AI, In Layman's Terms

Rachel Cossar: Hello, everyone, and welcome to another episode of conversations in the future of work. Today, we’re gonna take a deep dive into AI, not just as it is today, but some of the journey that it’s taken and some individuals who are experts in this space have taken their discovery of it. And I thought there would be no better person to welcome on the show than Mark Kelly. Mark Kelly, welcome. Thank you so much for having me on the show. 

Mark Kelly: Delighted to be here. I am calling or speaking to you from Dublin, Ireland now. So we’re we’re at probably about just coming up to twenty to three. And you’re based, Rachel, in New York. Is that right? Boston. Close. Boston. Boston. I’ve spent quite a bit of time in Boston too, so delighted to be on the show. Nice. Awesome. Yeah. 

Rachel Cossar: We’re, really excited about this one. And, I mean, I’d love to hear just for the audience, if you don’t mind giving a little bit of background just in terms of, you know, things you’ve done in the past, things you’re working on now. Yeah. 

Mark Kelly: So I’m I’m seventeen years in the world of technology staffing, and I would say god. What twenty twenty two years in the world of business. I’ve got, you know, four years business degree. Got a master’s from strategic management planning for for, Smurfah Business School. But I was always interested in applied and practical applications and everything that that I’ve done. So I started off working in kind of retail, so working in Brown Thomas, which was kinda like it’s kinda similar to, like, Nordstrom or or Harrods when I was in Arliss, that’s what you call it. So I really got a good understanding of customer service there. Then I went on to enterprise rent a car where I did a graduate program, and I got to learn how to actually run a 250 car fleece, how to sell, upsell, how to make sure that customer service is past 95%. Because in the world of enterprise rent a car, you could be making loads of money for the branch or you could be upselling loads of cars, and your insurance could be really low in terms of your premiums, in terms of damage. But unless you got 95% where people would refer you to a friend, you didn’t get a dime. Right? So when that is fixated into your mindset, you go well above what the customer wants. So that’s been ingrained with me from the last kind of twenty years in everything that I do and everything that we’re looking to do. So I spent eleven years with a staffing firm, multinational staffing firm, whereas very, very, very, very effective recruiter, then became a manager in predominantly in tech and then diversified in life sciences. But I always wanted to do my own thing. I always wanted to kinda get involved in business and actually put some skin in the game. And what I was seeing more and more when I was attending Gartner events, you know, good god. What was this like twelve years ago? I was seeing that AI and cybersecurity becoming more and more prevalent. And I said, yeah, I really have a feeling about this. I think this AI team is gonna take off. So what I did was I said before I’m gonna jump ship, I’m gonna go and meet people and ask them a simple question. Is AI hyped? Is it not? And I did 84 interviews in Ireland with people from academia and industry and saying, is AI hyped? Is it not? Should I set up a staffing firm? I I’ve got kids in a mortgage. What should I do? You know? And they were, like, thinking, yeah. I think it I think it’s something there. And I handed my notice, cofounded the staffing firm called Aldus, and we look to kinda grow across Europe and US. What we want to do different though from regular staffing was we wanted to get build relationships with customers very, very quickly, and we wanted to try to get social proof within the marketplace that our brand could be trusted. And I felt that the simplest way, this is probably, like, my only million dollar idea Mhmm. Was to interview leaders in the field of AI, asking them how are they using AI in action, what’s the impact. So we create this AI in action podcast. And then the second was AI mentors, which was how do you go build a data science team, but generate return on investment from us, and how do you actually scale that? And they’re they’re difficult things to actually do, and we created the podcast series. And coming up to, I think, this month, we’ve already got close to a thousand episodes. So we just kept it running, scaling us, and then kinda got it out into more of the ServiceNow field. But what that enabled me to do was really understand the problems that our customers are facing day to day. Mhmm. But also enabled me to understand that if we spent a lot of time with our customers figuring out their the culture, what they do every day, they’re gonna be more inclined to actually wanna work with us from a staffing perspective because we took that time. Locally, with all the learnings I had from those 84 interviews, I said, why don’t they create enough for profit and make these interviews accessible to people in Ireland? And I created AI Ireland. We took 50 of those interviews, made it available for the podcast series, and then we’ve been writing blogs ever since and putting them out there. Our goal is to help a million people become literate AI, and then we created the award ceremony. And the award ceremony is going for the last seven years, and we showcased over 650 applied applications of AI in the island of Ireland. So we’re trying to learn and we grow and engage with people through conversations that matter to them. But, fundamentally, probably what’s helped us most is we meet people where they’re at. So either on the staffing perspective on all this where we get to understand where they’re at from a technical journey, what type of skills they need. And then from the AI audit perspective, we’re just trying to give literacy to people just so they’ve got, like, a bit of a point of view if they should use technology, if they should not, and they got more of a balanced overview. So that’s a little bit about myself. That’s awesome. Thank you for sharing. 

