Fix the System, Not the Recruiter with Adam Kovacs Live on Shally's Alley

Boo! Did I scare you? Happy Halloween, everybody! It's not often that I get to have a Shelly's Alley episode on Halloween, but I guess that does happen once in a while. So this is kind of the scary episode.

Do you have anything better to do?

No, no. My children are off on their own doing their own thing for Halloween, so I get the house all to myself. I'm debating whether or not I turn the lights off and hide in the basement so I don't have to hand out candy. But then the problem with that is I'll eat it all. So I'm probably going to have to hand out as much candy as I can, then turn the lights off and hide in the basement. But yeah, I'm going to be— even my wife has plans with a friend, so I'm literally home alone for Halloween. That's— that is— if that— if the world wasn't scary enough, now we have another reason to be afraid. But yeah. And we have an Eastern European guest today, so even more scary.

Oh, come on. We are friendly people. We're just way too direct sometimes.

Yeah. Well, that's where I'm from. My family's genetic origins, if you will, trace back there before the war. So definitely familiar with that. Which one? Yeah, right. Well, actually, they were there before both wars and then left after the second one. So, um, yeah, they, they, uh, they went down to South America. So, uh, anyway, welcome to Shelly's Alley, man. We are live and it is Friday and it is Halloween and it is getting ridiculously cold for at least Atlanta. I don't know, Seattle, you said it was rainy and nasty.

The darkness started. It's the long darkness started last—

the long darkness.

It's like this until March, uh, but Hey, it's just, it's just autumn. It's fine. Uh, yeah, snow, it would be worse.

So the two seasons in Seattle: wet and dark and wet.

Nah, actually, in the summers are okay. In the summers, somewhat heat waves and all that. Like, not like heat waves as we had like back in Dublin where they caught like 50 degrees Fahrenheit or like 15 Celsius, a heat wave. Not like that. So It's nice weather. Just people who complain about Seattle weather, they just don't like the proper seasonality of it.

That's right. Yeah, no, I definitely spent a lot of time with Microsoft in Seattle. So, I'm familiar with the weather patterns. And believe it or not, it rains more in Atlanta as far as what like water accumulate, whatever accumulation, rainfall inches or something, much more in Atlanta than in Seattle. It's just that in Seattle, it will mist, you know, quite a bit. And so it feels like it's always rainy, but in Atlanta it's really raining and you have humidity. But that's, that's all year round. There's nothing you can do about that. The heater sometimes makes it a little bit better. But, um, anyways, so man, uh, welcome to the show. I'm super pumped that you're here. We got a bunch of stuff to talk about. Hopefully we don't run out of time and they kick us out because I have to rent this place by the hour. No, I'm just kidding.

Basement.

That's my basement. No, it's my garage, actually. So folks, for folks out there that don't know Adam, he is an— I'd say probably AI architect is a good word. I don't know if that's what you call yourself, but I just named you that. So forevermore.

Solution finder.

Solution finder. Solution finder.

Solution finder.

That sounds a lot like what I do, only I do it in the dark with my hands and stumble on the solution. Oh look, I found one!

That's a fixer. That's the thing.

Okay, gotcha. Yeah, the difference between an engineer and an architect and tech support.

Um, oh yeah, oh yeah, oh yeah. Uh, well, look, yeah, um, I wouldn't just even say, yeah, like, it's, it's— AI is not a destination, it just became, right, the technology, the enablement piece of it especially around, like, you know, operations. Like, I— my original background is in, in sales operations and sales management and economics. Then one day I could not find a job, and that's how I ended up in recruiting. Uh, but like, I, I did that for quite a while, and the scale in like sales outreach and everything is like way different than, you know, what the average recruiter does. Um, so when I entered the recruiting world, I was like coming with this like mindset of efficiency, of do more with less. And, you know, started working for Amazon and do more with less became sort of like the mantra. And there is always a need for do more. And yeah, the numbers were crazy, right? Like, I know when I joined, they were close to 100,000 people or something like that. And when I left, Amazon was like 1.5 million. Like, okay, not all corporate, but still like crazy velocity growth. And operations matter when you're doing anything like people related over there.

Yeah, definitely. Especially if you're not like a name brand that people recognize and you have to work twice as hard to get the right kind of attention and things like that.

I think it has like even the other side of it is even if you have the name brand, it's You always need it. Yeah, but you always work, and it's sometimes like you wish you, you wouldn't because maybe the funnel would be cleaner.

No kidding, right? Stop the applicant flow! Too many applicants.

Yeah, yeah, like, and that's becoming worse, right? Hey Jonathan, hope all is good, buddy. Long time no see. Yeah, uh, Hopefully you survived the unfortunate news of the Bezos Reaper doing its another round, uh, like yesterday. Close to 30,000 people. I, I know that a bunch of PXT folks in Washington were hit, like 100 and something people, 119 people.

Turbulence. Turbulent times.

Yeah, is it turbulent or is it like default normal?

No, it's turbulence. I've seen—

when was the last time it wasn't like this?

I've, I've been, I've been in this doing this for 30 years, and the, uh, turbulence now is not as bad as the turbulence one other time. So in, in, in my— I've seen 4 cycles like this. One was worse, and we're not anywhere close to that right now. Um, I'd say we're not even going in that direction. The other 3 were not as bad, so it's the second worst out of 4. Uh, if this is the fourth one, so, you know, still beat by one. It's the second one.

Still pretty high. Like, I, I'm not sure if you want to be on the worst top.

No, you don't.

No, like, I saw the numbers.

