E060 - AI or Not - Brandon Lozano and Pamela Isom

Welcome to "AI or Not," the podcast where we explore the intersection of digital transformation and real-world wisdom, hosted by the accomplished Pamela Isom. With over 25 years of experience guiding leaders in corporate, public, and private sectors, Pamela, the CEO and Founder of IsAdvice & Consulting LLC, is a veteran in successfully navigating the complex realms of artificial intelligence, innovation, cyber issues, governance, data management, and ethical decision-making.

AI can feel like a mind. Brandon Lozano argues it’s something far more practical and far more dangerous when misunderstood: a prediction engine fueled by data quality, incentives, and whatever slips into the pipeline. As we celebrate the anniversary of AI Are Not, we talk with Brandon, lead AI engineer at IsAdvice and Consulting, about how a career that blends computer science, data science, and policy awareness changes the way you build and govern AI systems in the real world.

We dig into Brandon’s path through research, including work at SLAC National Accelerator Laboratory and a formative look at how policing algorithms can inherit bias from skewed datasets. That thread leads straight into today’s generative AI reality: large language models can hallucinate, sound confident, and still be wrong. We unpack why LLMs are best understood as advanced “autocorrect at scale,” why context and memory features can quietly contaminate outputs, and why data integrity and traceability are the foundation of trustworthy AI.

From there, we get hands-on with applied AI for energy and infrastructure, including the challenge of accelerating microgrid adoption when information lives in silos and important inputs sit in unstructured formats like scanned notes and handwritten documents. We also talk AI governance, cybersecurity, and why “ethics” can be called rules or standards but cannot be optional, especially in an AI race mindset. We close with a forward look at robotics, edge AI, digital twins, semiconductors, and rare earth minerals shaping what 2030 may look like.

If you care about AI governance, data quality, cybersecurity, and practical digital transformation, listen now, then subscribe, share with a colleague, and leave a review so more builders and leaders can find the conversation. 

Transcript

[00:00] Pamela Isom: This podcast is for informational purposes only.

[00:26] Personal views and opinions expressed by our podcast guests are their own and not legal advice.

[00:34] Neither health, tax, nor professional nor official statements by their organizations.

[00:42] Guest views may not be those of the host.

[00:51] Hello and welcome to AI or Not the podcast where business leaders from around the globe share wisdom and insights that are needed right now to address issues and guide success in your artificial intelligence and that digital transformation journey.

[01:06] I am Pamela Isom and I am your podcast host.

[01:10] And so today we have a special guest, Brandon Lozano.

[01:14] He's the lead AI engineer at iSAdvice and Consulting LLC and a former intern at SLAC National Laboratory.

[01:24] Brandon, welcome to AI or Not.

[01:27] Brandon Lozano: Thank you very much Pam. A very warm welcome. I'm very happy to be here, very excited and if I'm not mistaken, I believe this is for a little bit of a special event for AI or not.

[01:40] Pamela Isom: That is true. So we are hitting our second year actually we had our one year anniversary and you celebrated with us part of that last year and and so this is our second anniversary.

[01:52] So we are excited about that and happy to have you be a part of it again.

[01:57] So to get started, will you tell me more about yourself, your career journey and how that has shaped where you are today and where you are headed?

[02:07] Brandon Lozano: Yeah, of course, I'd be happy to. So I'll give just a little bit of a blurb that's not so academic or professional. Just so people can maybe get a little sense of my personality.

[02:17] I guess so, yeah. My name is Brandon Lozano.

[02:19] My family is from Colombia, South America. I really like Star Wars. I've always been kind of nerdy, geeky. I like to build computers kind of growing up I played a lot of sports except baseball.

[02:32] I was just always pretty active to be honest.

[02:35] And I was always just like really curious. Love to read, definitely love to read. Always loved paper planes, model rockets, stuff like that.

[02:42] So now that I've kind of given a little personality blurb, I am lead AI engineer here at izz, Advice and consult.

[02:48] I've actually also been here for good a little bit. Probably going to be celebrating anniversary soon as well.

[02:55] But yeah, I've been here for since just around the end of 2024 in some capacity,

[03:01] you know, writing proposals, engineering solutions, designing things, working on different things here.

[03:07] But before that and how I actually kind of met Pam was as she mentioned my internship at SLAC National Accelerator Laboratory.

[03:17] So that was a really good time.

[03:19] I was a material science intern at SLAC which for my degree, which was political Science and computer science. Double majors with a minor in data science.

[03:30] That was very much none of those somehow right. With three areas of study.

[03:35] Material science was not that. And that's because when I was filling out the form I really wanted to kind of shift a little bit more into hardware. UNC Chapel Hill where I went to school at is very theory based.

[03:46] It's a liberal arts school and there's some state politics between the schools of like engineering being a little bit better at NC State.

[03:55] But I wanted to get a little bit more in a hands on work. I kind of wanted to get more into like really, really stemy STEM science and kind of interdisciplinary work.