Rachel Cossar: Very impressive, background and Thank you. Longevity with technology and AI specifically. So let’s let’s dig into some of those things that you mentioned. Right? So I feel like for a lot of people, AI didn’t exist, and then all of a sudden, one day ChatGPT came online. Right? 

Mark Kelly: But, of course, like, that’s that’s that’s certainly not the case. 

Rachel Cossar: So can you share more about, like, your early days of discovering AI? I know you mentioned a little bit briefly in your introduction, but, like, what were you seeing in the AI space? What did it look like? And what made you think, actually, I think this isn’t just hype. I think there’s a real, like, revolution waiting to happen here. 

Mark Kelly: It was funny because you’re around AI since 02/2017, 02/2018, like myself, you kind of were were like if you you know, that kinda everything’s a nail if you’ve got a a kinda hammer. So I was kinda thinking AI was kind of everywhere, anywhere because I was kinda biased because all my journey was AI. Right. But when Chativity came to town, it put an interface to a technology that Google already had, the transformer, and they gave Chativity where people could actually start to use this and get instant feedback, good, you know, poetry, images, whatever it might be. And then within five days, a million people were using us, 100,000,000 people in two months. I was sitting with, CFO of OpenAI two weeks ago, Sarah Friar. She told me a billion users are using, like, Chachibuty, and 85% of those users are actually outside of America. Like, Germany is one of the biggest markets, which I was kinda shocked about. Right. So suddenly, people had access to this technology. But before that, we had recommendation engines like Netflix, right, or card or making a purchase on Amazon or Zalando, where if you were to get the recommendation right by, like, two or 3%, that could be like a huge significant shift for that business profitability because you’re you’re getting so many different users now opting in to either make an additional per purchase or or Netflix spend more time on the platform. But there’s also, like, computer vision, where computer vision if anyone sees any of the autonomous vehicles like Waymo, where we’ve got, like, 750,000 drives happening a week out of going in the in the car, that’s all done by these kind of vision systems that can see someone walking across the road, a dog, cat, pigeon, and actually stop and kinda manage through all that different area. So they started to become really, really prevalent back, in 02/2018, 02/2019, 02/2020. And then you’d obviously have engaged with Alexa, natural language, voice, and all that kind of stuff too. But it was only when ChatJubit and generative AI came about that that was significant. Because, interestingly enough, AI investment was actually dropping, but generative AI came straight up that it really kinda get the line there. And if anyone’s watching this and they don’t know generative AI, it it’s kinda tight. We’ll probably give it away, but it’s creating images, creating text, creating video, and this is that kinda multimodal experience. But people wanna know, how do I use it to be more productive? How do I actually use it Yeah. For my business? How do I how do I upscale myself? And that’s probably my passion area is the applied stuff Right. To actually make people more literate in what they do. Right. Absolutely. Yeah. I mean and we’re we’re, like, big big on that too. 

Rachel Cossar: You know, you mentioned computer vision, virtual sapiens. That’s where we started with. And it’s interesting because when we started, we were so focused on the computer vision, behavioral aspects, nonverbal aspects of communication, and that’s what our AI did was provide people with feedback on that. And then when GenAI came on board and and was available via APIs, like, it made something that would have taken years probably to develop well available with, like, a half a day of work. Which is crazy. Which is just crazy. 

