24. AI Solves All of our Problems – Tech Field Day Podcast

Although AI can be quite useful, it seems that the promise of generative AI has led to irrational exuberance on the topic. This episode of the Tech Field Day podcast recorded ahead of AI Field Day features, Justin Warren, Alistair Cook, Frederick Van Herrin, and myself, Steven Foskett. Considering the promises made about AI and the reality that's likely to result, Welcome to the Tech Field Day podcast, where we bring together a group of IT technical experts to discuss a single idea about key concepts in the industry.

This podcast features a variety of perspectives from members of the Tech Field Day delegate community, and is often recorded in association with one of our events. Today's episode is recorded as a predecessor to AI Field Day, which actually starts tomorrow, September 11th through 13th in Silicon Valley. Tech Field Day is part of the Futurum Group, and this podcast is also published on our sister company site Textron tv.

On this episode ahead of AI Field Day, we'll be discussing, well, the promise of ai. Does it solve all your problems? Well, it does if you're a marketer.

So before we get into that, uh, let's meet who's on the panel today. I'm Fred Van Hern. I'm the founder of ens, a consulting and services company, active in the HPC and AI markets.

I'm Justin Warren. I'm the founder and chief analyst at Pivot nine. We consult to vendors and customers on what they wanna do with technology infrastructure.

And I'm Alistair Cook. I'm a CTO advisor at the Futurum Group, writing about all things enterprise it. So it does seem that no matter where we turn these days, AI is there solving all of our problems, whether it's buying a new car, buying a new phone, probably buying a new refrigerator or washing machine.

Um, certainly, uh, everywhere we go. Um, last hotel I went to, the entire experience was integrated with Alexa. Now, I don't know if you wanna call that AI or bad, uh, but it, but it, but it was, AI was solving all of my problems, that's for sure.

Of course, it created a lot of problems too. But, um, you know, this is the marketing promise, right? And this seems to be the overall promise as, uh, we discussed on the, uh, rundown, the news rundown.

Alistair, uh, California is so worried about AI solving all our problems, that they wanna install a kill switch mandatory in all AI systems because they're worried that maybe the problem is us. Uh, what do you all think is AI solving all of our problems? Um, Justin, I wanna start with you because you, well always have something to say on topics like this.

Yeah. Look, AI solves the problem of what are we gonna do with all these GPUs now that Bitcoin is, uh, no longer the, the topic of, uh, discussion. Uh, I mean, it was really great timing for all those, those stranded investments.

Uh, other than that, yeah, look, it, it's another hype cycle. We've lived through a bunch of these, unfortunately, there's a bunch of really useful things that, what we call AI and machine learning there, there's a bunch of useful stuff it can actually do. That's not what we are really talking about these days.

Unfortunately, everyone just lost their minds over generative AI with large language models, and that's kind of taken over. Uh, it's, it's really a proxy for what people are actually wanna do, which is to replace humans and tasks that they don't wanna pay for. That's largely what I, I see this being useful.

It's an excuse to fire a bunch of people and, uh, replace them with, with cheap slaves. Largely. I think the firing a lot of people is, uh, typical Justin, uh, give, giving us a a great headline.

Um, but there is absolutely a whole lot of work that humans aren't very good at and that we've relied on computers to do for a long time. And the practical uses of AI are often about that, about dealing with vast amounts of repeated activities, doing the same thing again and again, exactly the same way. Exactly what computers excel at and what humans, particularly me, are really bad at doing the same thing repeatedly, solving all of the problems.

Yeah, you just have to define your problems. Exactly right. And AI will solve all of them.

Just leave out all the problems. It doesn't help you with. Uh, the other element I think Justin hit on is this hype cycle.

So I think we are still at that overblown expectations for most customers and certainly for what's written about in, in the mass media. I think as always, the future is here, although it is unevenly distributed, there are people who are gaining huge amounts of value out of even large language models. But I think a lot of the value of AI is coming from more conventional, smaller neural networks, not the sort of generative AI that makes wonderful headlines, but not necessarily so good for business.

Yeah, I think one of the, one of the interesting pieces about artificial intelligence is the ability to learn from the past, right? And certainly when we talk about large volumes of, of data, I mean, nobody has an, uh, you know, books of encyclopedia at home anymore. You can use AI to do that.

