The Far Reaching Impact of AI – Techstrong AI Podcast EP21

Hello and welcome to the Techstrong AI Podcast. I'm Amanda Ani. With me today is Mike Baard.

How are you doing? I'm doing great. How are you?

Doing well. We have a lot of stories in the AI world today. Some very interesting ones.

We're gonna start with AI election fears, and there's an article over on Textron ai written by Frank Baard, and it's about, it's about, uh, AI being used to influence elections. And it's, it's a mixed, it's a mixed review from different countries depending on which country you're in. For example, India seems to embrace AI being used for messaging, even us utilizing deep fakes of deceased individuals to spread that message.

And it seems to be met positively. Uh, other countries, not so much. So what are your thoughts?

There you go. Well, first of all, Frank Ard is otherwise known as the brother follows all this stuff pretty closely. Um, I think it's, we may be a little bit ahead of the curve and we'll see because, um, we have this election cycle here in the States and the debates are now starting up, and we'll have some of those in the next few days.

And, uh, maybe, yeah, we'll start to see more deep fakes in the wake of that. But right now, I feel like a, a lot of this stuff is still early days. We haven't quite mastered it, or at least the bad guys haven't mastered it.

The Russians are clearly playing around with it. But, um, it may just be a matter of timing in the sense that I feel like the bad guys are no better than the good guys at experimenting with this, and we're all kind of just stumbling our way through this stuff. And, you know, eventually people will hone their skills and get better at doing whatever it is that they're gonna do.

And the good guys, who knows, you know, they might create deep fakes to advance their causes or to counter the various misinformation campaigns that people may be launching around some of these videos. Um, so I, I'm not quite a hundred percent certain that our deepest fears are gonna come to life, but it's still early. But I think, uh, I'm hopeful at least that maybe people will get smarter about looking at this stuff and art.

And my soul just says the more we inundate people with false information, eventually they just decide to distrust everything. And until something is proven to be true, they'll just disregard it. And that's not a bad way to be.

Yes. Well, you know, the other side of this coin that I find very interesting indeed, is we have another article over on Textron ai, about an AI avatar actually running for office at the UK Parliament. And, you know, supposedly it's gonna be able to analyze the data quickly from, you know, everyone's, uh, what their vote is and make decisions based on the majority vote of the population.

So I'm interested to see how that plays out and if he actually, this AI avatar gets voted into to Parliament. Well, it's an, I'm not sure that, uh, a non person can hold the seat, but, um, it might be in advance in the sense that at least this thing is looking at some data to make some decisions. Uh, I'm not sure if you can instantly poll your constituency on every issue.

And ultimately we are electing someone to make decisions on our behalf, but maybe there'll come a day where we can all vote on every stupid little thing that comes down the pike. But I think I got other things to do. Um, but I sure wish that whatever we do going forward, and we do look more at the data, if I look back in time, most politicians that I know barely read some of the bills that they're voting on.

They don't have a lot of insights into what's going on in terms of why they were created. And they certainly, uh, only look at, uh, a small portion of the data. And some of that data is presented to them by people that are clearly lobbyists.

And a lot of these bills are written by lobbyists using data of uncertain provenance, shall we say. So maybe we can have an AI agent that will just be a legislative aide that works from the politician a little bit and says, you know, here, you know, the actual facts in the manner. Now, you know, again, the AI is only as good as the data it's exposed to.

So ultimately it may just come down to like making sure we had enough data shared around for everybody to make an intelligent decision. But hey, can it be any worse? Right.

It it'd be nice, you know, if the population feels like, you know, they are, the majority is being heard and, and maybe AI can be a useful tool for that. Mm-Hmm. So moving on, this is a, a another use case for ai, and this is by our writer Brian.

And it's talking about how the movie industry is being impacted. Um, you know, there's a lot of, um, voiceover technology, um, that's being used with AI and it's matching different languages up with the lips and sinking it perfectly with the lips, the inflection, the tone, uh, really matching it up nicely, um, and more efficiently with AI instead of, um, for example, um, dubbing over it and it doesn't match up and it's really stilted looking or, you know, just having say the words, um, following along. So what are you seeing in this regard?

Well, first of all, I always enjoyed those Japanese Godzilla movies as a kid when the dubbing wasn't quite synced up anyway. I mean, that was part of the charm I thought, but that was just me. Um, I think This is going back to the writers and directors strike and all that stuff that happened last year, and some of that stuff was apparently settled and released to somebody's satisfaction.

But the truth of the matter is it's here. And a lot of that work probably isn't gonna be there in the future. 'cause the studios are just gonna take the, uh, a voice of someone or maybe even an amalgamated voice of several people into something to create the voiceover and do that thing.

I remember there was a fellow, I think if you remember going to the movies and there was always kind of the voiceover for the reviews of the movies. And, you know, there'd be some guy who would always say, you know, and in a world where, and blah, blah, blah, blah, blah, blah, blah, you know, that was a person and that was their job. And apparently they were used very often and then they passed away a few years back.