Mark Kelly: I get mad. Like Yeah. If people say, oh, AI, it’s it’s overhyped. It’s overhyped. I think they’re referring to GenAI. And from a market value perspective, I totally get why you’d say AI is overhyped. But from a business value perspective, particularly for small to medium enterprises Yeah. Like, it is totally underhyped the value that you can get specifically to your use cases if you double down on it. And I just don’t I think we’re actually just scratching the surface. Yep. Yep. I would agree with that. 

Rachel Cossar: But at the same time, I do feel like there’s a little bit of, like, a a contrast between the opportunity that does exist and this, like, global concern peep people have that we can get into later, about AI, like, replacing humans and taking over the world and, like, all that. And I don’t know. I I like, working with AI so closely every day, I’m like, I think we’re kind of we’re just, like, really not like, we have in order to for we use AI role plays. Right? So AI will be there as your conversation partner. No problem. And it’s amazing. But, like, the amount of guardrails and the amount of things that we have to tell the AI from a prompting perspective that most humans would just know, like Yeah. Or have been taught over, like, their entire lives of social interactions with other humans. Like, you have to you have to tell AI to do that. So in my mind, I’m like, AI is amazing, and it is going to completely transform the way we behave and act as humans. But it still requires so much human in the loop. And I don’t I don’t I I’m not really a believer of, like, AI is gonna totally replace humans and take over the world Yeah. Personally. 

Mark Kelly: But Yeah. Maybe twenty five years from now, but, hopefully, I’ve had a lot of systems in place to prevent that. 

Rachel Cossar: So it’s I got to see this firsthand. 

Mark Kelly: So for episode 200 of the AI Ireland podcast, I had Chativity, Gemini, and Copilot interview me. Okay. I didn’t get to use Claude because it didn’t have his voice didn’t find his voice before it, but I can do it now. So I I should try it out. And for the most part, it was okay if asking the question. So what I did was I said, okay. I want you to research everything you know about the ARRAN podcast, all the guests, all the episodes. I come up with 10 questions Mhmm. That you think it was suitable for. It came up with incredible questions. The questions were so good. I had to actually take some time to to to write up my answers because they were so Yeah. So good. But then when we actually did this, the wheels start to come off. So they would go through the questions Yeah. And answer. 

Rachel Cossar: And then suddenly there’d be the silence. 

Mark Kelly: I’m like, what’s going on? Yeah. Why don’t you ask me to ask and it would stop. So I haven’t released the bloopers by short, where it’s me just, like, trying to fix this and be, like, teaming over going, why are you asking that question again? Or why are you pausing? You know who I am, but are you referring to me back as do I wanna ask you a follow-up? So it was really kinda clunky. But for the for the twenty minute episode or so, it looked really, really seamless Yeah. That it can actually play out. So the more you spend time with these models, you can see Yeah. The kinks in them. Yep. And, also, in some cases, they’re actually regressing with the answers they give you. Right? And I I don’t and I don’t think they actually know the answer to why that is. But this is where I from the people losing their jobs perspective, you still need to understand when it’s wrong. So that means you still need to understand the process in terms of what you’re doing. Right? So Absolutely. So the example I’m giving is, you know, there’s certain consultancies out there at the moment where their graduates are just all in on Chativity CoPilot. They’re literally doing everything with us. But, consequently, when they when there’s an error, they can’t actually pinpoint when there’s an error. 

Rachel Cossar: So there’s there’s paper over those cracks really, really fundamental as an issue. 

Mark Kelly: And then the the consultants, more veterans will say, hey, guys. Let’s get off get off this now. Let’s teach you how to use this. But we’re not expecting you to be a twenty year vest unless you can actually spot where it’s telling a lie or that doesn’t sound right or or or the aspects. So you still need to know the process. Now from a school perspective, you I think you still need to learn from an early day early age to start to use the prompting if and if and when to start to use it. I saw you probably saw it as well, the ChattyBitty for your brain. Does it make you dumb? Does it does it does it not all this kind of analysis that was shared out. My kinda simple take on this is if you’re gonna get Chatopati to do every single thing for you, yes, your your brain is not gonna be in the best possible place. Yeah. But if you’re using this as an augmentation tool and and you’re actually genuinely working with it, I don’t see any issue for it. But like everything, you you have to be careful of the extremes. Yeah. For sure. 

