(0:00) Intellectual property exists at all stages of this pipeline. (0:03) I think the critical piece is the human-created (0:07) data that can be used by the machine for training purposes and emulation purposes. (0:17) Welcome to Tech Unhinged.

We have Stack Mellon in the house today. He’s the director of NASA’s (0:22) JETS program, which is supported through the Texas State University. He spent decades working (0:27) at the intersection of energy systems, smart infrastructure, networking, and intellectual (0:32) property.

Over the course of his career, he has held several senior engineering and executive (0:36) roles across different organizations, including HP, General Dynamics, Compaq, etc., and also held, (0:42) you know, multiple deep tech startups and kind of getting off the ground and taking them to (0:47) fairly large scale as well. So alongside his industry work, Stan is the author and editor (0:52) of several foundational books on smart cities, data security, and overall trust and connected (0:57) infrastructure. Stan, welcome to Tech Unhinged.

All right, so you’ve worked across aerospace, (1:02) energy, startups, and academia. So what was the one technology within these domains that you were (1:07) certain would change the world, but it just suddenly, you know, kind of changed it in the (1:10) way that you actually thought it would change it? Mobile phones probably would be number one on that (1:16) list. That was obviously a technology that was going to be very, very popular, but did not expect (1:23) way that it would be attaching so broadly to people.

I think that would be number one. Number (1:28) two would probably be drones. Interesting.

So you thought that, so you think that mobile phones (1:33) have become much more greater as part of our day-to-day than you were thinking they would be? (1:39) Yeah, they’ve pervaded daily life in a much more intense and intimate way than I would have (1:46) expected from a communications technology. Right, but I think that’s probably more because of (1:52) the social network effect as well, right? So being able to be able to connect with mass number (1:57) of people just through that device or through that app. So I guess that would be one of the (2:01) factors that kind of, you know, gave it a rocket boost, the entire technology.

Yeah, I think that’s (2:06) right. The apps are the most important thing on a communications device, and the social media apps (2:11) have really been very powerful. Did not expect the communications platform per se to become, (2:17) let’s call it a weapon, to be weaponized in the way that it is.

Yeah, yeah, that’s true as well. (2:24) So that’s an interesting thought. And why drones? You know, when drones, when I first saw the (2:28) commercial drones coming out, I thought that’s cute, that’s fun, that’ll be a lot of (2:33) entertainment.

But then the Ukraine-Russian war started, and Ukraine began to use the drones (2:40) in military offensive ways, which are absolutely fantastic. I think absolutely fascinating. (2:46) More is never fantastic, but fascinating.

I think the use of the use of enterprise class (2:53) or commodity class drones in that fashion was not something I would have expected. (2:57) So we were more thinking of them like as toys, but they ended up in a very different, (3:00) I think affected a lot of industries. I think tourism got impacted a lot by the videographies (3:06) and the kind of videos that you can produce using drone technologies.

War, obviously, (3:09) is one of them. Spying generally and those kind of things. So yeah, that’s an interesting fact.

(3:15) You’ve been building complex systems, you know, long before AI kind of became mainstream. So when (3:21) you look at today’s AI boom, does it feel more like a breakthrough or more like deja vu? (3:26) Neither one. I think it feels more like market texture.

I think we’re not seeing AI, (3:31) we’re not seeing the entire picture of AI. What we’re seeing pervade everyone’s consciousness is (3:38) large, the use of large language models. And large language models are a small subset of the general (3:43) concept of AI.

That’s true. So I think the market texture or the promotion of the use of large (3:48) language models has kind of gotten ahead of where it probably ought to be. No, you’re right.

Because (3:53) so LLM just kind of mainstreamed your access to artificial intelligence to general people, (4:00) right? So I think that’s and they kind of picked up the most use case that might be most common (4:06) throughout and then directly, you know, allow people to have access to those day-to-day where (4:12) they can employ AI day-to-day usage. Otherwise, it was largely fixed towards, you know, large (4:16) systems, specialized industries, specialized use cases. So that’s one of the things that maybe (4:21) helped make it more mainstream.

