Latest / Technology Now / Vibe coding, bad data, and the fourth industrial revolution: the unstoppable advance of AI in 2026 | Bob Friday
Transcript
- MICHAEL BIRD We are starting off today's episode with a riddle SAM JARRELL at 8:00 AM in the morning. What is the riddle, Michael? MICHAEL BIRD All right. Why is artificial intelligence like winemaking? SAM JARRELL Because both wine and AI improve over time? Is there a correct answer? MICHAEL BIRD Yeah, there is a correct answer to the riddle, but you will have to stay tuned, as will listeners, to later in the episode to find out what the answer is, because this week we are asking the question: What does AI look like in 2026? I’m Michael Bird SAM JARRELL I'm Sam Jarrell And welcome to Technology Now from HPE THEME MUSIC (added in edit) SAM JARRELL Not to get too philosophical early in the episode, but Michael, if you had to explain AI to someone who had never come across it before, what would you say? MICHAEL BIRD Sam, I'm abysmal with analogies. I think I would say it's a computer trying to emulate the functions of a human brain. SAM JARRELL I agree. But what AI looks like is changing rapidly. We've gone from large language models and chatbots to agentic AI and things like self-driving networks in just a couple of years MICHAEL BIRD Yeah, it feels like it. and that's exactly who I wanted to talk to this week's guest, Bob Friday. Now Bob is the chief AI officer of HPE Networking, SAM JARRELL Oh, well he sounds pretty perfect for what we wanna discuss today then MICHAEL BIRD He does indeed. Now, before we get onto where AI is now, I wanted to look backwards a little bit and take a moment to find out from Bob how he got to where he is today BOB FRIDAY I was trying to convince a very large retailer that they should put this Wi-Fi mobile experience app onto their customer's phone, and I can remember the CIO telling me, "Bob, we're not putting any business critical app onto this Wi-Fi network until you can promise me that, this customer's gonna have a great mobile internet experience." And that's when I realized there's a fundamental paradigm shift from just managing network elements to actually managing this client to cloud user experience. So we still have to keep net AP switches and routers green, but, the question is how do you make sure that, they're gonna have a great mobile internet experience? MICHAEL BIRD So Mist came out of that? BOB FRIDAY That's when, Sujai and I started Mist. And the bet was that this real-time, day two cloud AIOps solution really required a new microservices cloud architecture, and That was the thesis of Mist, that we were gonna change, bring cloud AIOps into the networking space MICHAEL BIRD So let's sort of help to set the scene a little bit. what does AI mean, and what does AI look like in 2026? BOB FRIDAY Yeah. So the first thing I always tell people that the AI is not like a model or an algorithm. AI is the concept of doing something on par with a human that takes cognitive reasoning skills. so AI is kind of in the eye of the beholder, is that AI or just fancy automation? I would tell you an interesting story. The other day I called up the hotel to get my invoice for my expense report. Lady answered the phone, real nice, told me that, she'll text me the number that, you know, give me a call. Called the number. They couldn't help me. So I called her back and said, "Hey, you know, that, that didn't help me." I started to realize, I said, are you a real person?" She said, "Don't I sound like a real person?" And then I started to get... I could realize it , you know. But the point was, it's gotten so good it actually faked me out, you know, for that hotel. So we're almost to the point of Turing test type of thing, right? So that is my definition of AI, is, where we start to build things that do things on par with humans. Self-driving cars, help desks that actually you can't tell the difference. that's kinda my definition of where we're gonna start. In the networking space, same thing, In networking space, do we get self-driving networking to the point where there is no more support tickets, right? MICHAEL BIRD and can you talk through the sort of the different technology that sits within the AI space again in 2026? BOB FRIDAY Yeah, I think the biggest change has been, for most part, classic AIs where we built models to predict things, right? Much more deterministic, type of thing. The thing that's really changing around AI is all this non-deterministic agentic AI. this is really a new programming language, and you can almost think of it, the new programming language is English, right? We are now writing skill files in English to tell a LLM how to solve a problem, and we're building tools to help that LLM solve the problems. so we're kind of evolving from classical