Artificial Intelligence Podcast: ChatGPT, Claude, Midjourney and all other AI Tools

Demystifying AI with Paula Paul

• Jonathan Green : Artificial Intelligence Expert and Author of ChatGPT Profits • Episode 348

Share episode

Welcome to the Artificial Intelligence Podcast with Jonathan Green! In this episode, we're delving into the ever-evolving world of AI with our special guest, Paula Paul, an industry veteran with a wealth of experience in software engineering and AI consulting.

Paula offers an insightful perspective on the expansive and sometimes misunderstood realm of AI, drawing on her extensive background in computing. We explore how AI has transitioned from deterministic decision-making tools to probabilistic language models, transforming the way humans interact with technology. Paula shares how AI has evolved past the rigid syntax of early computing days to a more intuitive interface where imperfect input can still yield accurate results, enhancing accessibility for non-technical users.

Notable Quotes:

  • "The challenges are around the way people relate to technology and the change management involved." - [Paula Paul] 
  • "Every job that's replaced by AI will just create new opportunities." - [Jonathan Green] "Technology is really abundant and easy, and the challenges are around the way people relate to it." - [Paula Paul] 
  • "AI is really a very helpful tool, but it's wrong a lot and you still need someone there to go, wait a minute, that's the wrong answer." - [Jonathan Green]

Connect with Paula Paul:

LinkedIn: https://www.linkedin.com/in/paulapaul/ 

Website: https://www.greyshore.com/

Connect with Jonathan Green

Demystifying AI with today's special guest, Paula Paul. Welcome to the Artificial Intelligence Podcast, where we make AI simple, practical, and accessible for small business owners and leaders. Forget the complicated T talk or expensive consultants. This is where you'll learn how to implement AI strategies that are easy to understand and can make a big impact for your business. The Artificial Intelligence Podcast is brought to you by fraction, a IO, the trusted partner for AI Digital transformation. At fraction a IO, we help small and medium sized businesses boost revenue by eliminating time wasting non-revenue generating tasks that frustrate your team. With our custom AI bots, tools and automations, we make it easy to shift your team's focus to the task. That matter most. Driving growth and results, we guide you through a smooth, seamless transition to ai, ensuring you avoid policy mistakes and invest in the tools that truly deliver value. Don't get left behind. Let fraction aio help you. Stay ahead in today's AI driven world. Learn more. Get started. Fraction aio.com. Now, I'm really excited to have you here because I'm always looking for unique perspectives, and what I've found is that so many people aren't even sure what AI means anymore. The definition has changed so much. When I was a kid, it meant sentient robot. And now it almost feels like it just means smart word processor. And in fact, now most people just think of AI as a chat bot. It's something you chat to. Is there anything more to it? And really, where do you think things are going over the next few years? Oh, wow. That's a great question because I always go back to my own history. I love that you brought up all these examples because there were decision making tools with us for a long time, and then even going back further I thought we've been asking computers to generate things for us for a long time, and I came into computing in the age of early CAD systems. So my father I had a little drafting board and I did mechanical drafting. So there was a type of a drawing that you could do called an isometric view. So you draw the front view, the side view, the top view of a widget, and then you draw this kind of 45 degree angle perspective view. I showed my father that on a CAD system you could draw the front view, the side view, the top view, and then you could just point at it with a light pen. It's called a light pen, and it would generate the isometric view. And like my father looked at me like I was a witch, burn her. So we have been asking computers to generate things. For us for a long time, including c, GI, and you remember, like uncanny valley. And maybe what's a little different here is we're using more not deterministic. It's more probabilistic. Like what's the likely next word in the sentence kind of thing. So I don't get to, it all. It's just another way to ask computer to do things for us. And I think the challenges we're asking it, we don't know exactly. I feel like this phase of AI is really the first time when you can not know exactly what you're asking. You can misspell a word or not phrase it correctly, and it can figure out what you mean. So I feel like the real value is that the variance of what data could accept.'cause when I started out, if you capitalized a lowercase letter, it would go, file not found. That directory doesn't exist. There was no room for error. Perfect spelling, perfect fonts. You messed up a space, it didn't work. And now we've gotten to the point where you can use the wrong word or you can misspell almost all of the words. Now, I've always been someone who's so obsessed with spelling. Now I misspell a bunch of words in the question, and I know it'll still know what I mean. And that's. What I think is the most magical part is that you can be less technical and less specific in your explanation. You don't have to learn how to code anymore just to talk to it. That's why when people say learning chat, GT is like learning Python. I'm like, no, it's not. Python's really hard. That's a really hard example. That's one of the hardest things you could say to someone. So it's really. The beauty is that it has a wider variance.'cause the biggest problem with computers, the first thing I ever learned was garbage in, garbage out. If you give it the wrong data, you'll get the wrong response. Now it has the ability to go, wait a minute, are you sure that's what you meant? And get us close to what we're looking for. And I think that a lot of people now have this fear all the time. When I work on projects with clients, they say to me, Hey, help me with this, but don't automate it so much that you automate me out of a job. And there's this fear at every level that AI is ready to replace people. And it's so far away from that. It's really a very helpful tool, but it's wrong a lot and you still need someone there to go, wait a minute, that's the wrong answer. There's definitely something not right here. I. I totally agree that I, worked through a lot from Google and Gemini and one of my favorite courses and that was where someone did explain large language models and how to do prompt engineering. And that really the most important part of about all of that is that someone needs to be at the end of value chain, really validate that model, produce something. You intended or that has value and. When you have the computer doing things probabilistically, right? It's that's probably what you meant. Or here's probably something that would be useful to you. So the human at the end saying, yeah, that is valuable, or No, let's do let's feed some more information back into the model and come up with a better solution. And I think it's every technology, it just creates more jobs and more opportunities. Yeah, I always say that every job that's replaced by AI will just create new opportunities. If everyone's being replaced with robots, I'll be the robot repairman. There's always an opportunity as soon as you get past that initial fear and now people that can are comfortable with AI are becoming more valuable or getting raises. There's a lot of opportunities because every company knows they want ai, but almost no company knows what that means. Very, do two people mean the same thing when they say ai? I find that. There's been this definition expansion that we've seen with technology. It used to mean