That was pretty, that was pretty atrocious. But there are things that are going in the right direction. It, it is turbulence, and that doesn't make it any better, you know. I mean, turbulence is still a roller coaster ride, and it can be very stressful and very emotional. But, um, it's not— yeah, I don't want people to get panicked and think that it's permanent like this. So you're on the roller coaster now, it's scary, and you know, ah, but what if this time is different? If it's— it's not, it's not, it's not gonna be. They all felt like this, they all did.

Did they?

Yeah.

I, I, I kindly disagree over there, but I, I'll—

well, it's 30 years and you know, some of these are long while. Um, yeah, if I told you the things that I had to do for to pay my bills during the one that was the really big one, you would be shocked. It's, it's not like that. But what I mean by it felt like this, like if you remember during COVID um, and not like right at the beginning, but like once everybody was like locked down and people got used to being locked down, and then there was like all these like, we're gonna hiring, we're hiring.

Oh yeah, yeah, yeah, yeah, yeah.

People were like, when is this gonna end? You know, when is it gonna be over? And then it ends.

Look, if we just wanna— if we just— and like, did it end though? I would still argue that it didn't end because right now we just mentioned Amazon. They caught more than they did during COVID So did it end?

It did. Yep.

I think it's the— I think it's the same wave just doing, doing the rounds. But like, we are going— do you want to be spooky?

I've never— yeah, you can be as spooky as you want, but have you ever been— you've probably been to lakes that are like really quiet, you know, like a lake that's really still. And if you are the one that disturbs the lake, like, uh, you put a boat in, or you, you know, it's something, something big, not like a pebble, but like throw a big rock in, or you put the boat in, it takes a while for those waves to come back down to where the lake was still before. And at first the waves look organized, but then they start crisscrossing and it becomes choppy, and then it settles. right? So that's the same kind of accordion type of effect. You're right, it does take time to settle, but it did settle, and then somebody else made another wave.

And I think that's what's like those waves being interconnected. Here's what keeps me awake as a workforce researcher type of person and an AI curious type of person? Yeah, we live in this— we live in this society where value creation and value add, the vast majority of the time, at least in the current economical setup that we have in the vast majority of the world, is measured in some sort of like financial metric. Right. And that financial metric is like, let's call it just revenue. And if you are an employer, your biggest cost center is the human.

Yeah, of course.

If you're manufacturing, of course, not in every business, but not every business, but like in let's say like early stage SaaS or consulting or maybe even like front-end retail, we are talking about 40, 50, 60, even 80% plus exposure to like the cost, the humans themselves.

So I mean, you know, if, if I'm a business owner, direct labor costs, indirect labor costs, there's also the cost of humans. God, that's horrible to say. But, you know, there's all these other things that are associated with humans that are the hiring, the firing benefits. Yeah, exactly.

Put it all together. Maybe we just say that this is just enterprise and like, oh, that's a different world and SMB will get itself together.

I think it's everywhere.

It's everywhere. But like, look, let's say you are a mom-and-pop shop. Maybe you have like a family business, you have a partner, you have 2, 3 kids, you make a decent living, the business is scaling and you have to make the decision of like, oh, I can do more, but that means I need to hire a person. I need to hire my first like project assistant or something like that. But also, well, the one kid needs to go to college, so we need to save for piggyback. And the other two kids really like Disneyland. And you know what are the prices of Disneyland. So I have a decision now. Do I hire somebody or do I go to Disneyland or do I pay for the education of my kids? And now you're making this decision where there is this artificial intelligence thing coming in. That can go and do good enough, like at the good enough level, a lot of things. And that lot of things are like 3 days ago, Gemma 3 discovered a new and novel way for curing cancer that was validated by scientists. AI can do that. AI is better at human-learned stuff such as creativity. Based on metrics, at least the average level human creativity. It is better at recognizing emotions. It is better at mimicking empathy. And right now we are at the stage where everyday consumer AI tools, you can give it a job. It's decent enough reasoner. Let's say it has an IQ of somewhere around equivalent of— I think Gemini 2.5 was measured for an IQ of 128 or something like that. So you have a You have a generically decent reasoner with correct scaffolding. And you can, you can make the decision of like, how much can I outsource to this machine for good enough? Or I hire somebody, or I go to Disneyland. And you as a small business owner, you probably want to go to Disneyland as well. So you are also motivated towards pushing towards automation. And And at the same time, we also have investors, trillions and trillions and trillions of dollars going against full automation of economy. And we also have a lot of smart people, like some of the world's best researcher minds thinking really hard at this. And at the same time, we also have this, well, it doesn't matter which side of the fence you are on, it's still presented to you as a race. So there is this like no stopping. There is this internal pressure that everything needs to be automated. And we are stuck in this shoulda and woulda state of like, but what happens with us? What happens when human productivity—

If I don't do it, I'm going to get my ass kicked. If I do it, I'm going to be out of a job. If I don't do it, I have to hire somebody. If I do it, I can go to— So, but basically what you're saying, is frankly just choices. I mean, you're talking about choices.

And so what happens to availability of what you're choosing out of?

Yes, but there's still choices. What is the difference between this and the advent of the internet in the late '90s? Okay, so the velocity, but that's it.

But it's not stopping. Like, we—

the internet didn't stop either. What are you talking about? Yeah, but like the capability— No internet, no ChatGPT.