But you’re right, people kind of confuse it with the initial, (4:26) like the way I like to call it is you have this real AI and then you have this LLM AI, right? So (4:31) that’s where I think the difference comes in. Let’s talk about where energy and AI kind of (4:37) come into the intersection. So AI is often discussed as a software problem, right? But (4:41) your work primarily has been more in the hardware and infrastructure side of things.

(4:47) So what do you think, where do the most AI conversations kind of get wrong about energy? (4:52) I think the use of AI in the management of energy delivery is something that hasn’t been (4:59) touched on very effectively. And I think that’s something that probably when that emerges as a (5:08) real conversation, I think that’s going to drastically affect the world that we live in. (5:13) We are highly dependent on electrical energy today.

We’re highly dependent on it. It’s not (5:19) going to decrease. And the electrical energy that’s delivered to us today is delivered over (5:25) ancient systems using ancient architectures that are highly human-dependent in terms of operation.

(5:32) And I think the application of some form of targeted AI in the optimization and the management, (5:40) the ongoing management of that energy delivery system is going to change a lot of things for us. (5:46) And I don’t think that has really come to the forefront yet. (5:50) So when you talk about energy distribution and management, what do you mean specifically? (5:54) The way electrical energy is delivered today is built on a very, very old system.

(6:01) A lot of hardware sunk costs, right? Wires and transformers. And it’s very expensive to maintain. (6:08) And it’s very expensive to make larger.

And what we’ve seen in the last several years (6:14) is a desire to cut back on the generation of electricity using pollution causing methods. (6:22) So you have the generation, you have the transmission and the consumption parts. (6:27) And the generation needs to increase.

In the US particularly, we need to have increased generation (6:33) because we’re not going to have enough capacity to meet the demand. But it takes a long time (6:37) to create new generation facilities and a lot of cost. And so instead of embarking on, (6:46) the AI boom is clearly going to consume more energy than we anticipated.

And it’s going to (6:50) continue also. So we have simultaneously this problem with constrained supply and increasing (6:55) demand in electrical energy. And we can’t increase the supply and we can’t hold back the demand.

(7:02) We’ve got to have some sort of intermediary play. And the intermediary play, I think, (7:06) may be the focus of some of the AI development energy, AI development effort towards the (7:14) management of the energy that it consumes. I can foresee there’s a lot of compute power (7:18) that is going to be needed.

So specifically for all these LLMs and everything, (7:23) there was a boom when all this cryptocurrency mining kind of began. Now there’s the next, (7:30) the way I see the next major power or energy demand will be coming in from these AI-driven (7:36) compute growth systems. Now, how unprepared are we today? I think that’s primarily the question (7:43) that needs to be thought about.

Yeah, that’s exactly right. And I think the solution to (7:47) the problem may be focusing some AI resources on dynamic management of its own problem. And by (7:55) extension, dynamic management of the problem that it creates for humans.

In Texas, Texas has its (8:02) own sort of separated grid. In Texas for a while, they’ve had controllable load resources to address (8:07) the Bitcoin mining issue. So if you have a Bitcoin, if you’re a Bitcoin miner, you can (8:12) basically create your own highly focused data center and you can register as a controlled (8:16) load resource and they can turn you off, right, to shed load.

That kind of partially addresses the (8:22) issue because we need to be able to turn on resources that optimize the load, not just shed (8:28) the load. Yeah, that’s true. I mean, shedding the load is like, it’s just like putting a bandaid on (8:33) the problem itself, but not actually really solving the problem itself.

That’s right. That’s (8:37) right. So for the last several years, I’ve been involved in an organization called Grid Pathway.

(8:41) And Grid Pathway has a think tank sort of idea. And there’s a Grid Pathway Institute, (8:47) which is a collection of people who are very well-heeled in this particular, the management and (8:54) structure of the grid. And this is sort of the outflow of that thought process is that we need (9:00) to focus some attention on management of the grid in different ways.

And one of those ways to manage (9:07) the grid is to use some form of agentic AI to optimize how the power is not necessarily used (9:14) or distributed. When you say manage, so what exactly, I mean, is it making energy systems (9:19) more faster? It’s just understanding how data, how energy is being used right now so that you (9:25) can then see where we can optimize it. So what exactly? All of the above.