So who knows, maybe we'll be bringing this guy's voice back to do all that. But yeah, sometimes the tech just advances and it will change jobs and careers and things that people do. I sure hope the people whose voices are being used to drive things get compensated for it.

I mean, uh, there shouldn't be this situation where they're just willy-nilly taking voices from folks that that's, you know, it's part of your personality, it's part of your person, um, that just can't be randomly used whenever, um, at least not beyond a normal sampling rate per se. And, uh, we've seen some actors and actresses get caught up in this conversation already and more to come. But I think reason and lawyers will prevail.

Yeah, I think most recently there was something about Scarlet Johansson's. I think her voice being, uh, she was suing 'cause she thought her voice was being utilized, um, in something. I can't remember.

But yeah, it's, it's certainly an issue. And, and I remember too, what you're speaking of, and some of those voices are so iconic and you recognize them right off the bat. They, they hold a lot of weight when you hear them and, and they're the ones voicing these shows.

So if they are using AI in this way, yes, hopefully there's compensation there. Right now, the New York Yankees at a legendary announcer who passed away, I don't know, maybe almost two decades now, I'm not sure, but at least a decade or more. And, uh, they used to record his voice for, uh, Derek Jeter's introduction when he came to the plate.

And it was the only lady, everybody else just had a normal announcer. But Derek Jeter wanted this guy's replay of his voice being introduced used throughout the latter half of his career, even though the fellow had passed on many years earlier. Um, so is that any different than what we're currently talking about?

I don't know. It seems like it's, uh, similar. I'm sure the Yankees compensated, uh, fellas family who did the original introductions, but he was iconic, did that job for, I don't know, three decades.

Uh, maybe we can just bring his voice back for everybody now. That would be good. Good.

Well, I'll say I do like this new voiceover technology concept because it is a little distracting if you have to read subtitles when you're watching a movie. And there's some really good movies that I've seen where I had to spend half my time reading instead of watching what's happening. So you miss, you miss a little bit.

Well, you bring up a whole other issue. I saw some survey that said half the people are watching movies with subtitles on it anyway, because, um, they're hearing a shot and that's how they gotta watch television these days. So I Yes, that, that's very true too.

Okay, so moving on the next one, this is how restaurants are trying to get in on the AI scene. And we have McDonald's who recently tried to use, uh, they partnered up with IBM and they did a little test run using AI ordering systems and they didn't have great luck with it. Um, the AI just wasn't quite understanding orders correctly.

And there was some, um, pretty big mess ups, like I think one of them was bacon and ice cream, although actually that sounds like it might kind of be good. I don't know, strangely sounds like it could be good, but you know, butter instead of caramel and all kinds of weird mess ups on orders that were getting, uh, you know, put on social media. And so they pulled all those AI ordering machines out.

Um, but we do have other, I think Taco Johns, I think was one of them that's also gonna be utilizing some ai uh, testing later on. Um, so what are your thoughts on this? Well, this stuff is coming for sure.

And part of the issue, frankly, is just that these fast food restaurants can hire enough people who want to take those jobs, right? Um, and you know, I seem to remember working in one of those in college many, many years ago. It's not a lot of fun.

It's just kind of like, uh, you know, it's really greasy back there, let's put it that way. Um, so in still early days though, and I got a feeling that sometimes we test things at a broader level of scale than maybe we should have. So maybe this was a proof of concept that got a little bit out of the box too soon.

I would also note that these restaurants, when they put it in these, uh, kiosk systems in whatever else, um, they're not great experiences to begin with, even before ai. I just feel like they don't really think it through in terms of how these things are gonna be used by normal folks. And, um, half the people I notice when you go into those restaurants, half the people avoid those kiosk and want to go to stock to the human cashier because they can't figure out the menu flow and what they're supposed to order and then just kind of throw up their hands.

And seems to me the older they are, the less likely they are to want to use that kiosk or engage with the machine. 'cause sometimes it's just simpler to look at somebody and say, I would like a Big Mac. I don't need $10 million worth of technology to handle.

Yeah. Uh, I mean, here's my opinion on it. It is not easier.

It is not a better customer experience. I have gone into several of these restaurants. I have tried to utilize the little digital touch pad and order instead of talking to somebody.

First of all, it's weird because there's literally people standing there at the counter that you could go over and talk to. But instead you have your bag turned to them and you're sitting there spending like 20 minutes going through all the different choices and adding in and subtracting everything till you have your order just right and then submitting it. It is utterly ridiculous.

It takes way longer. It's harder, and it's awkward. Especially, I think I went into a Wendy's the other day and that ordering pad was literally on the counter where people actually are working there at the counter, but you are supposed to enter.

It was just weird. Okay. Yeah.

And, and what's even more troubling still is the line behind you is actually longer than it was if you were talking to a human. Yes, yes. It's very strange and nobody knows what to do, whether to use that, the ordering tablet or go talk to a person.

So everybody's just scrambling and looking all over the place. Where do they go? It's just not working.

All right. We're telling the restaurant industry, go back to the drawing board. It's not worth it.

