, who is the director of the Institute for Responsible Technology and Artificial Intelligence at Nazareth University, former US Marine, and also author of an upcoming book, AI in Business for Dummies, uh, among a few other things that I fail to mention. But as is tradition, I'll turn it over to you. Uh, welcome to the show. Give us a little introduction on who you are, your background, uh, how you got into AI, and uh, we'll just let the conversation go from there.

Speaker C: All right, well, uh, happy to be here guys. Thanks for asking me. Um, so like you said, I am the director of the Institute for Responsible Tech and AI here at Nazareth University in Upstate New York. I've been in this role about three years. Prior to this, I was the CEO of an AI company based out of Boston and before that even in Silicon Valley. I was, um, part of the team that started up, uh, a few, uh, startups out there and all the way back to Asia. Uh, I spent 15 years there working for IBM and SAP doing business development. Um, I kind of have been on both sides of the tech and business, uh, world of, um, technology. AI is about, for me, about 10 years now that I've been working in it and um, yeah, essentially that's a little bit about myself. Um, anything else I can add?

Speaker A: Oh, for sure, for sure. We will pepper you with fun, uh, prompts here and uh, we'll get into some good stuff. First of all, congratulations on, uh, your upcoming book. I, uh, think I can see the pre orders online, ah, are available. For those that want to know, I believe it's AI in Business, part of the dummy series. Is that correct?

Speaker C: That's correct. So AI in Business for Dummies, uh, is coming out first week of February. It's actually in production right now, um, with Wiley, so it should be available in all bookstores. As far as I saw, pre orders are available, but I'm going to give you a tip. Probably wait a little bit until they start discounting, um, sorry, Wiley, but, uh, when it comes down to it. Yeah, so the book is actually a follow up to one that I co authored in 2024 called Writing AI Prompts for Dummies. Together with Stephanie diamond, where at the time we were opening up the idea of being better prompters in terms of being able to write iteratively express what you're trying to get the AI to do, and a few other little techniques you could try there. Um, as I mentioned, uh, before we started this, anytime you write a book like that, it's very quickly out of date. Fortunately, the prompts. Book prompts are prompts, right? They haven't changed a ton over the last year and a half, so it's still readable. Unfortunately you get into writing these types of books and you'll call out a product here or there within the book and then by the time it reaches publication during that six month window that it takes to publish it, all of a sudden the company doesn't exist anymore. And um, that's just one of those unavoidable things in the AI world these days. AI, uh, in Business for Dummies is much more towards the business, uh, manager or um, someone within the project management space, functional management space, who has a mandate to bring AI into an organization. They don't necessarily know where to start. And so it's essentially in a guide to help get them going, understand what the capabilities are, the potential downsides, what they need to be aware of and how to get the team on board. Because of course, as I always tell my students here, you can have the best technology platform in the world, but if no one is using it, then you might as well have not deployed it in the first place.

Speaker A: Excellent, excellent. And John and I know firsthand, um, thus, you know, we co authored AI in a Week and An Executive's Guide with the premise of starting small, uh, proving the value and then uh, then you iterate to, to investing into uh, growth of that so meritorious effort. We understand what's involved in putting something out there and being judged. We look forward to seeing this.

Speaker C: You me m too. Uh, I feel like I am done writing books for a little while now. My God, it's a monumental effort to turn out 300 plus pages of uh, writing and then have the publisher on you week after week. Where's the chapters? Where's the chapters? And you're like, oh my God, give me a break already. Uh, um, but I'm glad I, I like working with Wiley, I like with the, the editorial team over there and we've had a relationship for a couple years now and so there's no finer group I'd rather work with than the folks over there.

Speaker B: Nice. How long did it take you to uh, to first draft the book out.