[04:05] So I ranked my preferences as like I think astrophysics number one, material science and then just pure computer science, just something simple.

[04:15] And so the matching I guess just assigned me to chemistry,

[04:19] material science. My mentor, Hao Chen, very great mentor, still have a lot of love for him, still communicate with him as well.

[04:28] But I was assigned to that group and he had no computer science knowledge at all and I had pretty much no chemistry knowledge at all.

[04:37] So that was a pretty fun summer of just teaching each other and trying to work on something cool. That was inspired by his colleague's work from I believe,

[04:47] I think it was Lawrence Livermore, some work he had done in 2018. Something in the physics arena that I was kind of trying to generalize into just general research science.

[04:58] Thinking back, it was kind of like trying to have or implement a really, really primitive form of rag of retrieval augmented generation. Because right when ChatGPT first came out it was really great and everyone's obviously thinking about how can you use this for science?

[05:15] How can we do this? Can we get robots to use this? Can we get robots to understand instructions through this? Can we get robots to do research for us? Right?

[05:23] Just completely like accelerate research. Cut out the need for, you know, human sleep or human focus of counting each vial or swirling everything up for exactly 30 minutes.

[05:36] And the biggest problem was that CHAT CBT hallucinates And back in 2024, I think ChatGPT originally comes out like fall 2023.

[05:44] It's hallucinating to the point where it might say like two plus two is five, right? Like this is something that you can barely really trust at the time. It's still kind of a novelty, a really, really cool novelty.

[05:54] But it's definitely not to the point where we see today people getting AI psychosis, right? Because this AI keeps talking to them so normally and egging them on and on and on.

[06:04] That's its own whole thing. But anyways, I was aiming to fix that by using basically like a text corpus, specifically a training text corpus of just scientific research papers and kind of having this like provenance traceability which again an AI with a lot of things being black box models,

[06:22] you don't really know why exactly it's giving you an answer. You just kind of trust it because it's been right in all your testing. Right.

[06:28] So it was aiming to kind of fix all those things by getting it solid scientific research papers,

[06:34] turning that into embeddings and having the system understand a little bit some of these relationships and maybe be able to do something. Of course this was a two person team.

[06:44] One a recently graduated computer science undergrad and then a chemistry PhD who did not know much at all outside of maybe like jupyter notebooks in terms of coding or computer science.

[06:59] So that was a really good time and I met someone else to name job. I think he was also part of the last year's roundtable. Ryan Coffey.

[07:08] I actually met him in the last few days of my internship there because they kept saying oh yeah, he's a computer guy, I think he knows something about it. And this was pretty much the department I was placed in.

[07:20] This was pretty much the only computer science guy in like the entire department. Everybody else was chemistry. I was sitting in on groups discussing X ray spectroscopy, quantum chemistry,

[07:33] all this stuff just jotting down notes, trying to understand and reading their articles. But I met him and we just really hit it off. I guess we just were really chatting.

[07:42] He really liked my blend of like the political science and computer science. He really appreciated that like not just looking for a STEM pathway, but not to be too corny, too cheesy, but trying to use those skills to actually better the world or make things easier, innovate rather than just doing it for the money,

[08:00] just following whatever the money says. Like a Looney Tunes character. Right.

[08:05] So that was how I ended up here. And then before that I'd done some undergraduate research. I won't get super super into it, but I did some research at the University of Virginia that was in 2023 and that was research into kind of like data.

[08:19] My case study in California specifically kind of into how data is being used by police and how these algorithms might be a little skewed and how these, how AI and policing can kind of give an excuse of like well it can be biased because it's an AI, it' a robot,

[08:37] it doesn't have a bias. But again this, this issue of data Data Integrity comes up constantly and always should.

[08:44] Right. In a conversation where we're talking about AI,

[08:47] if you feed it bad or not perfect data, really,

[08:51] then this AI model is not going to be perfect. Right? So we can't pretend it's a perfect neutral arbiter, you know, maybe locking people up or giving people tickets when it might not even be, you know, trained properly or taking the right consideration.

[09:06] So that, and that was actually my first big introduction into like data science. That's what kind of made me really like and be into like data eventually, AI, machine learning,

[09:17] and just kind of like ethics or like data storytelling. I've heard kind of thrown around as like understanding data, interpreting it, manipulating it, being able to understand and tell a story from that data.

[09:31] And then before that, it's more just kind of general, like computer science, software engineering,

[09:36] and just like little student groups. I was very involved with SHIP at unc, the Society of Hispanic Professional Engineers. And I think that also really encouraged me, not necessarily maybe in a academic interest or professional interest.

[09:50] Like, it didn't really convince me to like, get into AI or anything, but I do think that being in that and being kind of like a student leader in some of these groups really helped me just not like, reject myself before like anyone else did, and just put yourself forward and just like,

[10:06] kind of gave me a little more comfort and community again, a little more experience with like,

[10:11] handling things, handling people.

[10:13] People sometimes get into conflict. People sometimes cause problems that you wouldn't really think to account for.

