AI's Limitations and the Role of Human Judgement — Stuart McClure Keynote, TIF 2026
WethosAI — https://wethos.ai/resources/videos/ai-limitations-and-human-judgement-tif-2026
Source video: https://www.youtube.com/watch?v=BfJNwYAGyz4

Hope everyone was able to catch up and um grab a snack and hopefully meet some new friends today during the break. It is my absolute pleasure to announce our keynote speaker, Stuart McClure, a true pioneer in the world of cyber security. Stuart founded and was a CEO of Cylance, the AI-driven cyber security company which was acquired by BlackBerry for $1.4 billion.

And a special shout out to the Canadians in the room who know how great BlackBerry was. Uh, long before AI became today's buzzword, Stuart was already applying it to solve some of the most complex challenges in security, helping organizations stay ahead of increasingly sophisticated threats. Today, as AI continues to transform every industry, there's no better There's no one better to help us understand both the opportunity and the responsibility that comes with it.

Please join me in welcoming to the stage Stuart. All right, good morning everybody. Thank you so much for joining and attending. Hopefully I don't have double bubble here. Um, all right. Well, a quick a quick survey I think of everybody would be helpful for me uh this morning. How many of you in the audience have been on an airplane? I I figured I'd get full participation.

Okay, now keep your hands up. How many have been over the Pacific Ocean on an airplane? Okay, now how many of you uh have done it in the middle of the night? Okay. Now, how many times when you were doing that did a hole open up in the plane and pull out people? I knew I always win that game. Um, yeah. So, look, today what I want to be able to share with you is a bit of my journey and the why.

For those of you who know me, um, profit is always my second or tertiary or fourth agenda. Uh, when I build something, I always, right or wrong or indifferent, anchor and begin an anchor on mission. And it's that mission that I was gifted many many years ago. So today I want to share that story a bit and then tell you what I'm doing next so that you get a sense of the big challenge that I have and I need everyone's help because this is probably one of the biggest challenges not just that I've set out to achieve but that the world has.

So, I might fail spectacularly, but that's what I love about it. All right. So, you probably don't recognize anybody on there except maybe the cherub in the middle, but that is me at age 19. And that is the red carpet club. I wasn't supposed to be drinking, by the way, but don't tell anybody. But that was the red carpet club in Honolulu, Hawaii.

My mother is on the far left. My brother is to my right, our left. He was 10 years younger. Is 10 years younger. And the rest of everybody that was there was in business class on this flight. So, as you can see on the right hand side, the person individual with the shiner, he was hit with debris. The gentleman behind him, believe it or not, was an aviation attorney.

The worst thing you could have on a plane accident. And um all of us sat there waiting for our flight, our first flight after the accident. So, we were all very nervous. I even think my brother sneaked a couple of sips of the bourbon sour that someone was drinking next to him. Anyway, so for that to give you a little bit of more details on what happened, basically a 20 by 40 foot hole opened up at 23,000 ft.

Okay, I was in row 41. Um, it sucked out not just bags, chairs, um, people, but pretty much everything that wasn't bolted down. So, if you were standing there as a flight attendant looking what the heck happened, you would be looking at this at 23,000 ft. The engines and much of the debris went into engine 3 and engine 4. The who is what matters to me.

So who didn't make it and these are the souls, the nine souls that did not. And many of them were Kiwis and Aussies. Um but the heroes in all of this were Kevin and Susan Campbell in the top right who were the are the parents of Lee Campbell who passed away that night. And believe it or not, we were offered the upgrade seats to business class.

So my mother would have been in that seat which was what was that 13G and I would have been up in the first row is what we believe as we got those tickets and we were sort of scrambling to try and move from 41 up to the business class and in this last split second of it taking way too long for us to get ready I turn to my mom and I go what thought why are we moving like let's just we have a whole

row here in economy let's just stay and it was honestly that choice which I wish I could tell you I had some sort of you know presence from you know the heavens and the cosmos I didn't it was just gut instinct right it's all like intuition all right so why did this happen after we got on the ground everybody thought it was a bomb obviously right Lockerbie had just happened like a month before two months before.

And so we all sat in the terminal for the entire day getting interviewed by FBI agents. But to discover what truly had happened, they had to go and recover the door, the cargo door. And at first, of course, as you probably would imagine, the NTSB and the FAA said, "Yep, just typical air um infrastructure failure. Uh it's an older model, you know, so many hours and so many flights and they wanted to resolve it with a simple sort of wave of the hand."

But remember Susan and Lee or Lee Campbell uh as parents, they were tenacious. They decided that they weren't going to accept the basic answer of what really happened, the why. And they set on a path all around the US. They were Kiwis. They are Kiwis to find the truth. And only because of their persistent investigatory skills and capabilities and just never giving up did they actually get to the bottom of it.

There were two fundamental flaws in the design of that 747-100. The first was that as the electrical wiring from the cockpit would hang it and with multiple openings and closings of the door, it would often push up and bang up against the metal frame. So the wiring on the the wires or the sheathing on the wires became frayed, which caused a short, right?

