AI Hype, Data Hoarding and the Impact of AI on Children – Techstrong AI Podcast EP25

Hello and welcome to the Techstrong AI Podcast. I'm Amanda Ani, and with me today is Mike Baard. How are you today?

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

We have a large lineup of articles to share with you today. So we're going to dive right in with this one is about the Gartner Hype Cycle. It's very interesting.

Um, I'm gonna get to read a little bit from the article. The general public's enthusiasm with AI can be measured through the Gartner Hype Cycle, which is a chart that graphs the life lifecycle of innovation. This chart bears a strong resemblance to a rollercoaster, and we are currently riding along the first and the tallest inline when it comes to ai.

And then we have a couple other articles on Techstrong AI that talk about how businesses are really trying to jump on and harness AI technology for a variety of different solutions. So what are your thoughts? Well, this may all come to a surprise to anybody who's outside of it, but, um, AI's challenging, nimble it and the hype cycle has been outta control for some time now, in the sense that, you know, it's been awesome to see what Microsoft has done and open AI and Google and all these other folks.

I don't, they spent an inordinate amount of time and money just training AI models and hiring people to, you know, tell the AI model that this is a cat and that's a dog and it took forever. Um, and you can't just replicate that as a business overnight. Um, they don't have the skills, they don't have the expertise, they don't have the technology.

The data's driven all over the place. So, um, it's, it, it may be a surprise to investors, I guess, or maybe it's a surprise to end users, but you know, the average company isn't gonna be able to harness all this stuff so easily. And what we're seeing that play out now is, you know, this hype cycle thing.

Gartner has been playing this, you know, hype cycle thing. Now I'm, for as long as I can remember, um, I think the hype cycle existed long before Gartner discovered it, but that's a whole other issue. Um, the point being though that, um, investors who kind of, uh, made some assumptions and, and, and all the folks who covered this stuff, especially out of the financial media, did never seem to talk to anybody at it about what all this stuff is and how it works and what the implications are.

And there was all this assumption that, uh, this stuff would just magically exist and we would invoke it because, you know, Microsoft made it available through a easy to use interface. But, you know, go look at what they're letting you do, it's not gonna change your world. Right.

I can write an email better now. Well, thank you very much. Um, and I can do some interesting workflow things here and I can, can create a summary of a meeting known and that's all great and fun too.

But does it change the GDP equation? Hardly. And I think that, um, we'll see in next wave of this and probably the hypes I go will kick up again and maybe this in the case of AI is gonna be that rollercoaster phenomenon where we go up and down, up and down and up and down.

The next big thing are domain specific language models, otherwise also known as small language models that are trained for specific tasks. Uh, and then those LMS will be used to create agents that will perform those tasks. The agents will be AI agents and they'll be trained to do something specific and helpful and useful.

And then we humans will figure out how to orchestrate these agents to perform various tasks across our workflow. And you'll be more like, uh, you know, we'll be conductors of a symphony as it were with all these things that are gonna come together. But I just don't see that happening, you know, pervasively this year and probably well into next, but I mean, it will happen, but it's just, uh, as somebody once said to me when I first got into this business, never mistake a clear view for a short distance.

Yeah, I think AI is, of course, it's a fantastic technology. It's exciting what it holds for the future, but I I do see that business leaders are placing a little bit too much faith in AI for too much. Yeah.

And I sometimes just wonder where all that comes from because business leaders haven't been investing in tech forever and day, and I'm a huge fan of tech, don't get me wrong. But it's not like tech has moved the productivity curve substantially over the last two decades and people are not that much more productive or companies are not, uh, by any metric that people seem to be tracking. I mean, maybe AI will be different, but these changes are incremental.

And to the degree that most of them may make the employee more efficient, it doesn't always translate into more profit for the company if the employee is more efficient just because, uh, there may not be enough volume of work for that employee. There may also be other things that employee needs to do in their life besides work. Yeah, exactly.

So then moving on, we're gonna talk about some news from AWS. You can find this on techstrong AI and techstrong, ITSM, that's our newest site and we haven't talked about that one as much, so check that one out. But can you share some of the new tools that they've unveiled over at AWS?

Yeah, they're up to something that's pretty clever, all in all. So, um, it just puts a spotlight though, on what is the future of application development and yeah, I wish there had been a little more clear about what they were trying to describe when they had this event in New York. But basically there's two types of applications now that can be built using, uh, a natural language interface that's invoking some type of LLM.

And the first is what they're calling personal apps, right? Which is, um, you know, what they're saying is you can put into a prompt a couple of, or maybe one or two prompts and uh, this new, uh, app Amazon service will create that app for you based on what it sees the prompt. So you now have your own personal app and they were pointing out that it would probably be quicker for you to create a personal app and then go find an app that did what you wanted it to do and then learn how to do it.