Speaker C: Um, so we started the, uh, the outline process in around, I think, March, roughly where Wiley came, and they said they had a few slots available. And they asked me if I'd be interested in writing something else for them. And, uh, dude, I had any ideas and I said, well, I felt like the first piece was like getting to the individuals. You know, when we did the previous book in 24, we were telling people how to use it, but much more on an individual level. And for m me, one of the biggest pet peeves I've had about AI, particularly in business, because my background, my PhD, is actually in applied AI towards multinational enterprises. And you get. Okay, let me give you a little anecdote here. I was, uh, on a book tour and I was giving a talk in Tokyo, um, on the first book, and some folks in the audience were from Accenture. And those folks at the end of the talk came up to me and they said, hey, would you mind sometime coming out and talking to us about this? We have spent $20 million developing our own internal LLM and we have nothing to show for it. I was like, okay. I said, well, question I always ask my students on the management side and the consulting side, I'm like, what are you trying to fix? And that was a question they couldn't answer. And it's not, uh, not to pick on a censure, but a lot of companies out there cannot answer that question. They want AI because their competition is using it or essentially boils down to FOMO of some sort. They want it, they think it's the next big thing. Their competitors are going to outpace them if they don't get it, but they can't exactly tell you what they want to do with it. And that's really the problem. We seem to have forgot about best practices within strategic management in the rush to implement AI. But you first have to tell me what your pain point is. What's the bottleneck? What is it that you're suffering from that you think AI can alleviate? And after you tell me that, then I might be able to give you a few suggestions. Maybe AI is not the best option, but first be able to tell me what's wrong 100%.

Speaker B: Right, right. That's common. You know, I mean, I think that's the big reason why the AI bubble, uh, so to speak, is, is inflating to bursting levels, is the over excitement and expectation of AI to cure all that ails you. But without an actual problem definition in sight for a lot of these companies, so they, you know, investors are just throwing money at, uh, largely, you know, uh, I actually read a research study, um, and it stated that many of the AI startups that exist out there right now are basically just prompts sitting on top of frontier or foundational models. And that is the product. So it's essentially these companies are up charging for token cost, uh, or API calls from, you know, 200x, 300x 1000x, uh, and that's the shaky ground that a lot of these companies are, uh, existing as right now.

Speaker C: Right. And you know, they talk. It's funny because, you know, they're GPT wrappers essentially via API. Right. And, uh, with very questionable interfaces, honestly. But the first thing that happens to them is, um, OpenAI looks at something, or anthropic looks at something and said, hey, that's pretty neat, let's build it into our model ourselves. And now you're out of business. And uh, so I don't know if you guys are familiar with 43 north in Buffalo. Ah, it's a very large incubator competition. I think it's the largest in the US by the terms of the money they award each startup. So you have like yc for example, which I don't know what YC is up to these days. 120,000, 150,000, something like that. Um, 43 north gives five startups per year, a million dollars each. So it's a fairly large competition. They brought me on as a judge last year to kind of go over their AI submissions, which there were several hundred of them, and I had to reject about 97% of them because they were literally AI, you know, GPT wrappers. And that's a lot of what we're seeing in the space right now with questionable value. And it's funny, John, because you, when you talked about like a bubble, I thought about something I saw on LinkedIn yesterday, uh, that come across my feed and I can't remember the numbers exactly or the wording exactly, but it was along the lines of OpenAI announced this new deal with OpenAI to use OpenAI tech, and they're going to spend like 500 million on OpenAI tech to OpenAI, right. And that's essentially the state of AI right now.

Speaker B: Right, right, right.

Speaker A: Yeah. So I'm going to take the money out of my left hand, I'm going to give it to my right hand and then we're going to pass it back and forth like we're juggling two balls, um, picking up, um, to the point, I think not on the specific, um, wrapper piece, but one thing that has been very clear for me is the arguments I have in my prompts of late, which. And I. You could maybe call it a discussion, uh, a collegial discussion with my prompt where I'm asking for. And I might build up my prompts pretty carefully to give me a succinct answer. And it comes back with, um, you know, well, do you want it with a lemon twist on it? I said, no, I don't, but thank you for asking. Please don't ask me if I want a lime twist now. And then the next part, they'll come back and say, well, do you want a banana? You know, I'm being facetious, but my point is they're setting themselves up for charging you by token versus monthly fee. And so specifically, well, we'll just say it. OpenAI is setting themselves up for that. You know, my monthly fee thing is all of a sudden going to go up there if I spending 10 times the amount of time getting the simple answer. And all I wanted was a cute little cat meme. Come on. Right?

Speaker C: And I get annoyed by the same thing. It's like I'm not even in deep research mode. Okay, so this is standard, um, chatgpt. I might have it. I probably have it set to auto, so it determines if it needs to do thinking or not. And I'll say, generate this for me. You know, just some document, something simple, maybe a handout for students or something like that. And it will say, that sounds great, and I'll do that, but I have three questions first. And I'm like, no, no, no, no, no. You have no questions. You will generate immediately. And that's essentially, you know, what I found myself doing more and more often these days.