Because you know, electrical t touches the metal and you're pretty much shortening that thing. But of course, the Boeing engineers that designed this are no dummies, you know. They said, "We have to have a hooking mechanism." So if if something does happen to this door in mid-flight, it can hold it. Now, Boeing engineers are incredibly smart, incredibly intelligent, but are they perfect?

No. So the the door itself and the frame itself is made of metal, right? Steel. What do you think the hooking device is made of? Don't say plastic, although that'd be a decent choice. Aluminum. So what happens to aluminum when steel runs through it? It pretty much melts like butter. So that's what happened. The door opened in mid-flight. It blew right past the aluminum latching which you can see right there in the circle and the door opened despite um a design that was supposed to prevent it.

So all of us survived 346 and nine um did not and this was the chaos that sort of came through it. Now I don't know how people had enough wherewithal to take pictures to be honest with you. I I was not I did not have the wherewithal to go and look for my camera, but some folks did and and luckily they did so you could capture what was happening.

So for about 19 minutes um the plane simply went down we thought over the Pacific Ocean and was going to ditch into the water. But thank goodness the amazing pilot had better ideas. So I was gifted with the uh pleasure of actually giving him a call. I found I got his phone number. He was living who just retired. It was his second to last flight before retirement.

I called him in his Lake Tahoe home. Caught him and I said, you know, I believe it or not, I'm one of the passengers on flight 811. I I really thank you so much for what you did. I don't know how you did it. Like all the news articles tell us that there was just no way. It seems impossible, but I knew if I talked to the talked to the pilot, you know, that the engineers in effect, you would give me an answer.

And his answer was, Stuart, I have no idea. So I said, well, what does that mean? You have no idea. He's like, "Well, when we got down on the ground, they pulled the cockpit flight recorder, which has all the telemetry of the plane, all the decisions made by the pilots, and we put it into a simulator in Honolulu. And we spent a week uh both myself, my co-pilot, and an independent pilot simulating that flight to attempt to get the flight to land as we got it to land that night.

And in over 200 simulations, every flight landed in the ocean. So, of course, once I heard that, I didn't need to hear much more. I said, "Okay, thank you. I know what I need to do now. So that really was, I think, the kernel, the seed for me to think about how do we prevent the preventable. That was clearly a preventable problem. I mean, nothing was like magic.

It wasn't like, you know, a laser came out from an alien spaceship and burned a hole through our that I would have probably couldn't have prevented that. This was highly preventable. So, why couldn't I apply that to the world of technology? And that's what I set on my path to do. So, the first company that I started here in Orange County is Foundstone in 1999, right out of Irvine, uh, right down by the spectrum.

And that was the sole mission of the company was to find vulnerabilities in computer systems and networks and prevent them by knowing they're there and either fixing them with patches or preventing them um through u configuration changes and layers of security. Then that company was sold to we sold to McAfee 2004. McAfee was a perfect alignment for me because it was now even broader not just looking at sort of network system vulnerabilities but full stack.

Um I eventually took over the office of the CTO globally. We sold to Intel at the time and I left in 2012 because I realized I had gone as far as I could with McAfee and the entire antivirus industry. So antivirus, as you know, you you probably know by now, it's a look in the rear view mirror to drive. It it does not look through the front windshield.

What what I mean by that is the only way that you can be protected is if your neighbor was hacked. That's it. If your neighbor was hacked, then fine. We'll forensically understand why. We'll build a signature detection and we'll give it to you and hopefully in time before the bad guys hack you. So, this is a, you know, trailing indicator system like there's no tomorrow.

And I knew that there was a way to predict and prevent this stuff. And at the time I didn't know it was AI. I knew decision tree learning. I knew computer learning because I had been um gone to college for computer science and realized this might be the way we can actually predict and prevent forever. Not just a point in time but for years in advance.

And so that's what Cylance was all about. Uh we raised about I mean 166 million um of raw venture capital. We had 100 million in debt but never really used any of it. Um and all of that started to explode. The concept of applying AI predictively preventatively at the endpoint at the phone to be able to prevent not just today's attacks but tomorrow's unknown attacks really started to take off.

We were just about to go public when BlackBerry came in and promised a synergy that would allow us to take AI down into the embedded world. So embedded cars, embedded medical devices, embedded operational technology, OT, right? Um, all of that was one of the biggest promises and why we it made sense for us to merge. Now you see me holding.

Does anybody know what that is? in that picture. Any guesses? It's a It's a little antenna, right? It's called a Yagi antenna. It's it's a directional antenna. So, if you point it at something, you'll pick up signal, RF signal for many, many more feet. Um, we use that on stage to hack into an insulin pump. And this was in 2012. and and we were prolific in going after hardware vulnerabilities.

That vulnerability was a feature inside of that particular insulin pump and that feature was present on that device for years before we found it and a good 10 years after we discovered it. Not only that, that vulnerability was present in every device from that same manufacturer. neurobrain stimulators, implantable cardiac defibrillators, everything.

