The Future of AI at AIE 2024

Hi everyone. Thanks for joining us today. I hope you've enjoyed, uh, the sessions from our virtual event today, the Artificially Intelligent Enterprise Online Conference.

And if we're gonna talk about the artificially intelligent enterprise, there's really only one person who I could think of to, to talk about it with, and he's kind of the guy who, who kind of coined the term, and it's the, the name of the newsletter that he puts out that I'm sure many of you subscribe to. If you don't, you should certainly do it after today's event. And he's also sort of the, the guiding light manager of, of today's event and our very first artificially intelligent Enterprise online conference, my friend, mark Hinkle.

Hey, mark, welcome. Thanks for joining me on this fireside chat. Thanks, Alan.

It's great to be here. It's, it's, it's actually great to have you here. You know, mark, it's interesting, we sit here and we're gonna talk a little bit about the future of ai, near term, short term, long term.

Um, but my, what a strange trip. It's spent, right, for both of us, we've both been in the tech world for, for a number of years. I always like, especially when we do these sort of chats, to give our audience a sense of how did you get here today, mark, let's talk about your journey.

Okay, that's a great, it's a great story. So, you know, my background was gonna be, my first job was in finance, and a buddy of mine was going for a job interview at one of the first internet service providers, uh, PSI net, which was huge. I remembering that.

All Right. com bubble. And, uh, he couldn't find his way to the interview, so went along with him and, uh, I'm like, well, I'm here.

I might as well interview. And, you know, fast forward, and now I'm gonna date myself. 29 years later, I've been working in enterprise tech from open source systems management, where I think I used to run into you at conferences and the speakers here, and you were talking security to cloud, to, you know, all sorts of, uh, emerging tech.

And when I finally saw ai, um, I had my own company. We're a venture backed company by, you know, Cisco and Index, and I'm like, this is gonna be the next thing. And it's a, you know, the, the technology of a lifetime.

So that's how I got here today. Excellent. You know, ha having observed you, as you said, running into at conferences, but certainly since this AI Burst on the scene, this whole, you know, generator of AI and everything burst on the scene.

It, it was a little bit Richard Dreyfus, Dreyfus esque, if you will, from, uh, encounters of the third kind, where you've been building this model of like doubles peak or whatever that was, if you remember the movie. Yeah. I mean, you've really, it's got, it's, it's grabbed you by the throat, meaning AI has really grabbed you by the throat and, and, you know, brought you right in to it.

But as, as many things you've done in your career, you've, you know, doing a little juujitsu wound up on top with this, and, you know, rather than it driving you, you are trying to drive it. Yeah, yeah. No, Good.

I think the thing that is unique about this is, I, I am always excited about new, new technology, but this is a whole new level. And whenever I learn something, I love to share it. And the, when you start sharing your story, then you hear other people's stories.

And this has been the most fun I've had since the early days of the internet, just because of how it's transforming people's lives from, you know, new careers, new opportunities for businesses, and just an overall, you know, cautions, let's say cautious exuberance. Let, like what it could do. Like, you know, I like to work, but it would be great if I could work on more things that are meaningful and less of the mundane.

And I think that's, you know, I look at, you know, 200 years ago, people with hose and rakes out in the field, and now they drive air conditioned tractors like that, that was, and it improved our food security as a, in the world. AI is gonna do similar things in whole new realms that we haven't even began to touch. Absolutely.

You know, it's funny, I I, I think I told you off camera, this is the third or fourth video that I, I've recorded today. Um, and AI has been a topic at every one of 'em. Matter of fact, it's very hard.

I, I start recording now when I do interviews to see how long it'll take till it comes up. Um, and without a doubt, it always does. Like you, I haven't seen anything since the onset of the internet itself that I think has reached this level of, of potential, of excitement, of investment, of just hype, right?

But you always, uh, you know, you're always wary of, man, can it live up to the hype? Is it gonna live up to the hype? And because invariably people will, if it doesn't live up to the hype, say, ah, it was, it was a failure.

This AI thing is a, you know, nonsense. And, and the, the expectations being set right now are so astronomical. I don't know if it can live up.