So those are great use cases for that. The the problem with hype, I think in general is at first when people don't understand the technology, but they see that they can do something with very simple commands, it's very easy to assume that it can do a lot more than just that, right? So the use cases become kind of infinite.

I think with, with the, the actual use cases out there, it's kind of, uh, being pushed out by technologies, like large language models, right? So if you look at how large language models actually were brought to the market, it almost like it escaped the lab, right? It wasn't really prime and ready for the market, but one small company decided that they want to show off the technology they had.

And from then on, the cat was out of the bag. Now, what that meant was the technology now was available in a very simple format. You know, I like Google search if you want, and you would get results.

Nobody's really answering the question, are those results correct or not? But that's where the hype starts, right? And then there is the, the disappointment.

But I think in general, it's not easy. It's not difficult to understand that people think AI can do a lot based on this. Well, it's a very convincing technology.

I love your analogy that it escaped from the lab because it does feel that way. It's, it's so cool, it is so impressive that no wonder it escaped from the lab. You know, the, the, the promise, you know, basically when chat GT three, uh, you know, came out ba it was so amazing what it could do or what it seems to be able to do that everybody wanted it to, you know, wanted to jump on it.

But that's why I think it's interesting that, you know, we're seeing this incredible backlash against ai. I mentioned the California kill switch bill. Um, even regular people though I was, I was, uh, sitting with my family, uh, over the weekend and, um, my, my wife was wondering if, um, if, uh, a skunk will eat nuts.

And so she googled, do skunks, eat nuts. And apparently that's one of those edge case questions that will cause Google to lose its stuff and give you an amazing AI driven answer. Uh, all of the listeners of the podcast are now gonna Google that.

Um, the point is these are people who don't know, don't care about technology, who are not in the industry and are just using regular tools. And suddenly they're confronted by the impact of this belief that AI solves everyone's problems and it's ready for prime time. Spoiler alert, it's not also spoiler alert, yes, skunks will eat nuts, but they're not as worried about whether they eat them for breakfast, lunch, or dinner.

Uh, skunks pretty much will eat whatever they can get their little mouths on. Yeah, I think that's, that's part of the problem, really, is it like, I mean, computers are already terrible as anyone who works with them understands, and it's amazing. Anything ever works, but everyone, like all the hype is, is pushing it so that people who don't know and honestly can't know, nor nor should they be expected to know the specialist details that we do.

Like they're being told that it's magic and that it will solve everything. But that's not true. So it's kind of fraudulent.

And I, I, I actually, that that annoys me. 'cause that's not in customer's best interests. Like all these co companies that say, you know, customers are at, we have a customer obsession, but you don't because you're putting this stuff that clearly doesn't work and you're testing it live on humans who don't get a choice.

Like try and find a PC that you can buy that doesn't have AI in it. Like, it's really, really hard. It's like trying to buy smart tea, like a non-smart TV when there isn't another option that, well, we don't have a a, a B case in our AB testing, so how can you know if customers really want this or not?

And a lot of them are finding when they go to use this stuff that it's terrible. But just like what we had previously with things like IVR where instead of being able to ring up and talk to a human, you get this weirdly designed computerized system that you can't escape and you can't actually find a way to get to the people to solve your problem. People hated that, but we still got it because it was cheap.

And that's what I see happening a lot of places is that a lot of companies are actually destroying their relationship with their customers because they've gone for this short term cost saving and they've bought into this hype that it's somehow magic. Generally it's people spending their money who don't actually know the tech. And it's gonna be a while before that kind of gets realized and, and they decide actually this was a terrible idea and maybe we shouldn't have done that.

Yeah, I mean, when we talk about hype, you know, there's a, this concept of an AI winter. So we already had TWI two of those. And so what's really an AI winter, it's really when the hype takes over and kind of overcast the actual capabilities of the technology, right?

And so I would say that the hype conversation is probably cyclic. Um, and it's not driven by the technology itself. It's driven by humans expecting a lot more or being disappointed in, in general with, with the concept.

And so when we talk about AI today, I mean, it, it's a whole different conversation than the AI in the 1960s, let's say. But I think maybe a lot of people are, are one, one abusing the term AI for something that isn't really ai. And then the second thing is because technology, you know, hard drives or SSDs and n VMEs and compute is so cheap compared to, you know, maybe a decade ago, it's, it's very easy for somebody with very little knowledge to put something together that provides an answer.