the person you call when your computer's broken. Now it has brought into me people that work at server farms never talk to anyone. It's frontend engineers, backend engineers are all considered part of it, and the definition's getting broader and broader. And I've seen it companies that now handle ai, that now handle cybersecurity. So there's been these mergings now it feels aI just means anything to do with a computer.'cause most of the clients, most people I talk to, they say, I say, gimme a list of all the things you wanna do. And 90% of it's not ai, it's automation, it's organization, it's basically the biggest problem most people have is in the area of data. Either their data is disorganized, they don't keep data, or they keep data and do nothing with it. It's just sitting on a shelf or they have too much data that no one can really find anything useful from it. And most of that is just organization and structure and building out a plan. And obviously the best time to organize your data is 20 years ago, but the second best time is today. Most people, even if they're not quite sure how to use the right words, want pretty similar results when they ask for your help with different projects in the AI field. Yes. And I think a lot of the people that I work with are just trying to sort out, where do I get started? And I am like, you probably already are. So in many meetings now I'll use an AI note taker that will do the summarization of the text or take the notes or capture the action items. And in most organizations, if you can get it through compliance and security which you can over, time, even in the most regulated industries, you can use an ai, what's called an AI notetaker, but really it's just large language model text summarization. And then from there you can say do you have a call center? Your call center representatives are probably going off of scripts and then, case notes and whatnot. And a large language model, text summarization can really help in that area as well. And you being, organ companies like Salesforce are really doing great work in that area with their model called I Einstein. So I, I think when you stop. Drawing the term AI around and you start talking about specific use cases that are really valuable to people that's where I really start to make headway. And then where it gets really kinda squishy sometimes is can you I get a lot of calls of, can you make my engineering team go faster? I usually answer with Yes, but it might not be the way you think. So I, I don't really believe that you're gonna replace all your engineers with a. Ai. But there are ways to use AI to make sure that maybe more test coverage is in place. So I think it's better to talk about the use cases than just say, I can get in. So a lot of AI and automation consultants and builders on the consulting side always have these conversations that say what do I charge for bot? What is a bot worth? What is a chat bot worth? What is this type of bot worth? And I try to explain that I don't come from that perspective. Like when you, that's really hard to say what is a piece of paper worth? How long is a line? These are all opinion based and subjective. And I say, I don't do that. When I approach a project client, I say, what's your problem? What's, and then you solve the problem that has a value. And it's really hard to say what's a medicine worth? The medicine is worth the disease of cures. So when you look at it from that side, it creates a lot more value and it's a lot easier to understand. When I work, whenever I approach a new client or anyone at one of my companies, I say, they always say a couple of the same things. Number one, please don't automate me out of a job. And number two, I don't know if this is possible, but, and I always try to explain. Don't worry about if it's possible, that's my job. If it's not possible, I'll tell you, but just dream big and then we'll see if AI can do it. And sometimes they dream towards things that don't exist yet, but 90% of the time it's doable. And I find something very interesting. The harder someone thinks something is, the easier it's to do. And the things that they think are really easy are always really hard. Do you ever encounter that? Yeah, one of my, for people who have worked with me on projects or at clients, they know that I always say the technology is the easy part because I do come, from an era. I started as a software engineer in the eighties at IBM and at that time, writing code was hard. It was batch files, it took hours and all the good stuff. So now to me, technology is at the wave of a hand or the press of a button. You could spin up a Kubernetes cluster or take advantage of a large language model. So technology is really abundant and easy, and the challenges are around the way people relate to it and the change management involved. When I started in the early days of cad, the I had to go and do the little dog and pony show of CAD systems to people who were on. And it was, don't automate me out of a job has always been the backstory of technology . So I think that it just changes jobs and it helps us become more productive if it's used in the right way. I do think that you need to. Look at the cost of technology in terms of how you're gonna measure the value. It's the, what's it worth question is what's it worth to introduce a large language model and text summarization into a call center? Let's measure how much how many more cases we can close in a day, or how long it takes to close a case. So metrics are always important in technology. And I think even more important when we start. Talking about emerging two mainstream technologies like ai. So you brought up something that I think is really important, and it's one of my pet peeves, that sometimes people or AI consultants or technology consultants say something like, when you've met this technology, you'll make your team more efficient, so you'll save this much money. And that's only if you fire one of them because you're still paying their salaries, right? You still have to, you have 10 employees, even if they're faster, your cost hasn't changed unless you fire someone. And there's this fear of consultants, right? That's oh, consultant comes in and fires half the staff. Does everyone work twice as hard? And then says, thanks. I see sometimes people get distracted by the wrong metric, and that's what I wanna talk about is that they measure the wrong thing. I'll make your team 