Rachel Cossar: I mean, I think that study, apparently, was done with, like, nine people, and, like, I don’t know if it’s So so then these spurious correlations you have to be cognizant for as well. Yeah. So in talk more about your your book and how you talk about AI and, like, part of your work is is, like you said, literacy and education in AI and getting more people to be able to understand how it works and understand its limitations and not be terrified of it and know how to use it without, like, totally atrophying. So can you share more about your values around this, you know, literacy and AI and education? So the easiest way I’d say is I think we need to meet people where they’re at. 

Mark Kelly: Right? So so from reading an awful lot of books and then writing books, I kinda quickly understood that most people don’t read books, and they don’t necessarily finish off the books necessarily anymore. So if you’re gonna grab someone’s attention with a book about AI, for example, you have to have practical use cases that someone’s gonna resonate to. And that that’s particularly how I learned is practical use cases. So I always talk about my mom, 75 65, and my dad’s 71. They are my guinea pigs for everything that I that I do. So with my mom, she uses Chativity to help her create a itinerary for a tour guide business. Literally polishes it up, makes it more accessible, and she could turn it into any different language she wants, or she uses tools like Canva to turn into little brochures or a little kind of social media copy. Now for her, that’s pretty much where she needs to be. Now one of their companies, GetYourGuide, reached out to her and said, listen. We’d love you to be able to do a tour for us, but we need you to have a website to be able to actually as a reference point. So what we did was we jumped on to, like, a Lovable, which is like a tool similar to to Replit, where we just created a design, created a website, and then we showed my dad. And my dad was like, what’s my friend Peter gonna do? Mhmm. And I was like, well, who’s Peter? Peter’s Peter’s a web designer. Mhmm. He’s been doing this thirty years. It’s like, what? Lovable’s gonna take his job. And I was like, well, I think Peter’s got two options. Right? He’s got he can use Lovable and do really responsive work and start creating things that he wants to create and have a better architectural experience, or he can compete against Lovable. And I don’t necessarily think that’s competing against Lovable is the place to be. And he said, my only issue is I don’t think he’s ever gonna know about Lovable, which brings me to the point of, well, he’s gonna be indirectly competing against them. Yeah. And if he’s gonna be indirectly competing against it because he’s because he doesn’t have literacy, that’s a real issue. Right. And what I wanna say to people is, why don’t you just make people aware of these sort of products, these sort of tools? Yep. And then if you wanna use them, amazing. And if you don’t wanna use them, that’s your choice. But at least you’re cognizant Yeah. That you might review that and go back in the next year or so and go, actually, I might try to do that now because something’s changed within my business. But my biggest worry is people will never know. And this is where the job loss, the fear, and they then they get isolated. They’re not going on the journey, and then it becomes it’s all the technology’s fault. But I just think it’s more of literacy perspective as well. Right. Yeah. No. 

Rachel Cossar: I think that makes a ton of sense because, obviously, like, with with knowledge comes power and, certainly, a lot more, optionality in terms of the decisions you make and how you interact with AI or not. You know? But I I I think that’s a really wonderful mission. In in all these conversations you’ve had with with leaders about trans like, technology, AI transformation, Are there any examples of either AI transformations or just transformations that have, you know, brought in these new technologies that have been done well? 