And, and the definition of an answer here is you get words spit out of the engine no matter if the words make sense or accurate. So I think if you put it all together, we're in this fast moving market where large enterprises are coming out with a lot of products that are very capable, but are really beyond our control to figure out what's really happening inside. And then at the same time, you know, AI will do everything, right?

You don't have to do anything. You just sit at home and everything will be done for you. The cooking, tv, everything.

You don't have to go to work 'cause AI will do it for you. And I think those, those items combined, um, are kind of the, the problem, right? There's a lot of education that needs to happen around AI in general and its capabilities.

I think this wave of noise around AI in, in public media is what's leading to elements like that kill switch act that the, um, California, um, lawmakers are gonna push through ev even if it's vetoed by their governor because they have a Hollywood education of what AI is. And so they have, they've envisioning the, the Skynet Terminator situation where you can sit at home 'cause everything's gonna be done for you. You will, you will be killed off, uh, by this Skynet set of AI robots, which really doesn't correspond to anything anybody's seen anywhere, but in the movies.

And so there is multiple dimensions to this education, both the, the general public education of this is what AI really is, but also the business users who are trying to gain some value out of this, trying to get rid of those excess stuff that, um, that, uh, Justin would, would like us to retain because the human touch is important, like human interaction. Uh, but there is business benefit from using these, these tools if you understand what these AI tools are doing and what your data is. I think Frederick is very much about managing that flow of data coming in, because like any other system building your AI garbage goes in, you're gonna get garbage out guaranteed.

Uh, even if good data goes in, AI can give you garbage back out again, which is one of its wonderful capabilities. It knows the context of everything and the meaning of nothing. Uh, it's an, an amazing tool for giving those those words that might or might not mean something.

But, you know, maybe AI will solve all of our problems because the over, uh, irrational exuberance toward AI will give us all a lot to work on in the next coming years, uh, as we have to make these systems actually do the things that they've been promised to do as we have to remediate the problems that these systems have caused as we have to, uh, install and configure the infrastructure to support all of these systems. You know, I guess it depends on who you're talking to. I mean, the listeners of the Tech Field Day podcast are probably people like us who are out there, uh, implementing IT systems and IT solutions.

Uh, AI might be solving a lot of their problems as we speak since it gives everybody a lot of new things to work on and new things to do. And frankly, it's really fun. I don't know if y'all have, uh, spent much time working on, um, creating your own AI agents or, uh, trying to, uh, feed your own data into, uh, LLM.

It's a lot of fun to work on it, and it is actually, uh, delivering results in many cases. Uh, frankly, a lot of the, uh, AI solutions that I've used, uh, as part of Tech Field Day, as well as in my own systems, they actually have solved a lot of my problems, a lot of the things that I was having trouble with I'm able to accomplish with ai. So, you know, let's get back to that premise.

Maybe, maybe we're looking at this wrong. Maybe we're looking at it instead of, uh, instead of AI solving problems by actually providing useful solutions, uh, maybe, maybe we should be looking at it as AI giving us the next wave of success for the computer industry. Does AI solve that problem?

Um, maybe I, I think it'll be better once the tide goes out and we see all the bits that get left behind, like things like automated translation and transcription vision systems that are able to identify things automated captioning for people, um, who, you know, have vision issues so that you can actually have audio descriptions or, or written descriptions of what images look like. There's a whole bunch of quite useful things, but they're being overshadowed by all of this other hypey stuff. My concern is that when the, the shine comes off, off this hype cycle, when we get a bit of a pullback, everyone has thrown so much into it and like the, the s and p 500 is dominate, like I think about 20 to 30% of the s and p 500 is like six companies who have all pivoted really, really hard into ai, Nvidia in particular.

And if the, the shine comes off that, and if investment drops off from this hype and we get another AI winter, I'm, I'm concerned that that will have such a big ripple effect on the rest of the tech industry that we'll actually have like a tech recession, like a big one, and it may actually spread further than the tech industry. Uh, and that will then mean that there's less money available to go and do all of this stuff. That, that's my my concern that we, we will get more of a hard landing than a soft landing.