I think eventually it will. But, you know, the tech world, especially we have, we have very short attention spans here. You know, this thing has about 18 months to live up to it, otherwise people will start calling it a failure and they'll be looking for the neck shiny trinket to chase.

Um, what do you think we, you know, so let, let's focus on the next 18 months, mark. What, what can AI do over the next 18 month months to make people say, wow, it is the future? I think the biggest thing is that ai, right now, it's a little bit of a magic trick, right?

It's, it's, what it's doing is it's looking at all this data and having instant recall and synthesis of this data, like the world's smartest librarian. And that when it first, the first time you use it, you're, it's pretty gobsmacking. But if you wanted to solve a problem, it's not much smarter than our dog.

So I know you love your dog and I love my dog, but they basically are the smart as a three or 4-year-old child. So what's really gonna happen is better reasoning. So it can solve problems.

And I think we're gonna see not complex agents, not Agent Smith from the Matrix, but you know, agent mail handler an agent, you know, report writer and agent, that research assistant where you can give it an objective within bounds. And it's really gonna start to become like a dedicated virtual assistant in all sorts of ways. That's what's gonna be the next 18 months.

And I mean, I think that we're starting to see the frameworks for that to be deployed, but at the end of the day, these models have to be able to do some reasoning that is beyond their capabilities today. That's the big, you know, um, aha moment. And then once that starts, we're gonna have the equivalent of Moore's Law for Intelligent, where intelligence, where it starts to get smarter faster.

And that's when things really excite. And I think that probably is the next 18 months. I've got a bunch of questions for you, just, uh, from that response, but let, let's start with, you know, the different AI tasks that you, you brought up, right?

AI mail handling, AI report, right? Or ai, this or that. Does that mean we're gonna have a series of different ais that are very specialized for each of these tasks?

Or do we have one YA super AI that does each, that does those tasks for us? I think what we're gonna have is a lot of ais and I think they're gonna be some kind of router in front of it. So if you ask your AI to, you know, write a report, it's gonna go to an expert AI that's good at writing reports for media companies or, you know, businesses in that sector.

If you have medical issues, there's gonna be an AI for that. And it's gonna be, you know, when you, you see some of these popular ones, like, like Mytral is a, a mixture of experts. That's what it really is, is the bunch of smaller models that are routing to, to have it, to have expertise in each way.

'cause you're not, it's not practical, in my opinion, to keep training these models on all this data. 5 and four, they throw those models out afterwards because they're, they're expensive to build and then they're just as expensive to maintain. So what we want to do is, you know, start segmenting and specializing in certain areas, and you'll have your own, you know, enterprise GPT, which may not be nearly as powerful as a LAMA or an open A-D-I-G-P-T, but it's, it's definitely going to have the context for it to solve these problems and it does better in a narrow context.

Sure. My, my next question, and it actually comes out of a extra gang I recorded this morning, was, you know, the future of, let's call it an A IPC or whether it's an A IPC or an AI tablet or an AI phone, but, you know, AI being done rather than up there locally, right at your device, whatever device that is. Is that an 18 month thing you think, mark, It's happening today, and there's two, you know, two places where I think they've done really well is Microsoft did their fee model, which is a small language model.

Um, you know, the, the sort of the, the, a continuation, my previous theme is these smaller models are gonna exist on the edge and, you know, they might be more, um, task related models. Um, same thing with, um, uh, Google. They have their Gemini model that is running on Android phones already.

Um, just the, the, uh, Google Pixel I believe. But what they're doing is, just what I described earlier, is there's a certain set of tasks that, you know, our Siri and Alexa have not kept up with. They're not, they're not able to, to take these multi-step tasks as easily.

And they're always dependent on, like you said, they have, they're not, they're gonna be local. You don't want 'em to be tethered to a data center. So I think you're gonna see, you know, models that do some of these action oriented tasks along with learning your personal preferences when you say, Hey, play my favorite podcast.

You're not gonna have to say the TKA Don Gang, it's gonna know that's your favorite podcast. Or it's gonna say, you know, gimme my newsfeed and it's gonna curate your newsfeed based on your preferences. And they're not really hard things to do, but they're a little easier to do when you have a learning model rather than one that sort of needs to be programmed and sort of that if then else kind of format it.