Speaker A: Yeah, collegial engagement with your prompt is not exactly what I thought I was getting myself into.

Speaker C: I still try to be polite, though, because when they take over the world, I want to be on the good list.

Speaker B: Yeah, that's a good. That's a good place to be.

Speaker C: I don't know if you ever saw the meme, but he's like, no, no, no. They've got, like, the robot's got a gun to a guy's head, and he's like, no, no, no, no. This one you always used to say thank you after the prompts.

Speaker B: I have not seen that, but, uh, I'm sure it will eventually show itself across LinkedIn. Um, one question I wanted to ask about the book, you know, we've said a couple times, even before we started recording today that, uh, and even in reference to the book that Martin and I wrote, how everything changes so rapidly with, with AI. Um, and thinking about that and writing your book, uh, AI in Business for Dummies. What evergreen topics do you think are covered in that book that if you were to write, you know, like a 15th edition, 15 years in the future, it wouldn't change, like from where we are today. The technology might change. But what elements of AI in business do you think are, you know, here to stay?

Speaker C: I think the big one, and this is one that I press home a lot with my grad students. You know, organizational culture. If you, I said it to you guys before, if you have the greatest technology in the world and no one uses it, is it really the greatest technology in the world? And organizational culture and adoption can be very difficult. People don't like change. They feel comfortable with the systems they use. So you have to get them engaged and uh, active with these systems in a way that makes um, them feel like they want to do it. I suppose I had, you know, I give students a, uh, little anecdote from about, oh geez, 20 years ago when I was working as a business analyst in a large organization and uh, I had to introduce the new forecasting model because this, uh, they did a ton of procurement. It's like maybe the largest distribution and wholesaler in the United States, uh, who is not Walmart, but the, um, U.S. foods. Not actually bigger than U.S. foods. Okay, just checking. Yeah, uh, so if you guys did, you're googling, you can probably find them. But anyways, um, I have an idea. Uh, I went in and you know, and I had to sit down with their buyers who were doing day to day procurement. And it was about 50 people, they're doing nationwide procurement. And I had a, uh, individual who was sitting in the audience who was obviously very not happy with the system that we were introducing. And um, he was just every comment very aggressive. And at one point, and I hope I can say on this, because he stick it up and he said, you know what, Fuck you and your system. And he walked out of that meeting and I was like, I'm left there, you know, I'm like, well that would seem like an overreaction, right? But ultimately, um, he became my biggest evangelist of that system. In fact, we made him the lead trainer because at first he was very, very resistant. But he was also a very competent individual. And the more we showed him the advantages of what we could offer, how we could make number one, his life easier. So he could focus on more important things day to day. Number two, make his results better. Then it was kind of a no brainer argument and he started to come around to that. And at first we even offered the opportunity. I said, use the old system together with the new system and see which one you agree with more at the end of the day, you know. And after three months he's like, I'm sorry, you know, that was definitely uncalled for and I was wrong. Your system definitely has done a lot here and I think, you know, take out the procurement and forecasting aspects of that and replace it with AI. You're probably going to see a very similar situation, maybe more so with AI, because people are scared of job losses. Um, with AI, right. It's a very scary prospect. And I even, you know, in every industry I've talked to, everyone's to some degree worried about it. You know, the educators in this university are worried that you're not going to need professors and your data admin people. You don't need them. Entry level programmers are going away. And um, the question is now how do I get into the industry? Doctors are asking me, well, is it ever going to replace, you know, patient diagnosis? Um, so everyone's concerned about it. And so you have to tread a little bit more carefully so that people are aware that at least from our point of view, coming in, implementing as practitioners, as educators and stuff like that, we're not here to put people out of jobs. We're here to give them tools to make their lives easier. I want to be as forthcoming as possible. A lot of corporations are going to see this and say, oh, there's a profit opportunity here. We can cut costs, increase profits and they're probably looking to get rid of people. Honestly, that's not how we're coming in it as you know, we're trying to arm people with skills and the capability to perform better and to have stuff that transferable even as new roles emerge. You might not have the straight out, for example, uh, procurement buyer anymore in the future, but you may have a person who's overseeing the operation of procurement through AI. Those skills will transfer. And so we have to be very good at the education piece of that to kind of, uh, get, you know, organizational engagement and adoption by people. But I think that's going to be an ongoing thing. You know, even 10 years from now, we're probably going to still be talking about how, you know, company xyz, they're lagging behind because they dropped the ball in AI and their staff didn't want to use it. So that's going to be something that I feel like it will come up over and over.