[10:20] So all of that was really good. And then,

[10:23] so I think that was really career journey, career background a little bit. But now what was the rest, I think you said, and where was I wanting to go or where, where am I now?

[10:34] Pamela Isom: So first of all, thanks for the, the background and helping me to understand even more about you. I'm sure that the listeners will find that intriguing as well.

[10:45] And so to add onto that,

[10:47] I am familiar, of course, with some of the folks from the laboratory. So. And yeah, we have done quite a bit of things together,

[10:56] some with you, some without you, but I think together there has been a lot that has been accomplished and there's more to come.

[11:03] What I want to know is adding onto that is how has your past research helped you with your current and planned endeavors?

[11:14] Brandon Lozano: Hmm. Okay, interesting. Yeah, that's a good question.

[11:18] It's really nice because I do feel, and that what I do really like about working on this current project here is advice, is that it does, like I said, it does feel like something that's actually helpful and actually,

[11:30] you know, innovative.

[11:31] And what's really nice is that it honestly feels like kind of amalgamation of everything I've picked up along the way. It kind of just feels like, oh, like this is what I was working towards.

[11:41] Like,

[11:42] it's kind of funny to think back that when I was in undergrad or just freshly out of college, that maybe one component, each one of these internships or each one of these final projects in undergrad really stressed me out, really put me through the wringer.

[12:00] And I thought those were all bad enough on its own. I was like, okay,

[12:04] made it through, made it through. And then now that's just work.

[12:08] It's just doing kind of each one of these things.

[12:12] And so my current project, the main project here is advice that I'm working on.

[12:18] Again,

[12:18] don't want to put too much information just because of research security, but we're working on something,

[12:25] a database capability that helps accelerate microgrid adoption by really addressing this issue of having a lot of like,

[12:34] data silos and kind of just really, just every step, every connection is just really fragmented. It's really hard to have kind of like a continuous planning process.

[12:46] And not just continuous, but also like cheap, reproducible. Because from what we've heard from interviews with, with some stakeholders,

[12:56] it's kind of just consultant spend, third party spend, or it's relying on just one piece of software's, you know, capabilities. And it's usually just maybe what they know or what they're used to because, well, we don't really see what this will really give and we don't even really see the economics.

[13:14] But if they were able to do more analysis and faster, right, Then they would be a lot more willing and a lot more capable of investment and acceleration in this industry.

[13:25] And so one of the biggest problems, right on top of the fragmented data is the different structures, right? So some of this data is really nice. It's really, really, really structured.

[13:36] And then the unstructured, right, the handwritten documents, that is, I mean, that's a whole other layer. I mean, humans got to take notes of that. Humans got to remember that.

[13:45] Humans got to process that. When you ask a human, right, if you're typing to this human,

[13:49] show me something about Alaska for the other engineers or for the other people who are kind of mathematically inclined, you know, the. Tell me the big O of someone having to look through documents and looking for Alaska, right.

[14:02] I can guarantee it's going to be a lot slower than even the simplest query or retrieval system, right?

[14:09] So like I said, each Little component of what I've done like at Stanford, at Slack,

[14:15] working with like neural networks directly,

[14:18] working with like data ingestion, data pipelines, data processing.

[14:23] That was really helpful. That was very nice.

[14:26] Very, very helpful. Again, working in like a scientific environment,

[14:30] very helpful for the current research.

[14:34] Just kind of understanding the scientific method process of course is always going to just be good reinforcement throughout like undergrad research and then specific data courses of course also fed into that Stanford.

[14:47] Right?

[14:49] But if I had to throw in data science and statistical modeling and methods would have been a whole other thing. Right. So and then I do really think that one particular class, in the same way that Stanford helped me with the pipeline and the processing and kind of these machine learning frameworks,

[15:06] right? Because in college they never want you to use the tools that are out available like on the Internet, right. They want you to show that you understand the concepts yourself so you make it right, you program.

[15:19] That's why even though it was just one little component in college for like a final project, it still put me through the wringer because I'm coding these data structures themselves.

[15:28] I'm making sure they have the exact appropriate behavior according to the theory.

[15:33] But one class was nunc was very theory heavy. One class was a software engineering lab.

[15:40] That class was extremely helpful for like industry practice because again,

[15:47] in kind of normal computer science, especially if it's not specifically called like software engineering, computer science probably means it's pretty theory heavy.

[15:55] But in those classes we're coding all these structures ourselves, like we're learning about something in the class and then we're building that. Right? We're doing that ourselves and we can't use like the easy online things.

[16:07] We're not really supposed to be like copying code. Of course this is like pre chatgpt,

[16:11] so I don't even know what's going on in those classes these days. But you're supposed to do all that and you save it all on your computer and you turn it in like any other like economics assignment, political science essay, whatever in real world, in the real life,

[16:26] first off, you don't have to code your own data structures because someone already did that and they did it way better than you ever could. And you're just going to waste your time if you do it.