And there's probably some truth in that. Um, the second thing that was pretty interesting too is they were also talking about giving it teams that don't typically have a lot of programming expertise, access to a low code tool that you would access through a natural language interface that would allow them to build applications. And you know, historically those teams have tended to build apps using low code tools, using, I don't know, everything from, you know, some sort of spreadsheet interface to a small database to Lotus Notes back in the day.

Uh, and these apps were extensions of essentially an IT service management framework and they were used to, uh, automate a particular workflow. They weren't really designed to scale all that high. Sometimes they did, and a lot of the times they weren't very pretty, but we're gonna have more of these apps than ever.

And that on the face it sounds good, but on the other end of it, this could easily wind up being too much of a good thing. Uh, if everybody in his brothers build an applications and we have more applications built in the next two years than we have in the last decade, I don't think we're quite prepared to manage that onslaught. I don't think people are gonna be able to absorb it.

And so a lot of this stuff may just wind up, uh, going nowhere. And it's not quite clear to me, like if I build my own personal app and you got your own personal app, how regards who app's gonna like interact with each other and maybe do something. There's a lot of unanswered questions here.

So I think, uh, you, one, the way we think about building software, it's fundamentally changing. Two, is I don't think we understand exactly what the implications of that are. Yes.

And then it, it brings up the question, do we really need all these personalized apps for certain, for certain things, yes. For other things, no, because as you said, the app sprawl is gonna be tremendous. Yeah.

And then what happens is, like my phone, which is a mess, I have more apps that have downloaded, installed and used for about a hot minute and yeah. But I'm can't be bothered deleting them. And I'm sure that people who made them who are thinking that I'm are hoping I might return it as an active user someday.

But I feel like, you know, if I can create an app on demand, uh, I'm an itch that I'm gonna scratch immediately and then I'm just gonna forget about. Yes. So I guess we'll be hearing more about that and how they handle that moving forward.

Next we're gonna talk about data. Of course, we're always talking about data on this show, and this is about data hoardings. So I'm gonna read, um, from this article, according to a recent study, all respondents who are adopting or plan to adopt gen AI have encountered challenges with 49% in the US claiming the quality of data is at the top of the list.

It's nearly impossible for businesses to use AI to gain actionable insights from outdated data. The bottom line organizations must prioritize organizing and classifying their data from its creation and initial storage to when it becomes obsolete and in need of permanent removal. So what are your thoughts?

Well, we've been talking about the fact that data management in most enterprises is a mess. And we've kind of allowed that to happen because each department created their own applications and then wound up describing simple things like company names that our customers differently in those applications. And they never really, we talked about, you know, cleaning all that up and data quality and centralizing that for years, number of organizations that actually did it.

Not very high. And there's been, you know, some chief data officers appointed over the years and not quite clear to me how, which progress they made and then along cave ai and everybody woke up one morning and said, my God, we gotta like get some value outta this amazing AI stuff. Trouble is, AI is always good as the data that it gets shown and still garbage in, garbage out.

And there's just some massive amount of garbage in enterprise and entire IT environments, uh, whether it's, you know, sitting in an SAP application that conflicts with a Salesforce application. It's just rampant. And I think, you know, going back to our earlier conversation about Gartner, we're just not having an honest conversation about where we are.

Yeah. I think maybe it's a struggle too for them to determine what is outdated, unusable, unusable data versus the data that they should keep. Maybe they're just afraid as we see there's this issue with data hoarding, they're afraid to let go of the data.

Yeah. And again, why everybody's kind of terrified. 'cause they don't know what data might be valuable someday.

And they think that the algorithms maybe someday will, uh, uncover some massive truth that, um, a data science team is able to uncover because they were able to compare 22 years worth of data. I don't know, I think that maybe, um, the, the more recent the data is, the more valuable it is and just start cleaning up, say, I don't know, randomly pick the last six months worth of data just clean up from there. And then over time, if you wanna get ambitious, great.

But, um, I don't think we can afford to, um, clean up this entire mess that we've been making for three decades in three weeks. Yeah. It's gonna take a little bit longer I think.

Mm-Hmm. Okay. So next AI government governance.

Uh, this comes from tech strong ai. It says, when it comes to implementation and governance of ai, there is a three step starter program recommended, which is leveraging existing data classifications, inventory, proposed AI use cases, and then implement the low risk high reward efforts. And this is a helpful article on, you know, how to implement and then establish that governance.

So what are the, the struggles business leaders are facing when it comes to this? A lot of this will go back to, uh, audits. And ultimately I need to show some providence of the data that I use to train the AI model.