Um, I would like it to be more of a soft landing because no one likes a recession. So AI as a, as, as, as considered as innovation should help at all levels, right? I think if we look at, at the world, it's innovation and new technologies that help us around, and I think AI definitely will help.

I think the one challenge with AI compared to other technologies, um, in the past is that the AI technology goes so quickly and so fast that there are no, no actual breaks on it, right? It goes, it goes and being developed by all the major, um, companies like Google and Facebook who generate all these large, large language models. There is a new model every six months or so.

It's very, very difficult to keep track of it, and it validates what is being delivered to customers. I share a bit of Justin's concerns about the huge investment and the, the, the business kind of, uh, calls that large organizations are making around, uh, generative AI and the investments and, and huge quantities of compute. And then looking at the, the math around how much power is this all going to consume and what's the environmental impact of this?

I'm like, Justin concerned that that might lead to a, a very poor time if there's no return on that, that investment. But I'm hopeful that we will see some returns on that. We will see businesses getting value outta their AI investments because these are huge bet the business kind of investments.

I really liked, uh, Justin's metaphor of the tide going out and seeing what's left behind, uh, because it's so true that, uh, you know, the, this explosion of creativity and new ideas and people trying to apply AI technology and, you know, LLMs and generative AI and, and, and all the rest everywhere is leading us to, frankly, a bonanza of very, very cool technologies that probably aren't products, but they will be left behind when the dust settles. And so we'll be able to go back and look at that and we'll be able to say, man, you know, here's, you know, five open source projects that are doing some very, very cool stuff that we can apply over here. We're already seeing that in many ways because a lot of the AI work that's being done is open source, and we're already seeing people basically picking delicious apples from the tree and saying, my goodness, look at this one.

Well, we could use this over here and we could use this over there. Uh, ultimately the, uh, even if the current wave of technology, uh, or products don't amount to anything, the, the benefit of what we're doing, I think will, are, are, are you guys in agreement on that question? I, yes and no.

I think some of the things that get left behind will be great, and I think some of the other things that get left behind are not great depending on who you are. So like other industrial revolutions, I mean, they were amazing for the people who weren't children working in factories. Um, and it wasn't so great if you made socks at home, uh, as part of a cottage industry.

But we also meant that, you know, we got slightly less great socks, but more people got to have socks because they were cheaper and more accessible. So it, it's going to be a mixed bag where how the distribution of the benefits happens is where I'm, that, that's what I'm looking at is, okay, how much of these benefits are gonna actually exist? Uh, and as Alistair mentioned, I don't think it's gonna be anywhere near enough to pay off all the investment, but hey, that's startup land for you.

Most of them fail, but what gets left behind, like, where do those benefits go? To whom will those benefits accrue and will they be more equally distributed? And that might have a lot to do with this question of open source, Justin, because, you know, again, we've never had, you know, you talk of the industrial revolution, you know, imagine if anyone could just replicate a factory instantly once the factory owner was no longer, you know, interested in pursuing that.

It, there's, this is an an unusual time because basically we can take all of the investment that Meta has been putting into llama and instantly use it for free in some other way. We've seen that with the web, uh, we've seen that with the technologies that make all of this internet stuff work. And, and I think that wide, you know, we're widely seeing that in ai, right?

Yeah, I think we've seen it in ai, but I also think we have seen that in the past, right? Think about medicine, right? So there's medicine is protected, nobody else can, can make that medicine, and then somehow it ends up in a lower cost country where not for free, but for a lot less, they, they produce that medicine.

And the same thing, you know, my father was in the electronics business for a long time, and you know, there at some point, uh, people were concerned that they were going to lose their job because the factory that was making, you know, the very old light bulbs, they thought they were gonna lose their job. While in reality, you know, the people with the vision were saying, well, we're going to electronics and LAD and people were saying, that's never gonna happen, but it did happen, right? And those factories then went from, you know, more from to a, a cheaper country, or the labor at least is cheaper and, and built that.

So I think, I think AI just accelerates this, right? The fact that you can, that something costs a lot of money, a lot of in innovation, and then somebody else can take it over. Now, when you talk about open source without open source, they would not have been an AI today, right?

Because if AI would've been created by a company that was closed loop, you would never have access to the algorithms, right? So Steven, when you mentioned you could build something like a chat bot really easily, that's because everything is coming from open source. It's not coming from closed, closed loop.