[16:35] Third,

[16:35] you can definitely look up code all the time because again, if someone's already done it and you're just trying to figure it out your own novel stuff,

[16:42] cool way,

[16:43] you're kind of wasting your time. So no boss would ever tell you, don't look up if anyone's ever done something similar to this.

[16:50] And then fourth, just kind of like the business environment. Again, things you don't think about at all as like a student,

[16:57] but like actually using GitHub for like collaboration.

[17:01] Like sure, I use it to like save my progress because that's what it is basically like a file code, storage, code repository.

[17:10] But when you're actually working with other people and you have to make sure you guys aren't stepping on each other's toes or completely erasing someone else's progress or making sure this still works with the newest version of his thing and working in like a huge code base of well,

[17:27] huge. Right? Again, for like a project at that time it was like 10, 20,000 lines of code, which I probably got scared at that time when I saw like 300 lines of code.

[17:38] So that was just really helpful to understand. Kind of like front end, back end app, like organization and development database, like connecting everything, full stack development.

[17:49] That was really good. That was definitely a super useful class for like software engineering and kind of like actual street industry practice and use of these tools and environments. So those I think are some of the biggest things that I think really helped me in this project.

[18:05] There's also some, you know, an internal kind of like policy or like government memo intelligence tool that I made a little bit ago. And that kind of in a similar way was also kind of like a refresher or kind of like again another, another practice in like pipelines like integrating real time data,

[18:24] having like visualizations and kind of working on like a user interface.

[18:29] So yeah, I mean I would say I definitely had a, a lot of things that have really helped me with this current research.

[18:35] Now I think Slack definitely helped in terms of like the research and like working with like a scientific team because again, had I not had that experience I think I would have been a little, a little bit of like a culture shock, a little bit of a just, just diffusion.

[18:49] A lot of unspoken rules or just things that people just know. Once you know it, you just keep it and you don't even really think to mention it.

[18:57] So.

[18:57] So yeah, I feel like I definitely got a lot of experiences to draw from to help me in this project.

[19:04] Pamela Isom: I think that's good. I,

[19:06] when I graduated years ago and then every day,

[19:12] I mean I'm always looking for practical ways to take what I learned and apply it and I really want some, want to do something that's impactful and gonna make a difference and so, and I think that's the best way to learn I remember the applied experiences from, of course,

[19:31] my university days and my graduate school days.

[19:35] I remember the exercises, I remember the experiences, the opportunities and all that.

[19:42] But I think the most learning opportunity was when I put it to work for business,

[19:50] because that's when you know that it's almost like that's where the rubber meets the road.

[19:54] So if you can do that while you're in school, that's great.

[19:59] And I think they try to do that with the curricula. They try to make it as much like reality as when you get out there. Yet it's still a lot to learn.

[20:09] And so it's good if you can take, and I would say that to anybody, if you can take the experiences that you've learned and when you get these exercises in grad undergrad, going for your doctorate, all that, when you get those assignments, know that those assignments really are equipping you for more intense work in the real world.

[20:29] Because right now our work is intense. And so I'm glad that I'm not having to.

[20:35] There's some things that we don't have the bandwidth. So I'm glad that you, you bring that to the table. So we appreciate that. And so for the undergrads and those that are learning,

[20:44] know that those things that you're learning matters when you start to put it into practice.

[20:49] Brandon Lozano: Yeah.

[20:50] Pamela Isom: And so building on top of that, what do you wish people paid more attention to or knew more when it comes to AI?

[20:58] Brandon Lozano: Gosh, I mean, since again, since my first endeavors into data AI, I always link them data. Right.

[21:08] AI is really cool. And like I said, it's,

[21:11] it's very tempting to like artificial intelligence. It just rolls off the tongue. It just sounds, you know, kind of royal already. Right. It sounds like something we should listen to.

[21:21] It sounds like something, you know, again, like out of a sci fi film, like something we shouldn't even have as real.

[21:28] But again, like, at the end of the day,

[21:31] I don't think anyone would tell you that any of the AI models are truly intelligent or sentient in any way. I don't believe that anyone, at least at these labs, like I said, some people using these products might think differently.

[21:52] But I think anybody at the labs would not make the claim that these are sentient or intelligent really.

[21:59] And at the end of the day,

[22:02] the current basis for our, you know, kind of understanding our notion of artificial intelligences,

[22:09] LLMs, large language models. Right.

[22:11] And at the end of the day,

[22:15] again, that's the core of the model. And scientists are pretty clever with the way in which they use those capabilities. But LLM is a very, very, very, very advanced version of like autocorrect.

[22:29] The same way where your phone can predict the next word that you're thinking, right?

[22:34] Using all of the data, of typos, of all those errors that it would suggest you to change it stored all that. Of all the messages or all the data that they use to train it so that they know what would sound right or what sounds weird, like what might be an anomaly.

[22:50] Like, oh, they probably meant to hit the O, not that I.

[22:55] So it all boils down to that. It's. The AI is not necessarily thinking or genuinely giving you like a new insight. It's kind of just predicting what sounds right to what you gave it, right?