No, I would say no transformers, no ChatGPT, but I think we sort of maxed out what we have in the internet and cloud with like Web 2.0 and like we had—

That's not what I was saying. I was saying that there's, you know, the advent of technology, right? I mean, macroeconomically, if you look at the trends in history as far back as we can analyze them, because there's also stuff that we don't know about, there were technological developments that are outside of the realm of what we currently can measure or even imagine because they've been erased. But on the stuff that we know that we have record of, there were some pretty big moments. For example, the printing press, right? That revolutionized and changed things. Yes, the speed, I agree. But there were other things, the telegram, radio telecommunications.

None of them had that wide and across-the-board generic impact.

They all did. Electricity.

I would say fire.

There you go. So again, you're proving my point, right?

Like, I—

the wheel, electricity. There's always been some sort of, you know, some sort of change because of advancement. And so this is an advancement. And in technology specifically, it may not be an advancement in other areas, unfortunately, but in technology. And so we are dealing with decisions from that. And that's okay. I understand that. But again, that's turbulence.

So my point is decisions have consequences, man.

The decision, they always did and they always will. And some people make good ones, some people make bad ones. And you don't know sometimes if it was a good one or a bad one until too late. I was part of one of those in one of the previous cycles and I paid the price. And so I can tell you, you know, the turbulence— turbulence doesn't mean it's going to be fine. Turbulence could also end in disaster, right? I'm not saying it's only turbulence, forget about it. I'm saying it's—

tsunami is a turbulence, a big one, right?

Exactly. Yeah, that's right. So it's gonna end. It's either gonna end and you'll be fine, or it's gonna end in something else. But right now you need to put your seatbelt on. Oh, that's what I'm saying, right? You gotta, you gotta put your seatbelt on and then you make the decisions that you're able to make. There's nothing else that you can do. The panic and the the hysteria. I do have, uh, and I don't, I don't want to get into the economics because I think there's a lot of misinformation.

You're showing graphs, I will show graphs as well. Like, I want to do a graph war.

Economics? No, no, no, no. I just, I mean, I present only facts and I don't— I try to interpret or analyze the facts for myself, um, not repeat somebody else's, you know, statistical analysis because I That's just kind of how I'm wired. But I'm just saying that the facts presented as they are, there were worst times in my career already. There already have been worse times in my career and it was terrible. I don't want to go back to that, but we are not in that direction with this, at least not based on everything that led to that and based on everything where we're going right now.

What were the numbers back then that you were mentioning?

Well, so we were talking about—

And the bounce back, of course.

Yeah. So we're talking about 10% unemployment, right? There was during COVID it was 15% unemployment in sort of like a sustained way.

And then— I thought we had like sort of UBI then.

We had a peak, but I'm not talking about a peak. I'm not talking about what happened that one day, right? But like during a year-long period, that was affected. So during the, the 2007-2009 time period, unemployment was sustained very high. We had a negative GDP growth, and that is the opposite of our current positive GDP growth. It's just—

and what if those two don't correlate? What if— well, so here's where productivity and GDP goes two ways, and we are in a world where we have GDP growth like crazy. We have 20% GDP growth with 20% unemployment. The chance of that happening is bigger than zero. And I think that is—

Yeah, yeah. But here's the thing about these. That's what I was going to say is these facts, they present themselves in a certain way. If you put them as patterns and you attach correlation to them, you can tell a story. But what I can tell you is that the facts as they are, the GDP growth, the unemployment, they're presenting a different story. Than what people are presenting when they try to correlate them. And you're exactly right in your estimation that, you know, what if they are or they're not? That's the problem is, well, we don't know. And economists that are worth their weight will tell you this. I'm not an economist, by the way, but the ones that really know their stuff, they will tell you that these formulas don't always work. They're not the same as like laws of the universe, you know, gravity and speed of light. So sometimes these correlations work and predict Sometimes they don't. Right now, this is a change and it's turbulent. So I don't think that I can say, me, with any certainty that if you took GDP growth and unemployment and some other number and looked at it, it's going to behave the same way. I doubt it. It's not going to behave the same way.

Let me ask you one question. Probably.

Actually, I shouldn't say it's not because then I'm making a call. I don't know if it's going to behave the same way.

Let me ask one question because we know that there is one persona that will behave the same way. One persona? Yeah, one persona. Who is either the first or the last person in a job approval, job requisition approval chain? It's always finance.

Did you say the first or last?

It's always either the first or the last who approves the job requisition open. It's always finance.

But is it the first or the last?

They have a veto, right? Like, they— in what, either at the beginning or at the end, they have a veto. And now let's say you are, you are a CFO and you have a board, and we are in this version of capitalism where we mean market value maximization for the shareholders and not necessarily for everybody. So, right, you, you are, you are there to maximize profits for a board, and your biggest, uh, like suck on, on revenue and on profits is your human capital. So suddenly there is somebody who is telling you like, oh, you can cut 10% this year, probably you can cut another 10% next year, and probably you'll cut another 10% next year as well. And I think those CFOs, it's just, I think it's just bigger. And we are not even taking into consideration like robotics Like, uh, when, uh, when that hits and all the plans of all the major big box institutions wanting to do lights out, uh, fulfillment centers, that will be painful, especially in some rural areas where one of the only employer is either like Walmart or USPS or UPS or Amazon or you name it, right? And like, just during my lifetime at Amazon, we, we— the company went from from fulfillment center workers complaining that they need to carry a pee bottle because they need to walk too much, up to complaining that they are not walking enough and they are just doing the same repetitive tasks and the pee bottles went into the carts. So I'm just, I'm just quoting the news. I'm not representing Amazon or any of my experience. I'm just quoting the news of it, just if you know what they call, uh, because you know, they sign— they make you sign papers that you don't say anything So we went through there in like 4 years. And now they have this idea of like the lights-out fulfillment centers. Basically, fulfillment center, it will be operated like a data center. Like a data center, huge, huge, huge building to operate it on a day-by-day. These days, a dozen, few dozen people. You don't need the hundreds of people. The same goal is there with—

Sounds a lot like something that a guy named Henry Ford did once.