On top of that, the companies that used to be closed loop for their traditional software, like an operating system, they're also participating in open source. So I think open source is very key and an, and a magical enabler for ai. Um, but I think it just moving so fast, it's just, I think the concern is, is who controls this?

Who decides which product is better? There's a lot of competition going on, and the competition between organizations is to do it faster, not more accurate, not better, faster, whoever gets out to the market first is considered the person who wins. And it does feel like it's a, a AI land grab that there is, there's such a huge rate of change that to me is always a concern when I want to run production.

Uh, that fact that I might need to be updating new models, new methodologies, new hardware requirements every six to 12 months is a huge concern in the production environment. Uh, but I agree this, that this open source is a huge enabler for technology and for the adoption of technology, for the longevity of technology provided we're actually protecting those sources. Uh, one of the challenges with open source is abandoned wear and, uh, projects that are no longer being developed over time and then just disappear because nobody's paying for the hosting of those projects over time.

So there's some governance and control requirements, and this is where some of the, um, open source control bodies become very important for the future. Uh, continue to have really mixed feelings here is that rate of changes is concerning for me for deploying something for production, but the capabilities that are being delivered are, are quite amazing. I do come back to one of Justin's suggestions of what this, what, what our topic could be is that AI is a 5% solution.

It's not, it's not the thing, it's, it's the thing that makes the thing better, as Bobby Allen said at the last AI field day, and you've gotta have a thing before you can make it better, and let's not lose sight of having the thing to make better. Yeah, I mean, uh, uh, from a, from an open source perspective, I think it's also important to, to mention that when we talk about open source, we actually talk about code, right? AI is mostly driven by, by data.

And so the data is typically not open source, right? So it used to be in the traditional HPC world that the, the code was your ip. Today, the code is is, I wouldn't say irrelevant, but it's, it's irrelevant compared to the data.

Your data is the ip, right? And I think that's, that's something I would like to see where from an open source perspective, there's more work being done on the data side, so we have a better understanding what the end product will look like. Yeah, I mean, I'm, I'm old enough to remember when big data was gonna solve all of our problems.

Um, and this, this just feels like another variation on that. So I, I mean, it's, it's difficult not to be skeptical having, having lived through this before. Um, I just see the same patterns happening.

I I see us there, there's a lot of this stuff automating things that probably don't need to be automated at all. They should just be gotten rid of. Um, and, and that's, that's actually part of my concern with some of this is that you, when you use a computer to automate stuff, it actually makes it harder to get rid of because you have all of these layers of things built on top of it.

And we, we are baking in a whole bunch of things that are actually terrible. Like a lot of enterprises would do much better to just look at what they're trying to do. And rather than automating it with a computer, say, why does that process even exist?

Like, this is just dumb, stop doing it. Um, you'd save a hell of a lot of money, but that gets really difficult when you've got a lot of technology being layered onto it. And so, oh, no, no, we have to use this techno technological solution to, you know, we got one hammer.

As Al said, I think before, it's like we've, we've got an AI hammer and we're just gonna hit everything with it. Um, and, and the future is more hammers. I, I, I just don't see that really working out.

But I think that we have to go through this, this transition period, because if one thing we've learned from, from history is that humans don't learn from history much at all. So we will have to make a whole bunch of these mistakes in order to then realize that, oh yeah, that was a terrible idea, we shouldn't have done that. Uh, and then we'll just go and hopefully clean it all up.

Um, hopefully there's enough of us left around at the end of this to be able to go and do that cleaning up. So I wanna wrap then by posing the, the, the, the premise back to you, AI is gonna solve all of our problems. Uh, I don't wanna say yes, no, because of course everybody's gonna say no, I want you to say phra.

Let, let's phrase it. Let's give the listeners the answer. How exactly is AI gonna solve all of our problems?

It will. How is it gonna happen? Uh, Alistair, I'm gonna pick on you first.

How will AI solve all of our problems? So I guess I'm known for snark, so I'll start with the snarky answer. Um, Skynet will come along and remove all the humans, and then we will no longer be ruining things forever.

Uh, more realistically. Um, AI is good at seeing patterns and things, and humans, as Justin has just said, are not good at noticing patterns. Patent, let's, let's solve that problem with ai.