So when the outcome goes wonky, I can at least show that I had some responsible behavior. Um, hopefully, you know, this three step program will prevent you from signing up from the AI 12 step program because you have to stand up and explain to people, I got a little addicted to ai and the next thing you know, I applied it to everything without any thought processes to what the outputs would be. So it's really is about maturity and it comes down, uh, security of the data, the providence usage, uh, who's allowed to see what data, when some data is more sensitive than others.

And so not everybody needs to see, um, a social security number. If you're using that within the context of some solution, you gotta really think through who's allowed to see what and what impact that has on how the LL works. There's just a lot of nuance here around the governance side of this that, um, you know, if you think we're not doing much in the way of managing data, well then what makes you think we're doing anything to governance?

Yes. And of course we always talk about communication and it's gonna require good communication between all involved as well. Yeah.

And I, it's gonna require an investment in some new tools to go do that. And again, it adds to the total cost of the thing. And part of the issue is, well, companies don't have unlimited budgets.

They would love to be able to tell everybody, go play with ai, but then I wind up with a million proof of concepts. Well, that's interesting. Maybe, but eventually only two or three of those things probably you're gonna get the budget to go forward.

So start sorting that out now. Mm-Hmm. Okay, last but not least, this is a very important topic actually, and this is how AI is affecting our children.

So I'm gonna read the, this article. Children of all ages are gaining access to AI tools with many requiring little to no form of parental consent, even though there are many profound benefits to the wide variety of tools available. If we aren't aware of the many risks of these AI chatbots on children, we face immense risks, evolving data, privacy loss, cyber threats, and inappropriate content content.

So yes, this is a serious concern how AI is affecting our children. And what are your thoughts? I don't know what to do about it.

I know it exists and I know that it's probably only a matter of time before, you know, a lot of the kids, they like to be on these Discord servers with their buddies and chat about certain, you know, topics that they're all interested in, or at least they know it's a, it's a small group, but you know, then one of them starts interacting with a chatbot and starts asking you questions and they all giggle at the answer. And next thing you know, the chatbot's part of the conversation flow. And this will play out time and time again across every one of these websites.

In fact, um, there's some stories floating around out there on the journal today talking about how, um, X tried to news gr to summarize headlines as it related to, uh, the assassination attempt on former President Trump. Um, you know, and the headlines have generated were a mess. They were kinda all over the place, both from everything couples said, you know, Kamala Harris had been shot at others said, uh, you know, uh, to shooter was, you know, some sort of a, you know, democratically inspired assailant.

Um, and this stuff's all not helpful and it's, you know, it's, it's accidental disinformation at the end of the day, but it gets into the systems and, and it's not clear to me that kids, you know, are in a position to understand how to process that or understand what's real and what's not real. And, and then they internalize it. You know, all kinds of things start to happen.

I'm, I'm generally fairly liberal in my outlook in terms of how folks should consume information, but some level of, um, supervision is gonna be required here. And I, you know, we basically have built up AI on the same frameworks that we're building social media on and we're not doing a good job of, um, managing access to social media based on age. So I got a bad feeling about all this right now and I, but maybe somebody's got a better idea.

Yeah, I do too. I mean, we know that children are on technology at a very early age now and they are learning and absorbing everything really fast at, you know, as they're, you know, younger age. And so if they're believing what they're reading from AI and it's false information, yes, this is a problem.

And, and then you get the older students using it for writing reports and things like that and taking it at face value as correct information for their reports or not even learning at all 'cause they're just giving it to, to chat GPT to do or something. So there, there definitely needs to be a little bit of, um, control there. I think There will always be lazy students and there are hopefully some students who find it difficult to articulate their ideas then maybe are getting some help from these things.

Mm-Hmm. So it can cut both ways a little ways, but the, yeah, the combination of AI and social media is a little dangerous just because the social media platforms are using algorithms to kind of drive people to click on the next thing. And the AI is sitting there trying to come up with, or guess what is that next thing that they might click on?

And it creates a little bit of a vicious cycle in terms of, you know, if I'm programming these things for an outcome that says more clicks that I'm just going to keep showing increasingly incredulous things to drive the clicks, right? Yeah, absolutely. And then of course, all the, the cyber threats stealing, um, important information.

So definitely some things to consider there. So, you know, just to be vigilant in that area and maybe pay attention to what these children are using on their different screens. Experimenting on people is bad.

Experimenting on children is unconscious. I agree. Well, that brings us to the end of our segment today.

Let us know your thoughts about any of these articles. And again, go check out Textron ITSM. We don't bring up that one very much here, so check it out and let us know your thoughts.

Have a great week and do you have any last words, Mike? As always, folks, lean in. If you don't know what it is, you can't control it.

We'll see you next time.