[23:10] Like the prompts that you gave it.

[23:13] And so again,

[23:14] that's only even possible right now because of how much data they have and how much data they collect and how cheap it is to collect that data,

[23:24] how profitable it is to sell that data, right?

[23:28] And then,

[23:29] you know, all the investment around the world that you see going into AI, right, Everyone clearly thinks that they need to spend all the money because then they'll win and everyone else will be the ones who wasted all their money on AI.

[23:44] So again, going back to this, just like models are what you give it, right?

[23:51] If we don't have like an idea of what's going in, or if we don't know the quality,

[23:57] or if we don't know how it's coming to certain decisions, then we really risk this really bad situation I keep bringing up where we just think the AI is just neutral, it's just perfect, right?

[24:07] And so having like data integrity so that we're not even thinking, we don't even want to think about malicious actors, right?

[24:16] But of course we have to data integrity in terms of that. Like, even if we did have a perfect model, what if a malicious actor all of a sudden starts poisoning the data so that it's like brainwashing our population slowly and in ways you can't notice, but it's starting to foment hatred or ferment certain feelings into population and we might not even know if we're not keeping a watchful eye,

[24:40] right? I see this misconception that that's just slowing us down, that's bogging us down. It's the AI race. It's the AI race. We need to get ahead, we need to do everything.

[24:48] We need to accelerate, accelerate, accelerate.

[24:51] But I think it's, it's, it is necessary to, to stop and take some considerations I know some people have issues with, you know, ethics or slowing down,

[25:01] but at the end of the day, like, even if you're not a fan of that, even if you don't believe in ethics, right? It still helps you keep secure,

[25:08] like, keeps you like a secure environment and helps you keep track of things. Right?

[25:13] That's always good, right? That's not a good thing or that's not a bad thing to know that you are moving data in secure ways, handling data in secure ways. Your training data is not being manipulated.

[25:25] Your training data is actually valid. It's not being skewed to any which direction, dimension,

[25:32] whatever, and understanding, you know, where this data comes from and where your pipelines are bringing. Right? Because there's already kind of this issue I've seen a lot of people mention.

[25:42] It's like, okay, generative AI, of course, has really taken the Internet by storm,

[25:47] again,

[25:48] a lot by malicious actors, because it makes a lot of things really, really easy and really, really cheap. But it's really taken everything by storm. And the more that the Internet becomes generative AI, like, output,

[26:01] the more that generative AI is just training itself on generative AI, right?

[26:07] So it's like at that point,

[26:09] even with all the data centers in the world,

[26:12] we're not moving. Right.

[26:14] It's the same thing as Moore's Law, right. It's moving really, really fast right now. But at a certain point,

[26:20] we've again hit a wall called reality.

[26:23] Pamela Isom: I have an example.

[26:26] So I like to use the AI tool sometimes to help me with a sentence to make sure my grammar is okay, right? Grammatical things.

[26:38] But there was one time when I said I want to really emphasize, and I'm trying to emphasize this particular point when I gave it the sentence to make sure the point was coming across.

[26:52] And so for this conversation, let's say that that point was misinformation. Let's say it was that.

[26:59] So now every time I'm using the model,

[27:02] it keeps bringing up misinformation.

[27:04] Brandon Lozano: Yeah, right.

[27:05] Pamela Isom: Like, no, no, I'm not talking about that. And that is because it learned that that was something I was interested in from a different conversation. And it retains information.

[27:16] Right?

[27:17] And so now I have to spend time making sure that what it is trying to communicate to me to if I want to get some. If I want to have a brainstorming exercise, for instance, with the AI tool,

[27:34] I have to make sure now that it doesn't sneakily throw in something about misinformation I'm not even talking about because it's holding it Retains.

[27:43] Brandon Lozano: Yeah,

[27:44] similarly, like, it's again, just trying to give you what you want. Right. Like it's, it's not thinking constantly.

[27:52] Oh, like. And this is a good place to talk about misinformation too. It's just thinking a tool, call whatever a tool called memory. Right.

[28:02] Look at some of the stuff she said recently.

[28:05] Oh, misinformation. Just throw something in there. It's not thinking. Right. It's not sentient. I think that's a big,

[28:11] big thing about the AI that a lot of people don't understand.

[28:14] Pamela Isom: So you said, what do you wish people paid more attention to? And so your point is things like that and also to be sure and check the context of the output that it's producing.

[28:25] Is that correct?

[28:25] Brandon Lozano: Oh, for sure. And then I'd also just quickly and I kind of mentioned it, but to make it explicit,

[28:30] cybersecurity is very, very important. Again, thinking of these malicious actors, because if you don't,

[28:37] that's a gold mine for these malicious actors. Right. The second you don't think about something, the second that you trust something,

[28:44] it can be taken advantage of. Right. So definitely very big. So, yeah, like I said, the biggest thing,

[28:51] data and AI, like, they're like inseparable. This AI cannot just be this perfect thing that fell out of the sky and is like bestowing hallucinations and knowledge to us. Right.