Yeah, but Henry Ford did it with the people. This tries to remove the bottleneck, the people. I think it's—

and I think it's still changing.

It is changed. It's like nobody says that it's changed, but do you think— how many, how many humans do you, do you know in your environment that loves change and adapts well to change fast?

Yeah, but most people don't like it. Most people want things to stay the same. But in times like that, so this comes back to the turbulence. The turbulence is causing the change, and I'm, you know, there's going to be, there's going to be things that we don't expect. I'm not being able to predict like what's going to happen, but what I can tell you is I'm gonna do the exact same thing that I did all 4 times, run ahead. And it worked. And it took longer on one of those times. If this is, if this time turns out to be more difficult than that one, um I don't think that's going to happen, but I'm just making a prediction because it could. But it worked 4 times, so it's going to work again for me. However, it doesn't mean that it's the same thing for everybody. And for me, it was constant evolution, constant reinvention because I made myself comfortable with change. So, I think you're right on something there that the level that you're saying is, I think what I would like to do is I would like to say it's not a good idea to panic because that gets you nowhere. So, instead of panicking, right? Put your seatbelt on and do something to get prepared for the change and maybe do two things because if you think it's going to go one way, do 80% of that but 20% of something else that you think might happen just as a hedge your bets because you never know, right? And that's what I did. So, I ended up, you know, just as a personal example, one of them was a career. So, I had 20% of my energy invested in switching careers because I thought this is all, you know, this industry is over, recruiting is over, it's gone. Believe me, it was a lot worse than now, a lot worse than now.

How many switching careers opportunities did you see around you?

That I've done or that I've seen?

Availability. If you want to switch careers—

There's no availability, zero availability. You have to make that happen yourself. Nobody's going to give it to you. But that's what I did. So, my point is that I had a small percentage of me was saying, I'm done. I'm quitting recruiting. This is it, right? Because I'm not going to make it. And the other 80%, I was like, all right, let me reinvent. So, I reinvented what I did and I also looked 20% of the time for a different job. Fortunately, Reinventing What I Did ended up getting me out of it, but I did have a backup plan. It was not the best backup plan, but I did. It's the same thing today. So, I have a backup plan, but I think the turbulence is still going to end at some point. The problem that I have right now, and this is actually one of the questions that I had prepared for you, one of the, you know, I don't usually make a lot of questions in advance, but I did want to ask you this one thing. Why do you think—

On the spot, like perfect.

Yeah. It's a segue into it. Why do you think— going back to just bear with me for a second. In the past, I don't want to go back to the wheel and fire and telegrams and all. Let's go back to like at least my lifetime. In the past, when the internet came around, lots of people were excited and lots of people were skeptical. Oh, it's not going to work. It's terrible. It's a bad idea. There's going to be crime. There's going to be this. So, there was a good balance of both. And then, whatever happened, it settled. Then, social media came around. And again, there were people that were like, "Oh my God, this is awesome." And there were people like, "Oh my God, this is terrible. It's the worst thing to happen to humanity." And there was skepticism and there were people that were full steam ahead and there were people that were like, "Let's pause for a minute. Let me point something out." For example, there were companies that had social media policies. When was the last time you even looked at a social media policy? right? We don't need them anymore because it's now part of media. But there was a time when we were scared of it. So, let me ask you the question. If that's the pattern, typically new technology, some people are scared, some people are excited. Why is it that with AI, this, which by the way, as a side note, what we call AI is really not AI. It's not. No.

And that's the danger. Right.

But let's not get into that, right? Because you're smart enough and I respect your opinion and I know that you know it's not really AI. So, I don't want to say AI and have you think, "Oh my God, Shali thinks it's real." It's not really AI. But what this thing is that we're calling this generative, whatever, this thing that we're in right now, why is it that every other time technology came around that was like big change like this, we had skepticism and we had debate? Why is it that this time nobody is asking any questions? It has been wholesale swallowed by the public, the media, everywhere. People are absolutely in love with it and it can do no wrong. It's like if I was to make an announcement today, and I'm going to use your own example, if I was to make an announcement today that I'm some biopharma or whatever, "Hey, our company just cured cancer," the whole world will be like, "Oh no, you didn't. Hell no. Show us. Prove it. We want proof." Hey, but you just said Gemini solved cancer and everybody's going to be like, wow, it did? No way. Really? Nobody even asks if that's real or not. Why are we accepting AI or whatever this is so wholeheartedly? Why is everybody, like you said, wanting to, you know, companies are telling their employees, use it or you might lose your job learning it. It's like being pushed on us. It's being— and it's not the CFO because even the CFO is being told to go do it. And they're doing it. Everybody is swallowing AI like it is the world's tastiest meal. Nobody is stopping to ask, where did this come from? And I'm reminded of the story of the animal that tasted really good. You know this story?

I did not grow up in American pop culture.

Oh, this is a worldwide story. They have it in almost every culture. It's an animal that tastes really good and people start eating it and they're like, this is the best tasting animal, this is awesome. And then when it's too late, they realize that the animal reproduces by consuming you when you eat it. So by eating it, you become the animal and that's how it spreads. And oops, it's too late. We are that fat animal eating AI right now. Why? Why did we not stop and ask any question? It's like we are just so— like, I feel like a fish, you know, I'm like on the hook, you know. Do you understand what I'm saying? Everybody, why?