All of the above. (9:30) Right now, a lot of these systems are managed manually. And what’s happening is that there’s (9:36) there’s becoming a faster event horizon on issues associated with energy.

So it’s built to be kind of (9:44) a static system in certain respects. And the requirement for faster changes or faster updates (9:50) is just outstripping human facilities. Because it’s humans that are making these decisions, (9:56) and they’re making it on a slow basis.

So I think we need some human assist, some AI-based assist (10:02) for the humans that are making these decisions. So it can, it’s similar to, I mean, I think it (10:07) parallels the automated driving sort of a thing, right? It parallels the autopilot thing. Pilots (10:13) can fly the plane 100% of the time, but it’s better for them if they have autopilot and they (10:19) can let the robot drive it for a little while.

Driving a car is great, but wouldn’t it be better (10:25) if you had a little AI-based thing that could help you drive it? We’re seeing that, right? (10:30) Same thing with managing the grid, right? Think of the grid as a vehicle that’s got a bunch of (10:36) people that are driving it around. They need some help. And so do those of us who are the passengers (10:41) on the vehicle.

We need the vehicle to be driven more effectively. And I think introduction of (10:45) agentic AI into the management of the grid, I think is going to be a huge game changer. (10:52) So where do you think they should optimize it first? At the sourcing level, at the infrastructure (10:57) or delivery level, or just basically the load management part? So in your experience, (11:02) where do you think we should really first start the optimization process or where AI could best (11:07) help? Probably the lowest hanging fruit is load management, right? So optimizing or fine-grained (11:15) load management would probably be a lot more that’s AI-driven or that’s algorithmically driven (11:20) would probably be very helpful.

And then from there it can progress back up the chain. Our (11:24) problem is that we don’t have enough. In the United States, we’re not going to have enough (11:28) energy to support all this stuff.

And we don’t have the human capital to build generation facilities (11:33) fast enough, right? United States, we can’t compete with China in terms of human capital. (11:40) They can build, and in terms of regulation, right? They can build generation facilities faster than (11:45) we can because they have less governmental regulation and they have better human capital (11:50) and they have lower costs. We can’t compete with building generation facilities.

So we have to do (11:54) something different. You’ve managed IP a lot, right? So both under your academia, startups, (12:00) and kind of in that place. So how does AI kind of change? And I know I’m talking very broadly (12:06) when I say AI, but I hope you get the gist of what I’m saying.

It’s the meaning of intellectual (12:12) property itself, right? So is there any impact there on that side? We’ve seen, you know, (12:18) voices or sounds being used from popular singers to recreate, you’ve seen, you know, all of these (12:25) things. So I guess it does. I mean, yeah.

So I’d just like to understand what are your thoughts on (12:29) how it has changed that meaning of intellectual property. I think the artists who are protesting (12:35) against the use of their materials are the canary in the mine. I think they’re on the right track.

(12:42) I think that what’s going to happen is we’re going to end up with a collection of (12:46) haves and have-nots, and the have-nots are going to be pushed into dealing with AI that’s garbage (12:53) in, garbage out. And the haves are going to be dealing with an AI-based system that’s got (12:59) refined inputs, and so it has better outputs. I think that’s the main sort of social concern (13:04) about this.

And I think the artists who are complaining about this right now, whether it’s (13:09) text-based artists, whether it’s people who publish things or people who write songs or whatever, (13:14) I think they’re the canary in the mine because they’re the source of human innovation. (13:19) And their human innovation is being consumed without their knowledge, and I think it’s bad. (13:26) That’s where you start saying that, because previously you could not, I don’t know if you (13:30) can trademark or IP a voice or, you know, the way I sing or something.

That’s going to be an (13:36) interesting way that that’s happening. And let’s say an energy and infrastructure-heavy AI systems, (13:42) where does the IP actually live? Is it in the algorithm? Is it in the data? Is it in the overall (13:47) system design? Or is it primarily on, you know, the way I have this operational know-how of how (13:53) something works? I think it exists. Intellectual property exists at all stages of this pipeline.

(13:58) I think the critical piece is the human-created data that can be used by the machine for training (14:06) purposes and emulation purposes. I think protecting somehow what the machine is going to use as an (14:12) input. It’s a human output and the machine is going to use it as an input.