Now, what I do think is pretty cool is the little AI robots that they're starting to use in certain restaurants to, you know, go around and deliver condiments or, you know, see if you need anything extra and things like that, that that's useful and probably efficient. Yeah, I see those robots on college campuses delivering pizza all the time. And yeah, those, the kids seem to really like it a lot.

Uh, then again, most of them haven't left their room in three days either, so who knows. Yeah. But those are cool.

Uh, I could see that going somewhere. Okay. So next is, um, oh yes.

Um, so who at Packard launched a managed AI infrastructure service and you were covering that. Can you share a little bit more? Yeah, I mean, this is an interesting idea in the sense that it's a managed service that will stand up the infrastructure.

You need to roll out your AI models or build or train or whatever you want to do. Um, largely designed for on-premise environments. So it's a, it's really a managed IT service, but HPE will set up the infrastructure including getting all the GPUs and provisioning all that stuff for you, and then manage that remotely.

And I think it's gonna be an interesting question. This concept's been around for a while, but with the rise of ai, I think a lot of organizations are gonna take a moment and do a gut check and say, to what degree do I really wanna manage infrastructure anymore? Do I or do I wanna just rely on a service?

Which is really no different than a cloud computing service. And you know, as long as I can figure out the math on the economics of that. But I think when a lot of organizations look at the total cost of this next round of infrastructure, which is gonna be expensive, uh, GPUs are just not cheap.

Plus the, all the people required to manage all of that, more people are gonna decide, Hey, you know what, just rather take what resources we have and use it to build software and do the fun stuff. 'cause the underlying infrastructure stuff, you know, it's a means to an end. Yes.

And I, you're absolutely right. You, you only have so much manpower and so much time. So how are you most efficient?

Yeah, we'll see, I know there's a lot of folks out there that make their living managing servers and infrastructure and working through all these data centers, but, uh, yeah, I don't think those jobs are going away. I just think that the people they work for are gonna be different. They're gonna be more like HPE or Dell or wherever else is it gonna be?

Mm-Hmm. Absolutely. com and this is about mind io adding tools to detect the presence of generative ai.

And then on Textron ai, we have an interview with Peter McKee, uh, talking again about, um, clean code and, uh, using, uh, tools to make sure code is clean. And so what are your thoughts about the accuracy of that Mm-Hmm. And what are you hearing from business leaders?

So all of this is gonna come to a head real soon. I think what's going on right now is developers who are hard pressed for time are using gen AI tools to help them write code. And they're doing it increasingly so, and they're not always testing it, right?

They're really testing it and they're not much inclined to actually won worry about if there's vulnerabilities in there. And all this code is now starting to move through these DevOps pipelines and finding its way into production environments eventually. And we could have a major problem down the road when it turns out that that code is either deeply flawed in the sense that a, it may be very expensive to run 'cause it's inefficient and worst case of all, it may have a vulnerability that is known to everybody but just didn't get caught.

And, uh, the bad guys are gonna exploit it and it is gonna be more trouble. And it's worth in the sense that we'll spend more time troubleshooting all this code that we're standing around going, isn't this an awesome gain in productivity? But it's not a gain in productivity if we wind up spending an inordinate amount of time cleaning it up later.

And yeah, especially if there's a security issue that winds up being more costly than anything we would've potentially saved in the development side of this equation, I think we need to have, you know, an honest conversation about what's going on with this stuff and, uh, and, and how we're gonna use it and how we're gonna review it. I'm not saying we shouldn't use it, but I am saying that, um, our existing processes and workflows for how we manage software supply chains are not up to this task. And so it's time to kind of sit down and the first step is what men was talking about, which is just identifying what code was actually written by a machine.

And then the second step is what your interview was talking about, is just making sure that the code is actually clean as it should be. Um, putting more, uh, substandard code into production environments is a step backwards, not forward. Yeah, absolutely.

And, uh, we do have to find a solution and of course the humans are not gonna just be able to, to cover all of it. They are gonna have to use tools and AI is probably going to be the solution, um, to augment this process. But, you know, it, it's gonna be a, a process.

Yeah, we're definitely gonna use AI tools to, um, investigate the AI code, right? There's just no ways around that. But yeah, right now we have more tools for the creating of the AI code, then we have AI tools for managing that process.

So I think we need to kind of like take a breath here and say, wait a minute, when exactly are our pipelines gonna be AI enabled enough to reliably manage all this code moving through these pipelines generated with the help of AI models? And you could argue that that cart is way in front of the horse here. Yep.

Well, that brings us to the end of our list. So do you have any final thoughts for our audience today? I would just say, you know, AI is gonna be, it's gonna change our lives.

There's no aspect of it that, uh, will not be impacted by it, but, um, we need to remember who's in charge of all this stuff and ultimately who's gonna be responsible for it when it goes wrong. That's the truth. Well, until next week, thank you for tuning in.

com to read those full articles in detail and make sure to share with us what you're enjoying reading about and what you'd like to hear more about. And have a great day. I'll see you next.