10% more efficient. Great. What does that mean? So I once had a social media employee work where he said, we have a reach of 4 million. And I said, what's Reach? And he said That's how many people could. I've seen our social media posts and I go, how many people did see it? Because I had, like in high school, I had a lot of reach , but nobody actually went on any dates with me. So I know reach is imaginary. I want a real metric. So that's really like a critical component for me. And what are the important metrics to actually. Track that aren't imaginary, what are the ones that actually matter from that? Like when you approach a business, you know the ones that matter the most to them that you need to bring up, and the ones they don't actually care about. Exactly that is the question. Because if you start talking to, the IT executives, they're often used to be treated being treated as a cost center. So all they care about is doing what they do and doing more of it, but at a lower cost, and that's. I think a downward spiral and a losing battle at the end of the day. So you have to start at the top of the house and say, how does this organization generate value? Is it revenue? Is it, and I do work with a lot of nonprofits. So is it the number of communities serviced in the number of, in recent engagement as how many. People in the community have adopted lower cost solar power. So there are metrics that the company uses to establish its value, and you need to tie the metrics of a technology organization to those organizational value metrics. Cost is, certainly very important but I don't think it is the thing that we should be focused on. And then the call center example is solving more cases or closing more cases with higher customer satisfaction. The cost of that comes after you deliver the service that you're trying to do that delivers. Yeah, I found that for some of the clients I work with, speed is the most important factor. So there's, if we switch this technology, how long will it take people to learn? And how much faster can we hit our milestones? So if we have 10 goals we're trying to hit over the next 10 months, can we hit them in eight months? Can we hit them in six months? And it's really important to know. What language or what metrics matter, and that's a really powerful question, which is what's the most important metric to your business? Because sometimes people ask for something and if they get it, they're still not happy 'cause they asked for the wrong metric. I find that a huge part of my job is figuring out what they actually meant. And sometimes it's like when you go to the doctor and you say, this is the medicine I want. And they go, did you see a commercial? And it's the wrong medicine. You don't have, and now they have to convince you first that you're wrong and then of the right thing. And I find that it takes a bit of investigation. So a lot of my job I find is saying, where's the end point where things end right now? And where do you want them to get to? And that's the gap I have to bridge. Figure out what are the steps in that bridge before any technology comes into place. And I try to explain that if you already have a process, automating it is so much easier if you don't have a process. First we have to build one and then automate it because we first have to design something that works. And so there's these different components to it and. You have these two ends of the spectrum, which is where AI's really far away, I have no idea to do, and other people go, it's a magic button. Push the magic AI button and solve my problems. I was like, I wish. I wish I had that button. That sounds amazing. Business and even back to like assembly line kind of work that you can have premature optimizations, that if your value as an organization happens when the car rolls off the assembly line, if there are eight different stations along the way from, getting the chassis together to putting the wheels on and whatever. If you optimize just one part of that assembly line, you can make it go as fast as possible, but it's not gonna add any value at the end of the day because you have an entire line. That's the value stream, that, that contributes the value. So I do think there's a lot of. Premature optimization that people are chasing after, without knowing how it fits into the big picture of how they're actually delivering value as an organization. And ai, with the kind of sparkly thing approach is gonna be a lot of premature optimizations, I think. It reminds me of there's this niche called efficiency where people try to get more and more efficient and you spend so much time become more efficient that you never get anything done. It's like you spend so much time perfectly organizing your closet that you never use anything, and I feel that sometimes we forget that. There's a lot to a transition, which is that not only do you have to design a new process, you then have to get the team to actually implement it and then keep using it. And there's that time period of the learning and the dip. And I find that a lot of the things I build for clients, they never use it. It increases efficiency, but they don't have a plan for creating adoption amongst their team. They don't transition to the tool. They don't make the switch, and it's really hard to create that cultural shift, especially because there's this. Phrase, which is, but we've always done it this way and it's once something's been 5, 10, 20 years. And I've seen we even see this where there are like systems that, computer systems that are 40, 50 years old that are still being used like air traffic control, still uses like tube vacuum tubes. And a lot of the federal government uses stuff programmed in cobalt, which is a language I think from the late seventies or early eighties like so we're. We're so far back that there's only five people left to even know how to use that language because it's been no private company's used it in so long. So this, I think, is where technology hits the human, which is. It's great to have a tool, but if nobody uses it, it doesn't make a difference. We have, in my house, we have a stove, a range, and then we have an oven, and nobody uses the oven because they're afraid of it, . So it doesn't matter if you have the tool, if no one knows how to use it, because they've never had an oven before in my family, and now we have one. They go, I don't understand it, so I'm not gonna learn how to use it. I know