deterministic supervised training this non-deterministic agentic era of learning how to work with and program LLMs with English MICHAEL BIRD So you mentioned agentic AI. can you give a definition between, when we say AI, agentic AI, what's the difference between them? And are we expecting most AI that people are interacting with nowadays to be agentic AI? BOB FRIDAY I don't think it's gonna be all what I call non-deterministic. There's still a lot of deterministic AI under the hood. So when you build a solution, that does something on par with a human, it's gonna be a combination of classic supervised models, like computer vision, recognizing stop signs. In our case, we're building large experience models that can actually predict Zoom team user experience. That's classic AI trained models. and then we have this new agentic non-deterministic, where English is a new programming language, And I, I kind of wonder, going forward in this old world, I hired PhDs out of school, computer scientists. MICHAEL BIRD I think in this new agentic world, you know, maybe more about English majors and people who are artistic. There's this, this concept of prompt engineering, isn't there? BOB FRIDAY so that's what I'm saying. I make a barrel of wine every year, and it's like I used to tell people that organic chemistry was harder than networking, right? It's like very, non-deterministic. It's hard to figure out why one oak makes a difference. I we're gonna be the same thing as we get into this agentic world. I could have two engineers create two different prompts and get slightly different results. for me, the engineers are becoming like the winemaker. You know the, barrels are becoming like the tools, right? so I think we're gonna see people start to learn how to use this non-deterministic automation solve problems, and it'll be in networking, and all industries will start leveraging it to solve problems they haven't been able, you know, automate things they haven't been able to in the past MICHAEL BIRD I think you sort of touched on this, but, Do you think we'll see, the roles of technologists changing? You know, moving more towards it being slightly more a philosophical question or, a slightly more of an arts type question rather than an engineering question maybe BOB FRIDAY in, my classic AI, I hire computer scientists who really know how to train models, how to, wrangle data. this new agentic world where you're working with prompts, it's gonna be a little bit more artsy. you're gonna have to have a feel for all the different large language models, right? That's what I'm saying. It's like a winemaker, right? You know, people are gonna develop a feel for what models actually help solve different types of problems. MICHAEL BIRD I'd love to talk about data actually, because, AI is essentially fueled by data. Good data in, good data out. what does a good data strategy look like when it comes to AI? BOB FRIDAY I think a good data strategy, in my space here is more about I really combine the support team with, to the hip of the data science team. 'Cause ultimately the domain experts are the ones who know what data needs to be used In a network, and that is a support team. they solve problems on day in, day out. So when when I started Mist, I made sure my data science team and my support team are tied to the hips to make sure we got the right data back to the cloud to solve problems MICHAEL BIRD and good data is vital for making, good models, good decisions, all that sort of stuff BOB FRIDAY Yeah, I mean, you need the data to solve a problem, you know? In my case, when I started, the support team was logging into devices to get the data. You know, the data's gotta be in the cloud. I gotta have the data. I can't have my data sitting in a AP or switch. I gotta figure out what data it is, and really it's the support team who knows what data. Now, I tell you, the other challenge in this agentic world now that's changing is the software QA. In my old world, it's like here's a piece of software, here's a unit test, In this agentic world, it's actually you gotta get domain experts. It's not quite as easy to tell, did that new prompt make things better? MICHAEL BIRD So looking at the way the world has responded to AI, how have you seen and how do you think the jobs market's going to change and be influenced by AI? BOB FRIDAY th-this is the classic question. If you look back in history, most technologies create more jobs than they destroy, right? there's a transition period. I think what people are looking at in this AI thing, is this just another technology, bringing new jobs, replacing new jobs with old jobs, or is this more of the Industrial Revolution? is this gonna be some sort of societal change, where we automate things so much that, the future looks radically different