Yeah. You know, uh, recently Chad, GPT-4, oh not just for came out just recently, I was a couple days ago, a week ago, whatever. Um, and to me that represents what I thought Steri and Google Home and a and Alexa were going to be, you know, we, so getting out of the chatbot interface to a verbal interface, and it talks back to you, but with the, with the magics, 'cause it's still kind of, you know, smoke and mirror magic to me of, of today's AI versus what theory or Alexa had, you know, prior when it first day first came out.

That was kind of disappointing to me when they first came out chat GPP four oh sounds like, okay, now they got it right. I have actually, um, in the newsletter last week, I talked about, um, integrating the open AI app instead of Siri. So when I say, Hey Siri, I say, Hey Siri, ask, um, chat GPT and it brings up that interface and I talk to that because it is that much better.

And, um, I think that what Apple got right originally was, um, uh, the, the sort of friendliness of the Siri voice that seemed reassuring, but open a, I was very, you know, clinical and cold and now they've added the perception of empathy there. And the, the back and forth is much more natural, like a conversation. And I think, you know, like you said, it's a little bit of smoke and mirrors.

It's still, it's still getting better all the time. The same thing with Google is, um, the day after OpenAI, OpenAI launched, I think last Monday, um, strategically the day before Google had Google io and they, they showed some of their Gemini age it, and the vision is, you know, I don't doubt that they'll achieve the vision of an assistant that is smart enough that I think they use the example of planning a trip. And as you know, as we both have a little bit more grow greater hair than we used to.

Mm-Hmm. Yeah. Or at least I have less hair than I used to, but go ahead.

But, uh, but you know, we have a very particular way that we like to travel. It's not like you're gonna pick the cheapest flight and stay at the cheapest hotel anymore. There's certain preferences that I want to be, you know, near transportation.

I want to be near restaurants, I want to fly at a certain time of day. I don't want to take a red eye and all of these things, it's gonna, it's gonna learn your preferences and it's gonna be the travel agent that we always wanted. And I think that kind of those same thing for grocery shopping and these are the sort of pedestrian tasks.

And then when you get to work, it's gonna be the, you know, it's gonna help you process information and the way that you like to receive it. Audio, video, text. And you're gonna have these smarter, more robust interactions 'cause they're not programmable, they're adaptable.

So yeah. What a quick, strange new world mark. You know, most of the people watching this today are tech workers.

You know, they're IT folk like you and I. Yeah. So I think AI's have a profound impact on society, on civilization, on everyday tasks that we all do, but specifically for IT workers, for our friends in the developer space, you know, we've heard say, oh, it's gonna replace, it's gonna write code, it's gonna write all the code, maybe it's gonna do all the tests.

It's going to, it's gonna do everything, all the IT workers, you know, and the IT workers. Every time there's a new innovation, some segment of IT workers are supposed to be made obsolete. And, and the reality is, is that there's always more jobs after than there were before.

Right. But where do you see the future of that, right? The, the future for IT workers in an AI powered world?

I think we're gonna do a lot less of the carrying water and chopping wood and a lot more of the high value tasks. You know, there's, there's no honor or I shouldn't honor is the right word, but, um, in writing a unit test, other than it's really great when you have a unit test done that tests what you just wrote. That's where AI does great.

Or how many times do you need to rewrite the same self routine that, you know, pulls a database. That stuff is, is from a developer side is not a high value, but you know, elegantly, you know, making the code faster and perform it and redundant, that's a higher level of skill that's not soon to be replaced. So, you know, on the developer side, we've seen these, you know, demos from like Devon, which, um, there was some great, you know, never estimate the power of a good demo because they raised I think a hundred million dollars afterwards.

But then you start looking at what they were demoing and, you know, they were definitely selling ahead of the curve. Um, and I think that in other spaces, systems management security where you, you've been big at, there's a certain level of, you know, things that we haven't been able to do because, you know, like think about the amount of data that comes in for an, a security system and actually drawing real insights from that. Even the best analyst can't go through, you know, millions of the, you know, log files and lines of the log file or, you know, a monitoring and, uh, systems management person can't digest the amount of information that we get.