how to use this. Yeah. Yeah. Then that's, yeah, it goes back to technology is the easy part. And even more so now we have such an abundance of compute and storage and just network. We have got starlink and now it's network is pervasive, so now everybody's struggling with what, we must be able to do something with all this. And it goes back to what is. A lot of times I'll start with trying to understand what the relationship is between the organization and technology, and that starts at the executive team. So if you've got really in any organization, if you've got anyone on the executive team who says, I'm not technical, that's a smell to me because, really in this world, unless you're really living off the grid, which is hard to do these days is non-technical because we all have phones, we all communicate, through sessions like this. So if an executive team relies on technology in any way, which nearly everyone does to deliver value to their constituents. Executive team needs to have a good relationship with technology. Like they can't be like your family in the, it's like I don't know what it is and I stay away from it and its a unhealthy. Relationship with technology, and if you don't have a good relationship with technology, you're not going to optimize your value because you depend on technology to get it. So it's really interesting, from my career perspective that I never met a technology I didn't like, that I didn't feel comfortable with. So I spend a lot of time trying to demystify or to improve the relationship that people have with technology so that they can get more from it, which is a very people oriented thing versus, how I studied in school. I'm in a very technical background, so it's just been interesting to see it flipped on me that way. Yeah. One of the challenges is that at first it's all about the hard skills. How well can you program, how well, how much can you get done a certain amount of day, how many? But then as you go up the ranks, it becomes about the soft skills. Do people wanna work with you? Can you lead a team? And it's very different. Most of the problems I deal with now, I'm not allowed to fix, like I have to assign it because I have so many, it's just so different. Leading versus doing. And. And once you move out of the coding phase, like you just can see it from a distance, but the principles stay the same, but you lose the ability to implement, but you need the big picture. And that's really the shift, which is knowing where everything's going, be able to get people excited about adopting a new technology, and show that it's not replacing them, that this is gonna help them. And I think this is the big challenge we face right now, is that we have, people adopting too fast and people adopting too slow. Some people are going all in not thinking about what they're doing with their data and what they're doing with security and that the other end spectrum of people that are so hesitant because they're worried about losing their job to ai, that they don't wanna give it a foothold. Yeah. Yeah. And people, and I mean it, it goes to the rise of agile delivery over 25, maybe 30 years ago. Now that, people forget that you can start by taking small steps and increment and learn, which is really all that agile is about, is take a small step to find the minimum business increments or minimum business value that you can do with something. Try. Lessons learned, mistakes made, but do something small. And then the other piece that I usually help people with is. Forgotten how to make good decisions and they struggle or, put it off for so long 'cause they can't make a decision, which is really in itself a decision. But even just making a decision in such a way that you can move something forward in a small step, leaving options open. The most important part of making a decision is making one that you understand. Whether or not you can change your mind later, because something's always gonna change. So those are all, you brought up soft skills, but I think of them as like strategy levers that people are not applying well in. The haze of ai, because there's a lot of unknowns and languages is well because people know what they mean anymore. Yeah, I think that you're exactly right that we're in this transitional phase and as what we're gonna find is in five years, AI becomes the assumption, right? We went through the nineties where you would see on job applications, like email skills a plus, and now you would never see that on a job listing. Do you know how to use email? Do you know how to use Microsoft Word? It. It's a plus to, it's a requirement, three to five years experience expected, and now it's so expected. It's not listed. If you showed up from a job and said, oh, I don't know how to use a computer. They would be so shocked. And I feel like it's gonna be that way with AI within a few years. So we really are in that adoption phase and I think that a lot of what you're saying makes a lot of sense. I think this has been really valuable for our audience who are . Leading their teams and figuring out how to transition in this AI period. If people wanna find out more about what you do, connect with you online and see some of the projects you're working with, where can they find out about you? Oh, sure. And I'd love to hear from people and thank you for the opportunity to say this has been great. People can connect with me on LinkedIn. It's linkedin.com. Wack in wack, Paul. So it's, I'm easy to find online. I also independently consult through my own LLC, which is called Grayshore . So Greyshore dot com, GREY. And I, always happy to communicate with people and help where I can. That's amazing. Thank you so much for being here today for another wonderful episode of the Artificial Intelligence podcast. Thanks. Thank you for listening to this week's episode of the Artificial Intelligence Podcast. Make sure to subscribe so you never miss another episode. We'll be back next Monday with more tips and strategies on how to leverage AI to grow your business and achieve better results. In the meantime, if you're curious about how AI can boost your business' revenue, head over to artificial intelligence pod.com/calculator. Use our AI revenue calculator to discover the potential impact AI can have on your bottom line. It's quick, easy, and might just change the way. Think about your bid. Business while you're there, catch up on past episodes. Leave a review and check out our socials.