than what we did? And that's what happened in the Industrial Revolution, right? We went from this agricultural society to a much more urban industrial society, I'm kind of leaning more towards the Industrial Revolution side. we have to fundamentally change how society works MICHAEL BIRD Do you think it's already influencing the job market? BOB FRIDAY it's definitely influencing the job market from a, efficiency point of view, right? if you look at the software community, right? you can actually start writing software, right? it's definitely making software engineers much more efficient and quicker, and that's happening as we speak MICHAEL BIRD Can I talk networking then? How has AI influenced networking? BOB FRIDAY so I have my customer support team actually use agentic AI, for every ticket that comes in, and we got up to maybe, 70, 80% of the time, it actually helped the support team solve problem. I think agentic AI is gonna let us start to solve the tail and the corner cases. these are solving hard problems like post-connection. Agentic AI is gonna really let us start to attacking those problems and start bringing that efficacy, north of 90%. MICHAEL BIRD so what makes good data BOB FRIDAY Well, actually, if you're training a model like a, large experience model you actually need both good data and bad data to train that model. MICHAEL BIRD Ahh I see. BOB FRIDAY and you actually need more bad data to train the model, 'cause if I don't have examples of bad, it's hard to train. Now, in the agentic world, it's not about data, it's about APIs, So if we go to classic AI, you need a combination of good and bad data to train models. In this new agentic world, it's about I need good APIs and good data to help solve problems. and usually you build that through benchmarking. is this API actually providing data, providing truth that's helpful in solving problems? MICHAEL BIRD interesting that, that concept of bad data, 'cause I think the, I suppose what you're saying is you can say to the AI model, "Yeah, don't do this, but do this." BOB FRIDAY Yeah. I mean, if you're doing agentic AI, you want good data to make good decisions. Uh, in classic AI, you want bad examples, you want bad data points and good data points 'cause you're trying to train models to recognize the difference MICHAEL BIRD Excellent. what should an organization be doing to best prepare themselves to be AI ready? BOB FRIDAY I think when we get into this agentic era, you really have to make sure your domain experts are part of your product development team, because we're basically trying to automate problems that domain experts are doing, In networking space, we're trying to automate what IT teams typically do to solve problems, right? And so you really gotta make sure domain experts and the development team are in lockstep, right? Because usually the person who knows a little bit about computer science may not know anything about the networking domain or whatever domain they're working on. Now, I would tell you, it's getting easier. the domain experts are like, "Hey, I don't even need any programmers. I'll write my own skills, and take care of this myself now." it is getting to that point where if you're a domain expert and you know how to write a skill they're starting to already automate what they do on a daily basis MICHAEL BIRD is that something that IT teams should be aware of, that sort of shadow IT, shadow AI, that sort of thing? BOB FRIDAY you see software engineering is already starting to leverage AI tools to make their jobs quicker and more efficient. I think support teams are doing the same thing. they're starting to figure out how to automate what they've been doing on a daily basis MICHAEL BIRD Well, Bob, thank you so much for joining us on Technology Now. Really appreciate your time. BOB FRIDAY Yeah. You know, topic dear to my heart. Always happy to chat SAM JARRELL Wow, I really enjoyed that conversation a lot. Bob is a fascinating guy. but what really stood out to me is that those of us with communications and English writing as a background, it's our time to shine. It's becoming a programming language. MICHAEL BIRD we've talked about prompt engineering before, and, and I think Bob just highlights like just how important that is. he used the phrase, uh, non-deterministic. So it's not one or zero. SAM JARRELL Yeah, exactly. And I appreciate that agentic AI sort of even changes the relationship further than LLMs have. Um, and I really also appreciate the idea that, he talked about this as like the Industrial Revolution, so, more people will get to participate in the job market in creating solutions, and it gives me hope for, AI not eliminating creative work or, the work of people who do things in fields like communications and instead just changes the way that the