Mark Kelly: Jared, there’s there’s one company that just really comes to mind. Is company called Hippocratic AI. I don’t know if you’ve ever seen them or or heard of them, but I think they’ve raised, like, a 155, $160,000,000 recently to get that series c, series d. But what they’re looking to do is and what they’re doing is giving you, like, an extension of your doctor via, like, an avatar. So let’s say it’s 07:00 at night, and you’ve got your antibiotics, but you’ve also noted your own meds, and you call up your doctor to say, hey. I totally forgot to let you know, but I’m on these meds. When you give me antibiotics, what’s what’s gonna happen? Right? And in the example, the digital avatar goes through a response as an extension of the doctor based on all the ladies’ medical history, and they could actually give that real time conversation to do. Now that lady can then call up the next day and speak to doctor whatever he want, but they’re so confident of the responses of these avatars as he’s accented of these trusted, surgeries or hospitals. I just feel that is the future of health care. Like, that to me is the future of health care. Like, they like, 1% of hospitals make money in The US. Right? I’m guessing it’s probably similar cost across Europe where health care is completely, it’s it’s broken. Right? So you need to try to reimagine that. And for me, that will transformational protect I think that’s lovely. The other aspect is if you think about how hospitals are becoming smarter about workflows. So one one of the biggest challenges you have in Ireland, and I’m guessing it’s probably across the world, is no shows at appointments. And it’s so difficult for people to get an appointment, first of all. And then when there’s a no show, you can’t even get someone else in because and it’s complete waste of time because the practitioners don’t even get to do what they wanna do. So simple ways to overcome that is to have an avatar call up and say, hi, mister Kelly. Your appointment is at 06:00. I just want to confirm, a, you’re still attending, b, will you need parking, c, can you pay the bill, now rather than, before? So it means we’ve got a timely, finished view, whatever it might be. So that’s one way that you could alleviate it as well. But another way is let’s say you come out of your your, appointment with your doctor, and you can actually record the conversation that your doctor gave to yourself, and that’s shared with you with an m p three. So think about it like notebook l m where you’ve got these kind of take these your transcripts from, video video blogs, or it could be it could be a PDF document at Stanford port. You can get it to turn into this podcast style where these two avatars talk feature. Well, these two avatars will talk about what happened at your appointment. And when you’re driving home, the avatars can tell you in real time what happened to the appointment just in case you forgot something or didn’t actually have the insights. Mhmm. But let’s say you’ve got a loved one who wants to know what happened at your appointment, but you don’t wanna talk. So my dad, got on grand. That was fine. Could you just share me with the MP3 what actually happened? Totally. Wow. Turns out you’re actually not fine. You have to do this, this, and this, and you also have to take this medication. Yeah. Whatever. That, again, is just a wonderful way to reimagine the health care process, start to use these tools in simple practical ways. So those types of transformations get me really, really excited. 

Rachel Cossar: And then you’ve got drug discovery where you’ve got the Moderna vaccine created in forty eight hours, one of the top vaccines. 

Mark Kelly: So they’re at dinner OpenAI house for literally pretty much every different division. So I’m always passionate about how they’ve moved the dial across people, process, the product, having a safe place to play and using it, and then leveraging their proprietary data. Right? So those five p’s Mhmm. Allow incredible transformations, but you have to underpin them with a perspective. Right? What is your perspective? And my feeling is people don’t get to have a perspective on AI or any of these tools because they’re so busy going down the hamster wheel that they actually need to have some think time, get off the hamster wheel, get into the sandbox, play with the tools, and then actually start to reimagine what the org structure needs to look like, what the job description needs to look like, what their customer experience needs to look like. But because we’re all tunneling, I have to get through to today. I have to get through to the next task. Yep. They’re not allowed to do that. Totally. Yeah. 

Rachel Cossar: I mean, we we I think especially for smaller companies, it could I I think smaller companies and larger companies have different issues. Right? But at the end of the day, it kinda comes down to that that same ability to have the space to look at things more broadly and think about some of the ways that transformation could take place, and then thinking about some of the processes to put that transformation in place Yeah. In general. Right? So I we’ve talked a lot about some of the things that you’re really excited about when it comes to AI. So what what makes you nervous about AI? 