So having some sort (14:15) of protection at that point, I think is absolutely critical. That’s why I think the artists that are (14:20) up in arms about this, they’re exactly on point. Once the machine consumes all that data.

(14:26) Absolutely. So that kind of leads to my next question. What is more important or what matters (14:30) more in the sense that I own the IP or I have the computing power and the energy to kind of scale (14:36) it, right? So… Unfortunately, I think it’s going to be the latter.

The computing power and the (14:40) energy to scale it, right, is going to overwhelm the individual ownership. Yeah. Particularly in (14:46) the United States in today’s landscape where it’s the IT billionaires that seem to be (14:53) controlling everything.

Yes. You know, so that game without some sufficient regulation in that (14:59) particular area, I think that game has already been won. No, that’s true.

I mean, even if you look (15:05) at the wars, the AI wars today, even though I feel that open AI won the initial battle, maybe, (15:12) but they haven’t won the war, right? So I have a very strong feeling the war is going to be won by (15:17) Google, let’s say, because they own the entire pipeline. They own everything from the start to (15:22) the end. They are completely vertically integrated.

They have the power to scale things. And I guess (15:29) that’s where, you know, it’s really going to be taking place. Yeah.

Google’s broad-based vertical (15:36) integration is absolutely fascinating. Yeah. So they would be one of the survivors of this for (15:41) sure.

Absolutely. Absolutely. Maybe Microsoft does as well, but let’s see.

It’s going to be (15:48) challenging. It’s going to be a tough world for the little guys, unless the little guys get eaten. (15:53) Yeah.

And little guys, yeah, I mean, they’ll just start… They’ve already started doing that, (15:58) so they started already adopting, you know, buying those smaller startups and everybody who’s (16:04) specialized in a niche, they can go out and buy it and simply… And we saw Facebook doing that (16:08) with Instagram and WhatsApp and all of those companies. So the same is going to happen over (16:12) here as well. So which is why at times, you know, you remember my first question, whether it’s (16:17) something new or it’s deja vu, I really feel at times it’s just a deja vu.

It’s just the (16:21) circumstances are the same. The scenarios are more or less simple, are similar. It’s probably the (16:27) technology that has changed, right? So it was something else before, now it’s something else, (16:31) but the playbook remains more or less the same.

That’s exactly right. Because at the heart of the (16:36) playbook is humans and what humans need to do. Yeah.

Yeah. Yeah. Absolutely.

And so they’re (16:40) the ones that are turning the pages in the playbook. Yeah. Yeah.

Absolutely. Absolutely. (16:44) I think, yeah.

So I agree. Once Energy Compute and whoever holds this power, that already… (16:50) that gives them a built-in advantage, you know, to acquire the smaller teams or trying to build (16:55) and protect their IPs and it’s… but eventually they will always have that advantage. A couple (17:00) of things.

I mean, we’ve been seeing that AI systems can be misused as well, right? So, (17:05) because the power of the brute force is kind of back. And eventually if, you know, quantum (17:11) computing comes into play, you’ll have this brute force much more stronger. I don’t know if you’ve (17:17) heard about the data leak from Anthropic.

And as more and more of these models are trained, (17:22) they’re hosted by third parties. How should organizations kind of protect themselves, (17:28) both from IP perspective, as well as just from a security or a cybersecurity perspective? (17:33) Yeah. I think this… we’re going to end up again in haves and have nots, right? Cloud computing (17:38) for the masses and my own data center, if I can afford it.

That way, if I have my own data center, (17:45) then I can keep my data private. And at the same time, I can welcome your data in. (17:51) Yeah.

So, we work with a lot of clients. I don’t know if you’ve… I mean, and they ask us, (17:55) how do we keep our data protected? We tell them there are two ways. We can turn that button off (18:00) in OpenAI, which says, do not use my data for training my models, you know, and just turn it (18:05) off.

Yeah, that works. (18:08) And the other is that I use these open source LLMs, you know, kind of deploy them on my own (18:14) server, post it on-prem and run it. I’m not sure if either of these options will ever work because, (18:19) we already use AWS, we use GCP, we use all of these.