What if, what if the answer to that question is because it works?

Sure.

In its own clumsy and—

sure, yeah, okay, good—

in incorrect way.

It works great, but nobody's asking.

Well, if it works, it works. And I wouldn't say nobody is asking. We are here on October 31st, Friday of 2025, and we still have people who, despite looking at like usage report benchmark data, like actual capabilities, they are still Some people rejecting reality. Are there people who say that this has a ceiling and maybe it's not going to be—

That's rejecting. So I understand what you're saying. That's basically saying, oh, it doesn't work. Like naysayers or whatever, deniers. I'm not talking about that. I'm saying that we had so many people investigating the internet, that what you can do, what you can't do. There were standards. We had so many people investigating social media. There were social media policies. With AI, I know firsthand with my clients, most of my clients don't even have even this much of an AI policy, right? I teach ethics in AI and the biggest gap is most of the companies that I talk to, at least 80% of them, they don't have any kind of guidance policy standards committee. This didn't happen with social media. This didn't happen with the internet.

Why the speed of adoption? Like, that's, that's why I'm saying is that, and that is the difference between all the previous times, like, change happened is not just like the virality of it. It, you know, the meme of like Will Smith eating spaghetti.

So it's, it's Asian.

It's, it's like how it, how it's shown, how fast it evolves and how, how fast it becomes actually usable. So it— oh my God, from the second that you change your approach to like, oh, is this good? Or like, haha, it could not answer this correctly. And you switch your mindset to the, but wait, how, how can I make it do what I want? What is needed for that? What sort of system of controls and balances? What sort of system instructions? Extra APIs for moderation or sanitization, like, you know, the good old principles that we have.

Yeah, that's what I'm saying.

It works. Like, maybe that's why. And because it works in bigger and bigger and bigger pockets, that pushes pressure on those other pockets, that pushes a lot of money into the system. And if there is a lot of money into the system, if it pops, it cannot pop because it will hurt a lot of people. So it's like self-generating.

Well, that's right. But that's actually a really good— so I can accept— I asked the question, so I'm, you know, I'm going to accept the answer that you're giving, which is because, because it works, that's why nobody is, you know, really pushing it.

How many— let me ask you this— how many people use the internet, um, between like, choose a random 2 years in the first like first public launch? Like, when, when did your first, like, AOL thingy come in the United States? What are we writing, late '80s, early '90s? You know, the discs and all that. When did that start?

Oh, you know, there was a company that was before that, um, com something or other. I remember, you know, I was very, very early in that, and it was 1993, '94 when, when computers first started, like talking to each other. And by '96—

and so those— let me just stop over there.

2 years.

Those 2 years, did the internet and the computers gain— like, and we're just talking about like OpenAI in this case and not about like machine learning and AI in general— did it gain billions of users? Or like, just OpenAI is now, I think it's like around like 700-something million daily active users. Like, the adoption and exposure rate is the big difference. And it's—

I accept your answer. I accept your answer. It's a flywheel, which means, by the way, to explain what that is, not everybody necessarily follows that. Flywheel means that the work gets easier the more that it— so the faster the flywheel spins, the easier, like a—

like, and it's multiplier, right? Like more people, there's more flywheels and influence.

So let's put this under the win column. Granted, it's— let's just say the speed of adoption, um, because it works and so on. Okay, I get it. But my question is, why are we not, even though we're doing that, also still doing the other thing that we normally would do? And I can tell you where it is happening because I have access to inside information of where it is happening. But where it is happening, it absolutely has to happen and there's no choice. However, they absolutely have to do it and they always do it. So, it's the only place where I see it. In other places where it's not like that, that restricted, that strict, there's just not a lot of, you know, people are not questioning it and we're completely unafraid of the consequences. For example, and this comes from the fact that I, you know, teach about the, the ethics of this. For example, if we give the AI whatever, you know, this tool, this automation tool, we give it access to a certain scope of information. It's very difficult, if not impossible. Some people would say, I don't know for sure for that, but it's possible that that information cannot be retrieved or retracted. In other words, we teach this, we give this to the system and now it knows this. And some of that information is really not supposed to be shared for other reasons other than efficiency or productivity. Let's just call it safety, security, whatever. So why are we not asking those questions?

I think we are, but the pressure in the system is so high that you're like, shush now.

You're told to ignore it.

Or like, just shh, don't ask.

It's nothing. Don't worry about it. Look the other way. Look the other way.

Do you remember the go fast, break things? It's nothing new. It's just the speed is different.

No, it's just— yes, that's right. But so what you're saying is that in the past, because it wasn't as fast, we had more time to react.

Yeah.

I get it. Okay. So now two answers, right? The first answer is why. It's the speed. And the other answer is why are we not looking at this? Maybe we are, maybe we're trying, but again, the speed means that we have less time to react. And so that, you know what that does? That makes me even more concerned for the turbulence because, and that may be the point that you were trying to make earlier, which I just connected, because that means that the turbulence is even deeper or whatever. There's more turbulence than we're feeling. There's turbulence that we're feeling, and then there's other turbulence we're not even feeling that's out there that's being caused by the fact that the people who normally keep us safe from things going wrong are not able to keep up with the speed. So do you think we should slow down even a little bit?