[29:02] It's completely powered by data. That's why we have the data centers. That's why we have all these things, all these problems. And, you know, that's why America is able to go really fast, whereas Europe, with a lot more, you know, EU GDPR regulations, other things to consider,

[29:19] they've been a little slow. I think that's also why China is able to go really fast is because, you know, they,

[29:26] the Chinese government has access to any data that it wants,

[29:31] no question.

[29:33] Pamela Isom: Yeah. Which makes me think about your concern around data access and access to information, period.

[29:39] And that goes back to the.

[29:41] So every company should have ethics, period. Every company has ethics, they have ethics policies, they have ethical guidelines. Those guidelines need to be applicable to the models and to the data as well as to the people.

[29:55] And I think that we can say we don't. There may be some that don't want to call it ethics. I don't care. Right. There may be some that don't want to call it that, but the ethics needs to be applicable to the machines as well as to the people,

[30:10] period. And so that's what I think I heard you alluding to, and I agree with that.

[30:16] Brandon Lozano: Yeah, for sure. Like you said, even if you don't want to call it ethics. Right. Like, at the end of the day, it's just, it's just rules or standards. Right.

[30:24] Pamela Isom: Moral. Like, where's your moral compass? Like that moral compass must be there.

[30:28] Brandon Lozano: Exactly. But even to not appeal to the moral compass, because money can definitely make that go away pretty fast.

[30:35] Even to not appeal to it, like I said, like, you just need to have like an inventory, like a. You need to understand what's going on. Right. You can't, I guess, unless you've just given up all morals, really.

[30:46] You can't just sell a product or release a product in the terms of AI, I think, is kind of a more appropriate word. You can't release a product into society, into the world.

[30:56] Right.

[30:57] Without having any understanding on what it does or what it could do or what it shouldn't do or should do during normal operating conditions and environment.

[31:08] Pamela Isom: Exactly.

[31:09] Brandon Lozano: So, yeah, I mean, I think it's silly to, to pretend like ethics isn't important.

[31:13] Even if you do, even if you do think that ethics is not important. I, I really, I just don't think you actually do. I, I think maybe at most you have a problem with that word or, or you think it's a little weighted or.

[31:24] Or something. But yeah,

[31:26] you just have to. I mean,

[31:27] there's no way around.

[31:29] Pamela Isom: Yeah, it goes to what I like to deal with, which is AI, security and governance. Right. So how are we governing our solutions?

[31:37] And how are we governing the people, process and technology?

[31:41] So let me ask you this because we were talking earlier and you like robotics, and I just kind of want to know more about that. Last. Last night I was at a birthday party with my little one and she turned five and she got two robots.

[31:57] And so it was so much fun, but I had to leave and go home. Like, grandma has to go home. Grandma go home.

[32:03] Brandon Lozano: Right.

[32:05] Pamela Isom: And so tell me about a different context. I know you like robotics, so tell me more about that. What's going on?

[32:14] Brandon Lozano: Yeah, I do like robotics. Unfortunately, so far it's been mostly as a,

[32:19] as kind of a hobby or just as like an interesting. I'm. Like I mentioned to you the other day, I am going through a move pretty soon, so I'm not really trying to be getting all that kind of stuff all, all into this space right before I leave, but I am planning on getting my hands on some stuff and building some projects just because that really interests me.

[32:39] Like I said, I used to like building my computers,

[32:42] but unfortunately all these AI labs are Buying up all the RAM and the graphics cards and everything. So that's getting pretty pricey these days.

[32:53] Pamela Isom: Are you gonna build a robot? Are you trying to tell me you're gonna build a robot?

[32:57] Brandon Lozano: Something like that? Yeah, for sure. An autonomous something at the least. But no, yeah, robotics is super cool and it's gotten very, very cool in the past few years. And I really think there's a few key developments that have really allowed this.

[33:14] So first, I think the most,

[33:17] not really most obvious, I think most pressing one, maybe people think about a lot are these rare earth minerals, metals, right, that allow us to make semiconductors or really, really small chips.

[33:30] That's like the biggest thing is that things have gotten really, really small and cheap, right? This phone, right, that we all have on us is an extremely capable computer.

[33:39] Even before, again, ChatGPT or all this hype about AI, AI chips, you know, before and Apple actually, with their M1 chips and stuff for their Macs and their proprietary chips for like phones, iPads and stuff, that was actually also a really big, really big jump, pretty big innovation from them and I think kind of woke up the industry.

[34:00] But those kind of circuit boards and things that are able to run AI,

[34:06] like inference, like edge AI that has gotten so, so small, right,

[34:12] these days and it's gotten a lot cheaper to do it.

[34:15] And so that's a big reason why things are able to move a little bit more. With robots, self driving cars as well, all that work with sensors and mapping out a physical environment,

[34:26] that's really helped.