So I just think it's, it's gonna inform our decisions better. We're going to, you know, be able to shorten our time to do things like root cause analysis and I think that our systems will adapt. But, you know, you could have a thoroughbred in the Kentucky Derby and they can run faster than any other one, but if you don't have a jockey on there to get it across the line, it's gonna run all over the place.

So I think that's, that's really where we're at, is we just have these big powerful, you know, thoroughbred systems and we still need, you know, something to direct them and it's gonna be people with critical thinking and analytics and the ability to convey what needs to be done to these machines that are still, you know, in their infancy. Absolutely. I wanna continue on that theme for a little bit.

You know, the Elon Musk of, so of course has sued open AI that they didn't say they're not doing what they said they were going to do, and he originally invested and was part of it. He started his own gen AI company now, and rumor has it, you know, there's must be like a $10 billion deal with Oracle Cloud. He's gonna host it in the, in Larry's Oracle cloud.

Um, is it still early enough where we don't know who the horses are gonna be? Could there be some dark horses that we don't even know about yet? Or are we already sort of locked into the usual Google, apple meta, Microsoft, Amazon, Oracle, kinda IBM of course with Watson, you know, is there room for some upstarts here?

Yeah, I think this, if we like use cloud as an analogy, you know, when they started doing compute and storage in the cloud, it was at a premium price and it was a little differentiated. Amazon was very, very, you know, cost effective compared to Google. But at the end of the day, to today, there's a dozen, you know, providers, all the hyperscalers are providing, you know, cloud and compute that's essentially the same.

And the cost of compute, compute storage went down significantly. Um, and then people moved on up the stack and, you know, provided other, you know, services and things on top of the cloud. I think that's what's gonna happen in AI is there's, ironically, Facebook is probably gonna be one of the big providers along with Microsoft and Google and obviously, um, Amazon, and then there's gonna be another layer.

And those layers are gonna be things that are, you know, addressing verticals. There'll be great financials, there'll be the equivalent of, you know, maybe Oracle financials will be Oracle AI financials or you know, Microsoft will have theirs. Or maybe Intuit who has all this data on financials from years and years and years will train a model that's very narrow and is the best, you know, virtual bookkeeper out there.

Who knows? I think those, those things will come about. And then there'll be a lot of things that are, you know, we haven't even thought of.

I think there will be virtual staffing agencies that'll have new staff. I think they will be AI agents that you will say, Hey, I need a, you know, I need a a a receptionist, I need a virtual admin, I need a, you know, marketing operations person. And those folks will, you know, they'll, they'll be robots for lack of better term, or there'll be a AI agents that will assist us in some of those ways.

And then we will have new jobs where, you know, there's nothing, you know, answering the phone and directing calls is not gonna text the human mind. But having them be able to do higher value things that really increase our business and what we can do, that'll be where we'll see, um, see a whole new industry that, you know, when we came out with the automobile, how many auto manufacturers were there and how many do we have today? I mean, and how different are they from Tesla to, you know, Hyundai to, you know, all these things.

It well, it's an interesting piece of it. Um, the, the, the, the, the, not the problem, but the gist of it though is look, the automobile, let, let's take the automobile for instance, right? The automobile changed, the American landscape, changed the American in many ways.

It defined the American lands, uh, way of life. Sure. Right.

Getting in the car, two car garage, you know, mass transit in the suburbs was probably forever stunted because of it. Um, and and around the world though, in the rest of the world, they, it had maybe a little bit of a lesser, uh, impact. When, when we look at the automobile, it probably took 40 years because really the automobile, the period of the automobile is kind of a post World War ii, right?

That was like, to me, the, the real boon, right? That's when suburbs pop up and all of that empowered by the auto. That was 50 years after, you know, the first automobiles 30 years after Henry Ford model t's AI here.

You know, if you talk to our friend John Willis, we've been working on AI for 30 years. Yes, we have. And probably longer than that, but when we look at, let's say the release of Open AI's chat, GPT as sort of the modern era, you know, BC ad right of, of Gen ai shoot, it's only been, was it less than two years, right?

And look where we are, we get 50 years, you know what, what's gonna happen in two or three years more down the line? Sure, sure. And I mean, 'cause it's a digital product.