I don't think there is a— are you familiar with the AI 2027 research initiative? They went around and asked 100 prominent AI researchers around best case and worst case scenarios. And the problem that humanity has is that we think that worst-case scenarios have a smaller chance of happening just because we don't like them. But the worst-case scenarios have like the, you know, the same amount of chance as the, as the best-case scenarios. And these are, these are people who, who were AI skeptics or anti-AI. There are people who created benchmarks to prove that this is not yet close to humans and we can show how unique we are. And these people starting to ring some bells really loudly and some of the people are not the ones to slow down and think a little bit. Right. That's what I'm saying. I think the pressure in the system is so high that everybody agrees that maybe we should consider slowing down, but we just can't. We are rolling.

The snowball is going down the hill.

Yes. And we are like, there is a race and there is sovereign AI and there is innovation war and all those things are like the external forces are there. So that's happening. But let me just finish what we're not talking about. And that is the alignment.

Is that from the 27 or that's you?

Everybody. Like, I think the 27 also says that alignment is a big issue. I also say that not talking about alignment is a big issue. Like, let's say some say oversight needs to start now. Oh, yes.

Like, we need to align around oversight, whether or not— so maybe not slow down, but just start the oversight. Gotcha.

How do you avoid singularity if you don't tell your super intelligent system that like, "No, we are nice people.

Like, you care for us." I don't know if we want to avoid singularity or not. I have no idea. You know, I saw a funny meme the other day. It was a laboratory. I think it was a college or maybe high school laboratory, mechanics, robotics. And, you know, they were working on these little mini robots that were, you know, sort of micro little AIs. And the student turns the robot on for the very first time. And then the robot wakes up, looks around, and then unplugs itself. So I don't know, maybe just it's a joke, but yeah. So I like what you're saying. Oversight needs to start now because we can't slow down.

It was good, right? Like, there will be a time. I, I, me personally, I think next year will be the year of the memory. If this year is the year of the agents, uh, next time, the year of memory.

Say that, say more.

I would say that this year is the age of the agents and year of the agents. And one of the big problems that comes with the agentic system is the non-existing layer of consistent memory. Like, as an organization, you have some sort of like tribal knowledge, you have like wikis, like you have people communicating with each other. And right now in agentic systems, this is a big problem and it's solved with like orchestration. Memory management layers and all sorts of like funky tech workarounds. But because this is the next big problem, everybody's working on it in one shape or form and good memory solutions for LLMs are coming out. So we suddenly have people or like you have not people, agentic resources, where probably in the workforce, every human will need to manage a group of resources, and the resource will be universal human and maybe AI. And if we decide that these two work setups don't happen beside each other in vacuum, in instead they like collaborate and co-work and co-create, that means that you will probably somewhere in your command of chain maybe have an AI boss. Who knows? But now, how many bad bosses do you have in your—

please, for the love of everything that is important to you, do not ask me that question in public.

So I ask you, if an AI that is well aligned has your best interest and does not—

it could still be a bad boss—

and it's aligned on not being a bad boss, maybe it's a good thing to remove humans from some parts of the process. Because this whole thing of human in the loop, this is a human issue. This is— we want it, like, we want it, not the AI.

The human in the loop, I think, is, uh, the part of the governance. But I don't think human in the loop means human in every little piece of the loop. Okay.

Like somewhere.

Somewhere in the loop. Yeah.

You know, it becomes an accountability.

We have already fully automated. There are systems that are fully automated before, you know, because you can build them in such a way where they don't need supervision, where they're—

But those were probabilistic systems. And these now we're dealing with non-deterministic systems.

That's true. Which, by the way, you're going to have to explain that because not everybody knows what that means.

Oh, apologies. Apologies. So deterministic systems and probabilistic systems differ in a way that current gen AI is non-deterministic, right? Like it doesn't know what is the end of the sentence when it starts to speak the sentence. There are workarounds around that. But anyways, it doesn't— when it starts generating, it shouldn't know what's the end of the generation.

That's the whole point of generation is not—

that's the whole point of generation.

Not having a determined correct answer means deterministic.

Versus probabilistic systems like good old algorithms. We've seen cars, when they see this sign, turn to the left 900,000 times out of a million. So probably we learned probabilistically that when, as a Tesla, I see this sign, I should turn left as well, because that's how probabilistic systems work. And I think the governance of those two is very, very different, because one you can trust—

manage by exception versus—

yeah, versus the other one is just an endless line of Guidelines and extra checks and balances. And depending on how much cost and latency are you willing to endure, you can make your non-deterministic system be like correct 90% of the time or like 99% or 80% of the time. But let me ask you, how many times are humans correct? And I think that's what you remember, Watson.

The cool thing about— were you, were you like in the business back then when Watson first came out?

Which one? The IBM Watson?

The first? Yeah, not that, not that Sherlock Holmes one. No, yeah, the IBM Watson. Was that, was that before your time, or?

I think it was before my time, a little bit.

Well, so let me just really quick— when Watson— time is just a concept anyways, not to the AI. That was a joke, by the way, for everybody who doesn't— AI today does not have any concept of time.

So, um, if you look at You can give it one, which is so weird. You can give it one, which is so weird, but not have it.

You can implant it, right?

Yeah.