[34:28] Another thing that's helped a lot is the advancements of machine learning and general hardware as well, allowing for more digital twins or virtual simulations, right? So instead of Boston Dynamics, right, Having to build a robot or 10 robots and run hours and hours and hours of just watching this robot,

[34:51] like try to learn how to walk, right? Try to learn how to climb a stair, try to learn how to react when it gets hit by a balloon, right?

[34:59] Instead of doing all that,

[35:00] just load it in a simulation,

[35:03] create hundreds that run for 24 hours, seven days a week on a supercomputer or data center or some kind of cloud infrastructure, whatever,

[35:11] and then bam, put it in the real world, have it tweaked for real world variables and everything,

[35:16] and you've got a pretty capable autonomous device.

[35:20] I think something else is all the, again, ChatGPT general advancements with like the machine learning, which is all enabled by the hardware.

[35:29] I want to really, really emphasize that it's all emphasized by the acceleration and developments with our hardware, with our process with our processors, with our chips.

[35:40] Another hardware design choice is rather than using, like, hydraulic joints, they're using, like, servo motors, which are a little bit more articulate and just seem to be a little bit better for this, like, balancing and things that robots have gotten a little bit better at recently.

[35:58] I do remember, I mean, back when I was kind of last interested in robots, before this recent boom,

[36:05] was when I was probably, like, in high school, and it was really just like Boston Dynamics robots at that time.

[36:12] And I do just remember how back then,

[36:17] something extremely impressive was the fact that you could kick this thing over and it could stay up,

[36:24] or the fact that you could kind of drop it and it could stay up. Right? It didn't have to get up and get into the perfect position. Right. It could actually balance and adjust for different forces and everything.

[36:37] Now,

[36:39] first off, that's just like a TikTok video is like watching a robot box someone.

[36:44] Now you can watch someone fight a robot. The robot jumps back up, puts in,

[36:50] which is crazy.

[36:53] Now,

[36:54] again,

[36:54] China is just hard to ignore in the AI space these days. And they had some kind of state propaganda, you know, showing or kind of demonstration with robots.

[37:04] These robots are doing, like, really crisp, really coordinated.

[37:08] I think it's kung fu, like a whole kung fu show. And they're doing flips and they're doing all this stuff, and none of them fall over. None of them have any progress.

[37:17] And of course,

[37:18] with the specifics of that, it's like, okay, was it programmed to just do that? Exactly. And they tested it on that stage for, like, six months, and they got it perfect, perfect, perfect so that nothing would mess up.

[37:30] Or were these robots that, right, maybe were given a video or were given instructions, like, once, and then autonomously executed this, of course, there's room in the air for that.

[37:40] But regardless,

[37:41] again,

[37:42] 10, 15 years ago,

[37:44] extremely impressive. If it can stand on its own,

[37:48] that's our bar, right? If it can stand on its own and move,

[37:52] actually, that's pretty much the bar. And you're already talking futuristic 10, 15 years back.

[37:59] Now we're thinking,

[38:01] well, did it really do that on the first try, or did they just specifically program it to do flips and to do all this stuff?

[38:10] And then like drone shows, you know, again, also show the coordination and the autonomous capabilities.

[38:17] It's. It's a new age. I mean, like. Like we say, some of it makes us kind of laugh at how incredulous it sounds. And then some of it is kind of like, oh, yeah,

[38:26] that's. That's it is a new, a new era.

[38:29] Pamela Isom: So you're saying robotics is basically the norm?

[38:33] Brandon Lozano: Yeah, I mean, robotics, I mean, again, economics and I think the most clear way, just because militaries are always going to be using experimental technology, they're going to be kind of at the forefront of the defense industry, usually is kind of where a lot of experimental technology ends up.

[38:50] A few years ago these things would be called cruise missiles, something that you could shoot and kind of glide a little bit to its target.

[38:58] And now. You mean, now these other countries are mass producing these things that are pretty much gliders with a lawnmower engine strapped on and an AI chip and then go ahead, you know, find your target.

[39:09] So it, I think it is going to be the norm, unfortunately. I think this rare earth metals, you know, you see it in some headlines. I think it's going to breed some conflict, breed some problems.

[39:20] Again,

[39:20] unfortunately, we're in a new age where that might be the thing that is starting to cause a lot of issues, a lot of friction between states. But that's a little bit of my political science bleeding into the tech realm.

[39:34] But yeah, I mean, I think robotics is going to be the norm. And it's again, really cute when I'm at the Japanese restaurant and little robot brings out my sushi with like a tuxedo.

[39:45] But yeah, I mean, I think it's going to be better. I've even seen some cool things from like Roomba, like Roombas that can like climb stairs or like pick up a sock, like,

[39:54] so it's, it's cool. I mean, I do think, like I said, we're going in a new era. I really don't think I could have imagined.

[39:59] I still can't imagine what 2030 is going to look like, but I don't think I could have imagined what 2030 was going to look like, you know, 10 years ago,

[40:07] five years ago even. I mean, it's, it is, I got to say. I mean, it is moving fast, for sure. I mean, we said we were going to move fast, we were going to accelerate and we, we definitely have.

[40:18] Everyone has.