We don't have to build, I mean, we have to build chip factories, but we don't, when you build software, you don't have to build a, you know, a plant you don't need near that kind of capital. You don't actually even need that, that number of, uh, workers when it comes to information products, you really benefit by having small teams that were big teams 'cause of the network effect de signal degrades over time. So you have things like Jeff Bezos and his two pizza teams with, you know, six or eight people on it that can, you know, do a lot of things.

And, and now we're assisting those people with, you know, infancy AI products. What happens, you know, every year we get a little bit better and we're seeing, gonna see exponential gains on what we can do. Now you mentioned, you know, digital, but damn AI's made hardware, cooling hat, right?

We stopped talking about hardware in our world for a long time. It's commodity hardware. Well, it's not so commodity anymore.

Nvidia are 2 trillion, or they were before their last run. I don't know if they're still 2 trillion, but you know, the fight is on for GPU chips and a MD and Intel are, are fighting for their lives. Mm-hmm.

Right. To get a share of this market. And the biggest sort of governor on, on even greater AI adoption is the availability of these GPUs and, and whatever the next gen AI chips are gonna be.

I mean, to me that's pretty darn incredible that all of a sudden we care about hardware again. Yeah. We care about hardware, we hear the, that power is gonna be a choke point.

We hear that network's gonna be a choke point. You know, it's, These are all the things we heard when the internet first went public commercial as well. Yeah.

And you know, look at where we started out with is I was happy as I'll get out when I could get 56 K as my download speed from me and You both. And then now the other day I got a thing that said, would you like eight gigabits to your house? And I was like, I don't even know what I could be doing to get to use that kind of bandwidth.

I Mean, I take it and then think about it later. I think that's what I told the guy. I called me.

I was like, yeah, I'll be interested. Yeah. But I, I think that, you know, it's funny because, uh, at the time I worked at that early internet service provider, um, uh, the guy who was the CFO for, uh, MCI, his gene, his name was Gene Gabbard, he was a bit surf spots father of the internet, and he was on our board of directors.

And, you know, he was, he's one of, he was a smart electrical engineer. He wasn't just a number cruncher and, you know, he was saying, oh yeah, you guys, this is how things work in telecom. It doesn't change.

And, you know, I'm sure if I talk to him today, it's like, yeah, things change really, really fast, faster than we expected that, you know, the way we were provisioning T ones and t threes and, you know, going from frame relay to a a s or DSL and cable. It just, it was transformative to the way that we could consume information. And that was, you know, unlike anything we'd seen five years earlier.

This is two years in, and since chat GPT came out that I guess it'll be two years in November, right? It just, it just opened the floodgates of the way things are changing. I, I mean, I, I talk, I wanted to walk with my dog talking to an ai, having it analyze data and write a report for me.

So when I get back to my office, I can sit down and have all this information collated and, you know, look, I want to analyze what I'm doing and I have an assistant that will, you know, if I write an article for Techstrong, it, it can say, oh, this would make it better. It'll look at trends. And it's, it's just such an amazing disruptive, in a good way to the way I'm working.

And I'm very much at the forefront of it. I mean, my, I have the luxury of having a job where I've gotta consume this stuff as it comes out. And then I get to see that, that sort of leap forward every time a new model comes out or a new, you know, kind of technology or riff.

It's, it's just like nothing I've ever seen. Not no, nothing like I've seen either. Hey, mark, we're about outta time.

I want to thank you for sitting in on this fireside chat with me for our, uh, event today. But more importantly, I want to thank you for really breathing light into the whole event today. This really was kind of your brainchild and you've managed and led it and led the team here and can't thank you enough.

And I'm, I'm sure everyone out here feels the same way, so thank you. Keep doing what you're doing, man. And let's, let's look at doing one of these.

I don't know if we'll wait a whole year because with AI time, you could probably do six months and we should do it again. Um, but thank you. Oh, thank you.

You've textron's been a great partner and, um, we look forward to this and hearing how everybody enjoyed the day. So, all right, Mark Hinkle, the, the, the intelligence behind the Artificially Intelligent Enterprise Online Conference. Thanks for joining us.

Enjoy the rest of the conference. Remember through the magic of virtual, you didn't miss anything. You could always watch any of the sessions on demand as well as one there, uh, streamed live.

And, uh, we hope you enjoy it. This is Alan Shemel for Textron.