So, it's non-time, like it exists in all time. But anyway, so when Watson came out, what made it so special and so super beyond was a leapfrog. And a lot of technology today in what we call AI is dependent on this. The machines that they were trying to build to make these deterministic predictions and things like that, they wanted them to be 100% accurate. What the Watson team came up with was exactly what you just said. Humans don't do that. Humans don't calculate a 100% answer before they answer every single question. Very often, we answer given an approximation of what we know because we're the Tesla that turned right 900 out of 1,000 times. So, if we are asked which way to go when the arrow is pointing to the right, we'll usually say to the right even though we're not 100% certain. So, Watson came out and said, when we have a tolerance of variance, right, that reaches 75%, so if we're 75% sure because of the data, then we're going to go ahead and say, yeah, or whatever the 75% means 100% basically. So Watson would answer questions that it thought was 75% probable to be correct, and they were sometimes wrong, but most of the time they were correct, just like a human. So it's what you just said. Humans are not 100%. And that's important to know because we expect ChatGPT to be 100%, but we're not 100%. We're not.

We are painfully mediocre. Neither are we. Like we are painfully— like if I want to—

Hey man, speak for yourself. I am not mediocre. I am well below average.

Right? You can say that. I won't say about that about myself. But like, that's the thing that all the capabilities that we think that we are human performance, like recognizing speech, recognizing language, image reading, predictive reasoning, some sort of machine intelligence already reached the human performance level, which is a lot of times, let's be honest, painfully average. There is a reason that a bunch of companies fire the bottom 10% each year. And there is a reason that the only top 10 or 20% gets promoted and gets the big bonus. Guess what's in between? The AI first only needs to be better than that, like bottom 10%. It only needs to be better than the starter 10%. And we see this with ADP data, right? The canaries in the coal mine research that came out a few weeks ago.

They look payroll way too far, man. Holy shit.

I'm watching this trend happening and I'm like, maybe we won't go into the direction that this is heading. But we are not doing anything neither to course correct or stop or slow down. Like, we are pedal to the metal. Let's freaking go. And that's what you said as well, right? Like, nobody's saying like, stop right now for a second.

And I'm just asking why. I mean, I'm not—

because we're fine. Like, why should we stop?

Because we stop almost every other time. And for some reason, this time we're not only not stopping, we're not even thinking about stopping. Yeah, we— people hit—

and why aren't you stopping? Why aren't we stopping?

That was my question.

Well, the difference is that then GDP was also down. Now you can maintain the same or higher range of GDP growth with the lower amount of productivity inputs in there as people. Again, I think that, that's why we like— what if it works?

What if it works? But it isn't.

Together with—

it isn't working because 80% of the size I need to see sectors rising that are not rising, that should be rising, that are not, which is ironic because technology, which should be going down, is going up, which is weird because it goes counter to the logic of all these jobs that are being eliminated. So it's like we're getting rid of all these IT people, yet we have a growth in technology. We're not getting rid of nurses, yet we have a decline in nurses. It sort of doesn't really follow the formula.

Let me ask you this. It doesn't really follow the formula. It doesn't follow the trends that we see. But because previously it was like this, sure, it will be the same this time. No, I know that.

That was my point before. I don't think that these numbers predict, right? So there's a larger macroeconomic analysis that I do trust a little bit more than GDP and all these other things that tend to be very country-centric. And that is at least the way I've been calculating the purchasing power parity and the ability to, let's just call it the economic spend freedom that you have. So, very similar to what you were talking about before with the choices. Do I hire someone or do I go to Disney? This is a kind of a mathematical way to try to measure that.

There is another way of that as well is that The reason that we're saying that it's not like we don't— it's not working for humans as, as we thought so far is that the projects that humans got like big investment in were always like venture scale projects that got the investment in so you can go and actually build that. And right now, that venture investment maybe is not going into the people, maybe is going into the infrastructure and into the automation, but it's here. It's in the— wherever you draw the borders of like the Western world, it got into the economy somehow. It, it got into the economy. But like, you start to look at like parts in Africa, parts in Latin America, parts in Asia, where here it's elimination of jobs. And like, maybe it's just a scapegoat, but it's still happening, right? Like, it— for one reason or the other, there is an elimination of jobs or an elimination of new jobs being created. All those new jobs that are being promised, where the heck are they? I sent— I seen one prompt engineer. Turns out it was a fad and it's not even existing anymore. So where are all the new jobs, right, like that, when we have hundreds of thousands of jobs disappearing all the time? But that's just in our bubble. You look at the other bubble, suddenly everybody is like a vibe marketer. SMBs can, you know, like fly in their like small communities. Everybody can be a small like white coder and create IT solutions for micro problems, right? Like when the internet came, you had all these internet millionaires. Where are the financials? Where are the financials?

I don't know. I don't know. It's a good question. But when the internet was, you know, first, all the people that were like early adopters, like cybersquatters, you know what I'm talking about? They would buy domains and then sell them. And yeah, I mean, there's going to be opportunities.

They still do that.

So, there's going to be opportunities for that in this economy as well. But I want to— I know that there's—

Let's end on a high.

It's been kind of a— we said roller coaster and we basically maintained the roller coaster the entire time during this episode. But one thing that we didn't talk about, and I do have two questions everybody needs to get asked on the show. That's just a hard rule. As much as I love you, man, you're not going to escape without that. So, leave room for two. But I wanted to kind of end on something, the conversation so far on something that may be a little bit more, I don't know. I don't know if uplifting is the right word. But tell me, what is one rule that you would tattoo on every recruiter's desk right now?

Curiosity.

That's a rule?

Yes, that's a Hardy Force rule. If you don't come to this desk with curiosity, get the hell out of there.

Holy crap, dude, that's like my number one rule. But I mean, not— it's my number one— I never thought of it as a rule. For me, it was just always my drive.

Like, don't enforce it, it's not happening.