Mark Kelly: I think the first thing is that people don’t necessarily know to use it correctly. Right? So that’s that’s the first thing. So when you think about Harvard Business Report came out recently, and it just blew my mind, the number one peep way people are using AI was as a therapist. Right? Right. So as a therapist, which I was just like, that is incredible. And then the second was organizing your life. So if you think about all those horrible emails, you guess, or those messages with your your partner that you don’t have to reply to, you’re you’re sticking into an AI to get you how to figure that out. Jeweler aspect is, like, character AI, Second fastest growing company of all time, and people are spending two to six hours with their perfect avatar having deep meaningful relationships to them. So so you know that boyfriend asked open end questions listens or that class bully that was really mean to you that you can go back. Yeah. Or that manager that doesn’t quite get you, but you can go and have role play discussions. And I would fear that, you know, one of the most searched terms still is I’m lonely. How do I make friends? That we’re we’re going to start to have these AI friendships for our young youngsters coming through who don’t wanna have more difficult, messy friendships and conversations because the AI gets you. Right? And, you spend time with people that get you. So if you’re gonna spend more time with these avatars that are just gonna be trained to keep you happy and engaged with you Mhmm. What does that mean for society? Totally. Yeah. That’s that that’s what kinda freaks me out. That’s what I kinda get worried about. Yeah. That’s so funny. 

Rachel Cossar: I’ve I’ve like, was it Replica AI? That company and some of the relation it it’s funny because I I do know some people who have these, like, AI buddies. And there are some applications or situations where you’re like, oh, that’s, you know, that’s that’s actually great. It’s great that that person is able to practice, like, social interactions and socialize with, like, a zero consequence buddy. Right? But, yeah, you’re you’re you’re right that there’s obviously a little bit. I think it depends on, you know, that person’s age too, you know, and, like, are they in that formative? Like, are they now going to expect that their, like, human girlfriend or boyfriend or significant other is going to just appease them and be like, you’re right. I love you. There you go. Right? Kinda crazy. 

Mark Kelly: Because our friendships then. Right? 

Rachel Cossar: So it’s kinda like we start to get more polarized. 

Mark Kelly: The algorithm just starts to show us what we wanted to see because we keep clicking in it. Yeah. So my updates are gonna be completely different than your updates. So it’s kinda like Netflix, but, like, 50 x. Right? Right. So that is an issue that I’m that I’m really, really concerned about. And then we also have the kind of the issue where we’ve now got humanoid robots coming in, and I think they’re gonna be wonderful for all age, homes and kind of companion robots, put really, really wonderful applications for that. But, again, what happens when we got one or two, kind of bad errors from kind of a hardware updates? What happens to when a billion robots kinda go rogue? Right? So so they would be more kind of practical applications that I would, figure out. But Yeah. If you’ve ever dealt with a really, really good AI that hallucinates, it’s like they are proper psychopaths. Totally. It’s crazy. Like, everything they’re saying has so much conviction, nuance, data behind us that you’re almost second guessing yourself to go, it is red. Right? It is red. Now tell me it’s blue green. Like, it’s just that scary. Yep. Totally. Yeah. 

Rachel Cossar: That’s I mean, it’s, it’s interesting. I think, though, that, like, the work that you’re doing around developing new awareness and understanding the boundaries and limitations of AI is gonna be part of what helps people, avoid some of those issues. Yeah. I think I’m in a balanced perspective. Yeah. You’re just saying to people, hey. This is the pros. Yeah. 

Mark Kelly: Here’s the cons. Yeah. If you’re gonna introduce this to your business, have an assumption log to say, hey. Here’s all my value or ethics on this. How did they resonate against your brand or business? Okay. Well, we’re off a bit. So we can go on this journey together and fix the corporate common source of the truth, or we’re gonna be way off. So I think it’s, yeah, it’s it’s multifaceted, kinda like life, though. Right? Yep. Absolutely. 

Rachel Cossar: Mark, thank you so much for joining today and sharing your perspective. It’s really fascinating to hear all the different ways that you’re, you know, coming at this, the future, future of work, future of companies, future of leadership. How can people get in touch with you or follow your work? Firstly, thank you so much for for helping today. 

Mark Kelly: That got well, like like, couple of minutes, so so it was really, really good fun. I think the easiest thing is just to probably follow me on LinkedIn. I usually am just throwing out ideas left, right, and center on that, and I try to share some of the content piece that we have. I think that’s probably the easiest way to get in touch. And, hopefully, if they’ve got anything they wanna resonate or talk about, I’ll be very, very happy to to share some thoughts. Awesome. Great. Well, thank you so much. 

Rachel Cossar: And as always, thanks to our audience. We’ll see you next time.