Supposedly, our data is protected over (18:26) there. I mean, it’s the same as the off button, right? So, it’s the same as the off button in (18:30) OpenAI. The only difference is that we don’t have to press any button there or choose an option.

(18:35) They just say that, hey, that is protected. And then you have these open source LLMs, (18:39) they’re constantly being evolved. You don’t know who’s updating them, where their code is going, (18:43) you know? So, I’m not really sure if that can really protect everything else.

But (18:48) organizations do need a way out, right? So, because what I feel is, like you said earlier, (18:53) it’s garbage in, garbage out. And AI or for these LLMs to make sense at a lot of places, (18:59) unless and until you give them good data, they cannot produce anything, right? So, (19:04) the orgs still have to find a way to figure out how to protect your IP. What can be one way of (19:10) doing it or what could be, what kind of solutions they can work with? You can adopt the Gen Z, (19:15) is it the Gen Z version of this, which is, I don’t care, it’s all out there anyway, just let it go, (19:20) right? That’s a good approach.

Yeah, that’s an approach. The other approach is some sort of (19:27) legal recourse against whoever you’re working with, right? That’s one approach, but that’s (19:34) very expensive to prosecute these guys. It’s the same approach that you would use with any cloud (19:38) servers, right? So, you were using any cloud servers, you were having data there, you’ll do (19:42) the same over here.

Yeah. You never know whether, I remember there was this challenge on Facebook, (19:47) you’ve seen it, but I think it was a couple of years back where they were collecting data on (19:51) how would I look in 10 years from now, right? So, something on those lines. How did I look 10 years (19:56) back? Sorry.

And people would update and upload this photo, which tells them that this is how (20:00) I looked 10 years back, this is how I look today. I really, really thought that was to train models (20:04) to understand how somebody would look from 10 years from now and kind of predict that. (20:08) Maybe.

Right? So, that still happens. That’s still my IP, but it’s being used. (20:16) Is it? Is it really? Are you sure about that? Yeah, absolutely.

So, all right. I think one of the (20:24) things that we’re going to see, we’ve already seen this with large organizations, but there’s got to (20:28) be some level of service level agreement. There used to be, when I was working in the (20:31) telecom industry, there were service level agreements that had to do with how the service (20:35) was maintained and what sort of service I got out of it.

The service level agreements, as they (20:42) exposed in terms of data, right now are done by a EULA, right? End User License Agreement, (20:48) where you give everything away. I think there needs to be some sort of service level agreement (20:52) on the data. That’s a more refined version of an End User License Agreement where it says, (20:57) I own this data, you can’t have it.

And if you use this data, you owe me money and you can have (21:01) this data. I don’t think anybody in their right mind would go through all that because it’s a (21:05) really difficult. But there’s something like that that needs to happen.

It says, (21:09) this is my data and you can’t use it. It’s very sort of HIPAA-like in terms of medical data. (21:15) Yeah.

And government regulations are also coming in strong. You see GDPR that’s being launched in (21:21) the Europe side. So, that’s a very good step.

You see organizations now complying to cookie (21:25) consents and that being enforced through browsers. So, that’s also good as well. You see a lot of (21:30) competition commission controls coming into action where they are coming against organizations who (21:34) are kind of creating monopolies in terms of data and everything and making them up.

And that’s also (21:38) in the right direction. So, I feel there are things that are being done. There are things that (21:43) everybody is also learning.

So, specific laws are now being introduced against how to use AI, (21:50) how not to use AI, what AI can use, cannot use. But it’s an effort at least. And those kinds of (21:57) things are happening.

So, I guess there’s a general understanding that has evolved over time. And Gen (22:03) Z, funny you mentioned that. I have worked with a lot of Gen Zs.

And their strength is in questioning (22:11) a lot, right? So, I feel every generation has their own advantage. So, boomers would have their own (22:18) advantage, the Xs, the millennials. I think Gen Z will bring their own advantage.

At one time, (22:23) it could be a parent. Right now, I don’t know where it is. But yeah.

Having had several children (22:30) and watching what happens to them as they become teenagers, they develop their own language. Because (22:36) they know innately that there are some things that need to be kept private. And they develop (22:42) a language themselves.