But I don't have to enforce it because I'm hungry for curiosity. Are you telling me there are people out there that do not have any? I—

can you teach common sense? Uh, it's like the same.

As well as you can teach Good taste.

You can enforce all of them, right? And that's how you enforce curiosity.

Enforce curiosity. Okay.

Enforce curiosity with questions.

I like it. I've never thought of curiosity as a rule.

Because what does curiosity as a rule enforce? It enforces change and thinking and creative thinking, critical thinking, asking questions, asking questions.

There you go. Nobody likes to ask questions.

And if you are not curious what's ahead of you, you'll never get there.

I don't know, man. That's right. I love that. This curiosity is a rule.

Very, very much wisdom over here on the front.

I like that. All right. So the two questions that everybody gets asked. The first one is—

Oh, I thought this was the first one.

No, no, that was the— that was the— I want to end on like, you know, give people something to stay curious. Like, stay curious, my friends.

If, if the ass is hitting the fan, the only— and the change is constant and you don't know what's—

just ask.

Just as in, you know, in The Walking Dead, the first way Don't ask ChatGPT. Don't ask ChatGPT. The first wave will be slaughtered potentially, but the people who are curious to find solutions and find answers to the problems will find a way. And maybe it will be a race of curiosity because there is only room for 1 out of 10 or 1 out of 100.

I don't know, but that's a good rule.

I like that rule.

Who knows?

But run ahead.

It's a very good rule. Stay curious is number 1 rule now. It's not quality. It's not safety.

It's stay curious. Be curious.

But you know what, dude? I want to rephrase my original statement about in the times of turbulence, put your seatbelt on. I was going to say after this conversation, I don't want to have a seatbelt on during the turbulence. I want to have a 5-point harness and a helmet. But anyway, all right. The two questions are, first of all, who should be our next guest?

Who should be your next guest? It's easy. It's my partner in crime. I'll talk home, Clara, because she talks a lot about mindset. And How to Learn in an Ever-Changing World. I'll do an intro for you. You might even know which one.

Done. Awesome. Now, you knocked that one out. Awesome. And then, the final wrap-up question. For a lot of folks, this is a recording or a podcast. And what I want to do is end with something that creates, let's just call it curiosity. And so, give people a question to ask themselves, not an answer, but Tell us a question we should be asking ourselves.

Any other guardrail or that's it?

Just something that we should ask ourselves. It doesn't have to be about the show, but make us think, force us to. You already said rule number 1, curiosity. Great. Now it's like rule number 2, question something about yourself. It actually doesn't even have to be about yourself. Just something, a question that you should ask, that we should ask. Sorry, not you.

When was the last time you could have taken action but you didn't? And how annoying does that feel?

Okay. I'm glad you added that. When was the last time you could have taken action but didn't and why not? Or what did you learn from that?

How does it feel to look back now that you know?

Ouch.

Ouch. Because people need to face themselves and need to face the consequences of their own decisions. And that's how they will realize that nobody's coming to help. You need to initiate that change. You need to be a self-starter.

I have a friend that is kind of gullible and easily influenced by popular opinion. And very, very, very early on, this friend took $500 that they had made in some kind of— it was a bonus, extra money, right? Pocket cash. And instead of buying, uh, something frivolous or whatever, they invested in this very little-known, um, stuff called Bitcoin. And they bought— I think it was— I may be exaggerating, but it might have been 500 Bitcoin or something like that. It may have been 50 Bitcoin or something, but it was $500 to buy some Bitcoin that wasn't worth very much. And then they lost the email account, or didn't lose it, but they forgot or whatever, for years, right? 5, 6 years went by, and they're like, hey, I, I remember I bought some Bitcoin. I wonder how much it is. And they went and recovered the account and went in there, and turned out they had made a ton of money.

Capital gain tax must have been a beyond that.

I don't know what happened if they gain taxed it or gate, or maybe it's still sitting there, maybe. But it was so early on, they say the answer to that problem. Exactly. Nobody really would lose money because it was so long ago, right? But if you were to do that now, that's not recommendable. Um, but this was like before it even— so, you know, if somebody had come to me and said, hey, you should take $500 and buy on Bitcoin back then, I would have been like, hell no. But he did it.

So don't forget that's something that it's It's somewhat out of control, out of your control.

I didn't even know Bitcoin existed. This was before I even knew about Bitcoin. Yeah, that wasn't a decision that I made. I'm just saying if it had come to me, I would have probably said no, and it's the kind of thing I would have been regretting myself.

At the time I was looking at it because there was this other website that, uh, is a clothing material and a road— I won't say out the name out loud. You're on the internet, um, and they were using Bitcoin for payment services. And that's how it first come into the, into the news. Yeah, yeah, yeah, before the times of legalization of it.

That's funny. Yeah. All right, well, ask yourself, when was the last time you had a chance to make a choice and you didn't, and then regretted it? And what did you learn from that? Dude, it's been awesome. It's been really fabulous.

Make that learning more active. No, just ask yourself, what could I have done? But like, what could have I learned today to make it easier tomorrow?

Love it. Dude, I'm going to have to— you're going to have to come back to the show because I think we got a lot of stuff we can cover. But let's keep in touch and hopefully I will see you back here. Until then, everybody, please be kind to yourselves. And take care and happy Halloween for those of you that celebrate it. Doesn't matter if you do or you don't, but it's still fun either way. And hopefully you come back for Brian Sherlock next week here on Shelly's.

If you're curious, you will, right? Like exercise that curiosity muscle.

You should be curious. That's right. You should be. All right, everybody. Take care.