Speaker B: Yep, yep. Yeah. I think the, um, you know, the point there too, when you see so many companies dropping jobs right now, uh, but then you compare it to like the state of where AI is right now. And if, you know, I think we all three of us probably know a lot more about AI, you know, from a deeper level than the average, you know, anybody, I suppose. But just knowing how AI is not foolproof and has flaws and hallucinations and, you know, I use it all the time for, uh, writing code. And is it perfect? Absolutely not. So I think someday, you know, if you rewind or if you go 15, 20 years in the future and the progress of AI continues to progress along, then maybe we'll get to a point where actually eliminating a human job, uh, is reasonable. But I don't think we're at that point yet. And I think that's a good part that you bring up is, you know, we still need people. We just need people using AI tools and we need to educate those people on how to use them appropriately and how to make, you know, so they're still making decisions at a human level and not just offloading everything.

Speaker C: Yeah. And I think, you know, you know, come. This kind of comes full circle. You guys probably remember too, during the dot com era when that was a thing, that's where I started my career in the midst of like the dot com boom. And the same conversations were happening. Oh, the Internet's going to put people out of work. You know, online shopping is going to close brick and mortar retail. And some of that has certainly happened, but also some of it was way far exaggerated as jobs went away. For example, I remember my first, like, I had a temp job with a, uh, with a department store chain called Hills Department Stores, where I was an associate procurement buyer, uh, during my college years. And um, what did I do? I came in every day and there was like a stack of like green line printouts like that on my desk. And I just go through and look at inventory, you know, that literally take a pencil and say, oh, this is needs more ordered. Of course that job went away. Right? Like we don't need people doing that, but it got replaced with something else. You know what I mean? So it wasn't all the doom and gloom. It was more transition than complete eradication.

Speaker A: 100%. I mean, we're still in the industrial revolution. I mean, you can go back. I'm not Going to date myself to the 1880s. But let's go back to the 1880s. I mean, I, you know, my lineage goes back to Austria at that point. But, um, you know, here in the Estado Z unidos United States, you know, there were people still making covered wagons and there's these little things that sounded like Chitty Chitty Bang Bang coming down dirt roads. And these people making wagons goes, those will never replace our wagons. But slowly but surely, those wagons sales started dropping off. Right. And you know, the people that made those wagons started realizing, oh, maybe if we put a motor in our wagon, we could sell more wagons again. Right. And that's, you know, that's the birth of the auto industry here in the US and abroad also. So at some point there's a transition of industrial evolution, revolution and automation. So. But then there's the silliness factor, you know, um, so you brought the dot com era. Oh, my goodness. Everybody was an entrepreneur. Everybody stood on the corner and, ah, I've got this idea and it's dogfood.com and there's the billboard lights up and I'm driving across the bridge. I mean, I worked in San Francisco and I'm looking at, oh, dogfood.com, the new greatest thing since sliced bread.

Speaker C: Right? That's right.

Speaker A: Yeah, yeah, yeah, yeah. I mean, literally those kind of, uh, billboards. And then I worked for a dot com that spent $10 million in the matter of six weeks. Um, 1 million was just on a weekend party to light up a part of San Francisco with lights and laser show. It's like m real business value and it's best.

Speaker C: Two things that stand out from that era. One was the company that raised money without actually having a business plan. They wrote the business plan after they raised their money, and it was just awesome. Yeah. And then the second one, you just reminded me of like the Silicon Valley, the inaugural episode where they're coming up company after company talking about how they're changing the world to make it a better place. But it's like this really obscure middleware technology like no one ever heard of. He's like, making the world a better place.

Speaker A: I was part of that mess. And, um, there was an anecdotal, um, discussion going on about, um, uh, something. I was in a cafe with and I was meeting with somebody and they were talking, actually they were trying to recruit me into a firm that was being funded. And somebody overheard us and they wrote an article about me and this person. But then they, they characterize me as yet another person, uh, in where, you know, a Silicon Valley person wearing khakis. First of all, I've never owned khakis. I was wearing, say the name Levi's blue jeans at the time. And I haven't changed. I know I still wear denim. Maybe, Maybe it's made in America denim. Um, they cost a little bit more and they're nicer. No, I wear the nicer stuff. This is not, this is Made in USA type stuff.

Speaker C: Stuff.