[40:18] Pamela Isom: I mean, well, if you do decide that you're going to start creating robots, then we should talk about how we can do something constructive. Yeah,

[40:29] yeah, for sure,

[40:31] for sure. With all the creative stuff we've been working on, that would be another thing to add.

[40:36] And we would put in the burdenless boundaries that we would need to put in place to guide our solutions and our robotics to more, even higher quality outcomes. So that's what we would do so let me ask you this.

[40:53] This is. So we're getting at the end of this show,

[40:56] considering your experience and the research that you are doing,

[41:01] I'd like to know if there is a call to action or words of wisdom or both that you would like to leave with the listeners. And I usually ask this as the last question.

[41:11] If there is anything else that you wanted to share with us,

[41:15] feel free and then wrap. Let's wrap it up with the the last question. Either a call to action or words of wisdom or both that you would like to leave with the listeners.

[41:26] And again,

[41:27] thank you for helping me celebrate our second year of AI or not.

[41:33] Brandon Lozano: Yeah, of course. I'm happy to be a part of it. I'm happy to be here.

[41:38] So words of wisdom.

[41:40] I think I will just say just to the other young professionals who are coming out and maybe not the most optimistic market or new is,

[41:51] I would definitely say don't reject yourself before anyone else does. Like, it sounds a little silly, but make it that someone else is rejecting you. Right. Don't ever tell yourself no or stop yourself.

[42:05] Right? Why would you ever do that? It should only ever be somebody else that's telling you no. And again, first listen, that sounds kind of weird, like wanting other people to tell you no.

[42:16] But I would much rather have someone else tell me no 10,000 times than tell myself no once.

[42:23] So I would definitely say that to the young professionals who might be tuned in. I mean, anyone who needs to hear that. Maybe it's not even just to y', all, but I know it's a.

[42:32] It's a little rough right now.

[42:35] As for call action in terms of more kind of subject matter or kind of AIE definitely just learn more and just educate yourself.

[42:45] I think a lot of people are kind of just listening to what's going around on headlines and not really sure what's going on or not really understanding. And you don't have to have a computer science education or mathematics education or anything to,

[43:00] you know, just get a little into it, just understand what's going on. Like what's its limits, what are its strengths, Right? Because right now it's like it's a time, right?

[43:10] AI. Everyone's trying to figure out what exactly it is. What do we mean when we say AI? Everyone's trying to figure out what is AI useful for. Everyone's trying to figure out what will people pay for to include AI, Right?

[43:25] So in a lot of different ways, in a consumer space,

[43:29] definitely learn more. Because I can imagine this is a risky environment to traverse because everyone is just trying to get your money and prove to some investor, right,

[43:38] they have some kind of viable product. Right? So be careful.

[43:43] As a tech person, I never buy the first iteration of a product because I know that when it comes to tech, that's usually the, like the beta testers, the, the early adopters tend to be the ones that get the kind of crappy thing that gets ironed out in like another generation or two.

[44:00] So to the consumers,

[44:02] be careful, be aware, stay informed. To anybody just kind of interested in the direction of society, direction of some of the most profitable companies we've got going on, definitely inform yourself again, learn the weaknesses, learn the strengths.

[44:17] Maybe the most reliable source of information is again,

[44:21] not the people trying to sell you something very specific.

[44:25] So, yeah, definitely inform yourself, learn more about it. Maybe there's even things that you don't think is, is right. Maybe write your representatives or write, you know, an angry, angrily worded blog post or something.

[44:37] But just, just get educated, get, learn about it. A is not going away. Even if it's a bubble, even if it pops, not going away.

[44:45] There's no way that can go away at this point. No way to put the genie back in the bottle.

[44:50] It's going to affect everything we let it affect.

[44:55] So not only get educated on how it works, get informed, learn how to use some models, learn again, differences between them, parameters. What does that mean? Like, is that a bigger number better?

[45:06] Is it lower? Better? You know what's going on.

[45:08] Just learn a little bit because, yeah, I really don't think it's going on. And, and the way I like to think about it is that I, I don't think it's like the end all, be all, like again, like divine intelligence that we're going to get in like the next year.

[45:19] But at the very least it's going to be like a screwdriver or like a hammer, you know, if you. Or a drill. So like a drill or like a hammer, right?

[45:27] Once you got a hammer, why would you still bang in nails with your fist? Right. Once you've got a drill, you're not going to be like, you know what, for old time's sake, let me assemble this table with a manual screwdriver, right?

[45:38] You're just going to be like,

[45:40] so definitely, you know, learn it, it's going to be around, it's going to be here. And you know, you never know. Maybe it's going to be the thing that enables you to do the very thing you've been wanting to do your whole life.

[45:50] Who knows.

[45:51] Pamela Isom: I like that. So we can look at and consider AI as an enabler and just be strategic about it.

[45:58] Yeah. So I really like that. That's good insight. This is. This has been very helpful and very informative. You have a lot of knowledge that came across in the discussion today.

[46:07] I appreciate, I appreciate it,