work is done and maybe there are actually new opportunities, as the barrier between technical and non-technical roles starts to blur a little bit more. MICHAEL BIRD Yeah. I sort of feel quite good about that. as somebody who's in a creative job, I think, yeah, this is my time to shine. I can just about write English and, I can come up with ideas. this is the right time for me SAM JARRELL He also kept coming back to, domain expertise. MICHAEL BIRD Why do you think he talked about that? SAM JARRELL I get the sense that, as, the technical and non-technical roles begin to blur, the people closest to a problem often know the best solution. And so success isn't, strictly about your technical coding or skills anymore, and actually having industry knowledge in a space becomes a more and more strategic advantage and allows certain experts to leverage AI on their own and automate parts of their own work MICHAEL BIRD it's sort of like, the people that are at the coalface. they're the people that know the intricacies, the sort of nuances of the particular problem, the particular job that they're trying to solve SAM JARRELL Yeah, exactly. I feel like we're moving into a world where, understanding the business problem is, I don't wanna say more valuable, but equally valuable to understanding the code behind everything MICHAEL BIRD Yeah, and how to solve it. And actually talking about code, the broader conversation about data. And the phrase that we've used on the show before loads is good data in, good data out. and I think you heard me slightly surprised when he said, "No, no, you also need bad data." SAM JARRELL yeah MICHAEL BIRD I found that realy interesting SAM JARRELL No, and it makes sense. I hadn't thought of it ever that way before too, but like when you're trying to learn any skill as like a person, you obviously want lots of really good examples. But it is helpful to see a really terrible example sometime and just go, "Okay, I know not to do that." MICHAEL BIRD 100% SAM JARRELL do have to say too, I was not expecting wine to be such a theme in today's AI discussion. But I really appreciated learning sort of Bob's view on that, and that like things aren't always perfectly predictable, and different prompts can actually create different outcomes, kind of like how different wines come out of, of what you're doing and there's sort of an element of experimentation and craft to everything. And success really your success is like based on experience and judgment MICHAEL BIRD different winemakers do things in subtly different ways. and I, I guess that's the point, it's sort of an art form rather than it being an absolute science. Now, Sam, I don't wanna sort of age us, but you and I are both millennials. SAM JARRELL Gosh, yeah MICHAEL BIRD We've been around for a while, and we had a front row seat on the evolution of AI. I would say the evolution of technology. But for those younger than us, AI has been a bit of a native part of their lives pretty much since the start. So the thing I wanted to find out from Bob was, is there any advice that he has for people growing up and entering the workforce in this new era of AI? BOB FRIDAY my good advice to, the next generation of kids coming up is be adaptive and ready for change. if I look at how fast things have changed over my career, when I started, there was no computers, there was no internet, there was no cell phones. I think exponentially things are changing almost yearly now in terms of... Agentic AI didn't exist two years ago, right? And now it's the latest and hottest thing. there'll be something next year. Things are just changing so fast. You have to be able to adapt very quickly SAM JARRELL Okay that brings us to the end of Technology Now for this week. Thank you to our guest, Bob Friday And of course, to our listeners. Thank you so much for joining us. MICHAEL BIRD If you’ve enjoyed this episode, please do let us know – rate and review us wherever you listen to episodes and if you want to get in contact with us, send us an email to technology now AT hpe.com subject line: non-deterministic SAM JARRELL Non-deterministic, sure. MICHAEL BIRD and don’t forget to subscribe so you can listen first every week. Technology Now is hosted by Sam Jarrell and myself, Michael Bird This episode was produced by Harry Lampert and Eva Higginbotham with production support from Alysha Kempson-Taylor, Nik Damarell Beckie Bird, Alissa Mitry, and Jenessa Ayache. Our theme music was composed by Greg Hooper. SAM JARRELL Our social editorial team is Rebecca Wissinger, Judy-Anne Goldman and Jacqueline Green and our social media designers are Alejandra Garcia, and Ambar Maldonado. MICHAEL BIRD Technology Now is a Fresh Air Production for Hewlett Packard Enterprise. (and) we’ll see you next week. Cheers! SAM JARRELL Bye y’all