Alright, Justin, uh, he, he used your name in vain. Uh, how is AI going to solve all of our problems? Uh, well, I, going with the snarky answer first.

Um, I think it's going to highlight the terrible, um, ravages of capitalism and, uh, accelerate the, the downfall of the capitalist system and usher in a glorious communist, um, future for everyone. Um, slightly, slightly more seriously. Uh, yeah, I think that we're, it's gonna solve a whole bunch of issues for people that have been overlooked by things that were very difficult for humans to do manually.

So things like automated translations, things like making things a bit more accessible. Uh, things that were deemed to be too expensive because we just didn't care about those people very much. The, these kind of automated systems are great for, for those, those areas.

Um, so it's gonna solve a lot of those sorts of problems. Um, I think it's gonna highlight the problems that we have elsewhere. So it's part of that solution that you, it'll make a whole bunch of, it'll make a lot of problems much more obvious, and it's only when things are really on fire that we'll actually address them.

So that's quite useful. Yeah, I think if AI, as, as an innovative tool will, will help and solve a lot of problems. But I think at the same time, uh, if you solve a problem, you might create another problem somewhere else, right?

So maybe, maybe the answer is a catch 22 where you're trying to solve something, but you created so many other problems that you end up back at, at square one. But I think AI is definitely a technology that can help. AI is a tool, so it's not necessarily good for everything.

Um, AI learns from data, so if you keep on feeding it new data and data from activities, it'll improve and learn from itself. Um, it can be a very dangerous tool, you know, bad input is, is is a bad output. But, you know, as an engineer, I'm very positive that it can solve a lot of problems.

Uh, I wouldn't like to be told that AI is, uh, driving my car for me or if it's attached to a medical device, but otherwise I feel free that it'll solve a lot of the problems. Well, I guarantee it's gonna be driving your car and attached to your medical devices in the future, so you might as well just unplug now. But, uh, I, I honestly agree that, um, with the, this idea that the technology that's left behind when the exuberance fades is going to be ubiquitous, it's gonna be used everywhere.

We're gonna see it in a lot of places. And frankly, I think many of the things that we as humans do, whether it's driving cars or, or using medical technology or posting on social media or crunching numbers or interacting with customer support or any of the things that have been brought up in this, I think a lot of those things are gonna be AI powered. And I think ultimately, I'm going to go a little far out here.

I think AI is gonna be used in the solution of many of the everyday problems that we have in the coming years because I think it's almost inevitable that the hammer that is currently in Vogue would not get used in many different places. So I think AI is gonna solve all of our problems, just not in the way we expect. So thank you very much for being part of this discussion, I have to call it to an end.

Uh, we've gone on quite a little bit of time here. Before we go, uh, first off, we're gonna see you all at AI Field Day. Uh, check out AI Field, a five at the tech field, a website, uh, you'll also see it live streaming on LinkedIn, on Textron.

Uh, let us know what you're looking forward to at AI Field Day this week. com and LinkedIn. That's Frederick Van Hern and I'm definitely looking forward to join you and the rest of the delegates tomorrow at the Tech Field Day.

And looking forward to, uh, what's new in the industry. I'm looking forward to hearing some actual solutions of things that have been built with ai, uh, at AI Field Day, as well as hearing some more of the cool data center infrastructure that people are building in order to, to facilitate this, you can find me as Alistair Cook on line or de task nz as my identity. And of course, AI Field Day is a great place, uh, hashtag AI field day five.

I'm looking forward to being in the room when, uh, we, we find out what these companies are actually doing with AI to find out if it's real, uh, or if it's just three linear regressions in a trench coat. com for consulting. com.

I do recommend The Crux. I'm a subscriber and a reader. Thank you Justin for that.

And thank you everyone for listening to this episode of the Tech Field Day podcast. If you enjoyed this discussion, please check us out on YouTube or in your favorite podcast application. And specifically, look back at our AI is just a fad and AI is not a FAD episodes, which, uh, were published a couple episodes ago, and those are a lot of fun too.

com or your favorite, uh, streaming platform for our forthcoming AI field Day sessions, which will be posted on YouTube. com/podcast or check us out on Techstrong tv. Thanks for listening and we will see you next week.