And they communicate in a private way. It started off with cell phones. It (22:48) started off being special characters or special sequences of characters.

And then it turned into (22:53) emoticon sequences. The young generation, they know innately that they have to keep something (22:58) private. They just have to figure out how to use the technology.

(23:01) How do you do that? (23:02) Yeah. That’s true. That’s true.

All right. So, coming back to a slightly bigger picture (23:07) with policy and strategy, we’re seeing a lot of AI-driven data center demands, (23:12) pushing electricity usage, as we talked about. And yet, when it comes to regulation, (23:16) most of the focus is still on AI models and use cases and not on the energy systems part of it.

(23:24) Is that a strategic blind spot that’s being created? (23:27) Yeah. I think going back to what I was saying earlier, I think an initial focus of AI (23:34) needs to be on how power is managed. And how power has to be managed in a fair way.

(23:41) And left to their own devices, the tech behemoths, they won’t care. They’ll just (23:46) do it for their own selves. That’s just human nature, right? So, there has to be some regulatory (23:50) input on how the power is managed and what power can be used for who and in what fashion.

And I (23:56) think the evolution of agentic AI in the management of energy, I think, is maybe a (24:02) significant contributor to that. So, you worked with NASA and national (24:06) programs. How do public institutions balance openness, IP protection, along with energy (24:12) efficiency, if that makes sense, in AI-driven systems? (24:15) Yeah.

That’s a tough one. That’s a tough one. Managing the privacy of data, especially in (24:20) academic research programs, what does a university do with the data? Do they put it in AWS or they (24:25) put it in a Google Cloud? Because you lose control of it at that point when you do that.

(24:30) We had a really serious issue not too long ago with Dropbox. The users at the university, (24:35) they needed to share data really fast with other people at other universities. So, (24:39) they started using Dropbox.

And when they started using Dropbox, and Dropbox is a great service, (24:44) but when they started using Dropbox, what they did was they circumvented a bunch of (24:48) policy that was really important, keeping control of the data. And so, you have this kind of (24:53) interplay between what the users need to do and how fast they need to get it done, (24:58) and what the institution can keep up with. And I think that’s a really big problem, (25:05) particularly as these services roll out and become more and more attractive, and they (25:09) solve more problems.

But the regulation is always lagging. And I think one of the reasons the (25:14) regulation is always lagging is because we have ancient politicians. (25:18) Yes, that’s true.

You need Gen Z in politicians as well, right? (25:21) Absolutely. (25:23) And I think this is the time right now when the old age politicians are being kind of, (25:29) this is a transitionary phase that I see across the world. If you look up US, if you look up UK, (25:34) if you look up Europe in general, if you look up the South Asian countries, everywhere, (25:38) you will have boomers and, you know, 70s plus guys leading politics.

And then the next generation (25:45) that is coming is in late 30s, early 40s, those kind of age gaps. We see that in the US as well, (25:52) Zohran Mamdani being an example, and you from New York, and then all of those guys, (25:56) I think that transition, once that takes place, then your regulations will automatically move (26:01) faster as well, because they would much, much better, hopefully, understand the challenges (26:06) that are coming and why regulations and policies need to adopt to this new world. (26:10) You get a sort of a collapse between generations, right? Because a generation, (26:15) when technology has progressed beyond a person’s generational capability to really understand it, (26:22) and then you have a secondary generation that’s grown up with it, there’s a collapse between (26:26) there and the old ones can’t keep up.

And that’s why the new ones have to take better control of (26:34) this. And for that to happen, the old ones have to let go. And that’s the hardest part, I think.

(26:38) That’s true. That’s true. That’s true.

That’s true across all, (26:42) I think, all countries and everywhere. It’s a general human behavior, right? So absolutely. (26:47) All right.

So I was reading a couple of your things and I read something related to where (26:52) you described major infrastructure threats with the phrase, forewarned is forearmed. And from your (26:58) experience with these infrastructure programs, what does being truly forearmed kind of look like (27:06) today? I think the politicians have to be forewarned and they have to have the ability to (27:10) keep up with it. And they have to have the ability to create regulations that matter, (27:14) not regulations that pertain only to certain slices of people that give them money.