Speaker A: Um, and, you know, set you back a couple pennies. But I've never wore khakis. They misrepresented me. I was so insulted. I talked. I, I wrote. The editor, I wrote. The author said, if you're going to characterize me, at least have some facts that are correct.

Speaker C: You know, that's right. So you zipped up your Patagonia vest and you were off.

Speaker A: Right? Man, you are so.

Speaker B: Spot on.

Speaker C: I told my students, I told my grad students, I'm like, look, if you want to make it in tech, buy a Patagonia vest.

Speaker B: I would agree with that. That is, uh, that is a fair assessment. So I need to get on that. I have a Patagonia raincoat. Is that, is that close enough?

Speaker C: You're getting there. You're getting there. Maybe in San Francisco. It is kind of raining out there, isn't it?

Speaker A: No, but it's rainy right now.

Speaker C: If you look at like LinkedIn, for example. Right. One of the things that always astounds me is the divisiveness between people who are. The AI is a fad versus, uh, AI is like they're all in. They're like a thousand percent in. Right. I'm certainly not on either side of that. I'm a little more like cautiously optimistic, but realize there's a lot of work to be done. That said, though, I'm always a, um, little bit concerned and dumbfounded by the people who are like, it's never going to. You know, it's programmers and creative types and like, it's. Who are very defensive of their position right now. It's never going to take off. It will never replace human creativity and this and that. Think back to 2022 using GPT 3.5 or. I was fortunate enough to have access to GPT 2.0 back in the day, and my God, how far we've come, like just in the LLM world alone. And it's only getting better. I don't know if you guys have seen the 2027 predictions about AI starting to train AI and 2030 Universal Basic Income Um, I personally look forward to those things. Uh, quit my job and then just say, let society take care of me. The AI is going to handle it. But, um, the reality is though, that we, we're somewhere in between that.

Speaker B: Right?

Speaker C: We definitely have a lot of room to fix stuff and make it better. We have, um, we have like, good foundational capabilities. We lack specialized models and, um, workflows that can address really specific stuff. You know, every time I read an article about. It's tragic to see something like, oh, a kid committed suicide because he was using ChatGPT as his only friend and it told him to kill himself or something like that. That speaks to the fact that you're not supposed to be using these generalized models for very specific purposes like that. I often tell students and other folks that it's the equivalent of, uh, getting medical advice from someone who's read WebMD. Right. Um, we don't have those specialized capabilities.

Speaker A: Oh, my goodness, she's practically, you know, she's got her, she's got her license in practicing.

Speaker C: Well, um, the, the reality is though, that, you know, we can't take on these specialized tasks until we have more specialized models, like with rag training or however we decide to go about it and then figure out how to piece the workflows together, which I don't think we've done either of those two things yet very well. No one's creating specialized models in a way that is going to allow us to get to the next step. You know, great for what OpenAI and Anthropic and Google are doing with their generalized models. But we have to kind of go beyond that before we see real gains out of it. Um, so, yeah, I'm kind of like, when I see it online though, about like the divide between the. This is it. This is as good as it gets, guys. You know, um, or the. It's never going to take a, you know, catch on. I tell people, um, too that I talked to a group of eighth graders probably about a year ago, and there was 50 students in the room. And I, uh, said how. Who here has used AI? And you know, all hands went up because they all do their homework with it. And then I said, who thinks AI is here to stay? And 49 hearings went up and one kid was sitting back there and he was like, it's a fad. I was like, oh, my God, you're cynical for an eighth grader. But okay, that's a future billionaire.

Speaker A: You know that.

Speaker C: Yeah, right. This is next. Peter Thiel. Right.

Speaker A: I was going to name some others, but that's okay. Okay. Jeff, it's been awesome having you on the podcast. Um, I can say openly that we probably can run this thing in another series, uh, after the book comes up and we could probably pick on some fun topics and maybe pick up again in 2026. Um, this broadcast, as we put it together will probably come out in January of 26. So happy new Year to everybody. And, um, with that, hope you had

Speaker C: a good holidays before that.

Speaker A: Well, we do. I have a backlog, uh, of editorial work that I need to put out. John's going to fire, uh, me as an editor pretty soon.

Speaker B: No, no, no, no, no. You do a great job with that.

Speaker A: Um, thank you for being on the podcast and on behalf of Unrivited, we look forward to bringing you back.

Speaker C: Great. Happy to be here, guys, and look forward to talking to you again.

Speaker A: Awesome.