I think (27:19) that’s a big issue. The younger generation, I think, is able to be forewarned and forearmed (27:24) because they’ve grown up with these technologies and they kind of understand natively which parts (27:28) they can slough and which parts need to be protected, particularly of their data. I don’t (27:33) think we have enough data protections in place to keep large scale AI from consuming all of our (27:40) data.

So we have to figure out ways to influence the politicians to create regulations that matter (27:46) for the data that’s going to be consumed by these AI machines. Because it’s going to happen. It’s (27:51) going to happen.

It’s already happening. We just have to prevent the damage, right? Some of it’s (27:55) being done for good. Some of it’s very, very useful, right? Things like chat GPT and stuff (28:00) like that, they can be incredibly powerful tools, but some constraints have to be put on them.

And (28:05) I don’t think the, I don’t think yesterday’s politicians can get their brains around what (28:10) really needs to happen. You’ve spent much of your career, you know, teaching and mentoring engineers. (28:14) And as AI becomes more coupled with infrastructure, IP, our day-to-day usage and everything, (28:20) what do you think is one of the, or two of the most critical things that we should, that teachers (28:25) or everybody should kind of teach the next generation of technologists? Today, when people (28:29) talk about AI, they’re primarily talking about things like ChatGPT and large language models (28:34) because those are the things that ingress on your personal space more.

So you’re more aware of those. (28:40) I think AI has been, things, AI type things have been happening for decades in hidden areas. And (28:45) they’ve been automating technology processes that we’re not even aware of.

I mean, one of the reasons (28:49) that we have really fast Wi-Fi today could be classified as a form of artificial intelligence, (28:56) right? It’s a, it’s an algorithmic adaptation of what’s going on to try to optimize how things (29:01) communicate with each other. That’s a form of AI in a loose sense. So AI is all around us.

It’s been (29:07) happening for decades in different pieces. Just the fact that we notice it now because of the (29:12) interface that we have with large language models is a symptom of a much larger technological (29:18) progression. And I think we have to embrace that with ways that, where we understand how our data (29:23) is being used.

I think a lot of this stuff is so complicated that it’s beyond normal people’s (29:29) understanding. And so it’s kind of incumbent on these technology companies to not abuse that (29:33) position of power. I think it’s incumbent on these technology companies, this huge technology, (29:38) to be kind with, or upfront with, or more contractual with, more positive acknowledgment (29:46) with how they’re going to use your data.

Because there’s a lot of people that aren’t going to care and just sign their life away and tough for them (29:50). But there’s other people that really are going to care and the people who care and the people who are cognizant of how their data is getting used and are aware of what needs to be kept safe and and are taking steps to do that (30:00). 

Those are the ones that are going to be protected and those are the ones that are going to leverage it more effectively. No that’s true I use this facility I think that’s by Facebook or by Amazon or pretty much everybody where they if you ask them to email me what data they have on me and they send you this email after a day or so (30:30). And it was about 300 400 MB of data it was all text so 300 400 MB in text is like huge and and I trying to decrypt it understand uh you know I spent about an hour I’m a CS grad and everything and I understand how the data is broken down and and all those things but still and I was amazed and fascinated and I was also kind of had a little fear that look how much they know me you know (31:00). 

And uh the amount of the way they’ve organized this the way they’ve put it together no wonder Amazon knows before me what do I need next in my shopping list next. Here here’s a fun thing to try to do try to delete your Facebook account okay (31:30). Try try to delete your Facebook account it’s really hard it’s really hard yeah yeah that’s not right that’s not right right Facebook from that perspective Facebook is not dealing above board with its consumers even though it’s providing a free service it makes it really hard to delete your Facebook account (32:00). 

Yeah it’s trying yeah even unsubscribing from an Amazon service is hard they they ask you like five times and then you it’s convoluted to find the place where cancel subscription button is and that’s that’s true I’ve never tried deleting my Facebook profile I’ve completely went silent on it but I’ll look into that just so that you you know good luck (32:30). All right Stan thank you very much I think this was a good conversation I hope you enjoyed it as well it’s really been a pleasure having you at Techn and uh we kind of appreciate your time and your kind yeah appreciate the opportunity and it’s been a fun talk (33:00).