So my mission was always to expose, to educate and to bring people onto the mission of protecting and preventing and wrote many, many books on this topic called Hacking Exposed. So by the way, has anybody Is anybody willing to say that they've gotten hacked before? Anybody? The brave ones. Okay. Good, good, good. All right. Yeah. And you know, identity fraud also counts, you know, like I think pretty much all.

All right. So, I'm also a philosophy student uh my life all my life ever since college. Aristotle once said at at their core all problems are information problems. reason from what is true, not from what is told. And man, wouldn't you love to exist in a world like that? We We don't live in a world like that to say the least. We are triggered by our bias, our cognitive unconscious cognitive bias.

And why do we trigger like that? Well, because we've spent years and years and years surviving and maybe thriving with that model. We just trigger on an instinct. We trigger on an intuition and we say that that's got to be the right way to go. And why wouldn't you? I mean, look, my intuition literally saved us. I I know that in my DNA on that flight.

So why wouldn't you just say, "Well, I I need to just trust my instinct all the time now. You can't. It's the It's the deadly mistake." So this is what we call first principles. Has anybody heard of that concept? When you take a problem and you break it down to the first principles, you're looking for the root root root root root root root root cause.

And it's only there based on truth and information that you build and experience. can you start to build upon it? And that's the concept today where I'm going. So I've spent my whole career fighting cyber vulnerabilities. Now I'm taking that to the human vulnerability from cyber to human. So the state of AI today is you many of you I'm sure use AI all the time.

How many use it like every day? Okay. How many use it like once a week? Okay. All right. How many don't use it? Actually, that would be a good one. Okay. Good. Okay. Nothing wrong with that. I think absolutely nothing wrong with that. Like fantastic. Like I want to know you. Okay. Um, but today's AI is nothing. No, let me back up. The AI we built at Cylance is very different than the AI that's being built today or that's being infused with trillions of dollars.

Okay. The AI back then was very simple. It was a um a precursor to today's LLMs, but it was specialized. So, we built a deep learning model on malicious and a deep learning model on safe. That was it. We had two models. Then when something came into a computer, we said, well, does it look malicious or does it look safe? If it looks malicious, we block it.

If it looks safe, we allow it. That's it. That was it. Simple, very complex, but very simple in the end. Um, today what they use the same deep learning models really a little extended, but the same concept, but on language. That's it. Words, right? That's it. So, it can predict words incredibly well. And what I call it, it's not really artificial intelligence because of that.

It's artificial knowledge. And and once once I got that sort of framed in my brain, I I realized this is a big problem. If we are leveraging artificial knowledge and expecting it to be intelligence, we are perpetuating the same bias and cognition that has failed us in the past. So what's what we're seeing in the space, you can see it here in the graph on the left.

All LLMs are general knowledge. They're not wisdom. They're not experience. They're not even intelligence to be honest with you. And because of that, and they've all trained from the same data. So ultimately, they all give the same answer. So what happens? You regress to the mean. So all of our answers, all of our actions, all of our suggestions, ways to fix problems, problem solve, to troubleshoot, they're all going to go to the mean, which is fine.

There's value in the mean, but it's the unique alpha. The left and the right of the bell curve is where you get your true creation, your true uniqueness, your true originality, and the true newness of life. And that comes only today from the human brain. So, you've probably you probably recognize something like this, right? A Zoom meeting.

And if you just take a moment and look at each pain, it's probably pretty real. There are people that are off video that you don't even know if they're listening. Who knows what they're doing. There are people that are in different time zones like it's midnight. Poor my poor India team, you know, like I don't know how they do it. They're amazing.

But they've got to be a little tired, right? And we want to work around that. people are distracted, rightly so. Competing priorities or maybe you just defer to the highest paid officer that's on that's that's on the call. Hey, Bill's on the call. Whatever Bill says, that's what we do. Like that's the best idea. Bill's got the best ideas.

I think that's true actually, though. Um or the loudest voice. Somebody's on there and just like, "Ah, we need to do this. We need to do that. We need to do this. We need to do that." Oh, yeah. Yeah. Yeah. what Bill said that or Jimmy or Janie that that sounds like a great idea. You're not getting a full engagement of true human intelligence.

All right. Does anybody disagree? And this for me as the CEO, this is what I see. I see a a price tag. I'm like, wait, how much did I just pay? How much did I spend in that hour? Did I get did I get my money money's worth on that system? No. No. No. So this is the problem that we want to solve for. Now the how how do we solve for this? Well, as I've mentioned before, I don't believe artificial intelligence, which is really artificial knowledge, is is absolutely not cognition and it's not human cogn.

It's not thinking. That's why, by the way, uh do you know um well, you know Yann LeCun? Does anybody know Yann LeCun? He's one of the more famous anti- AI AI guys. Okay. So he started a new company and there's others like thinking machines and stuff like that because it doesn't think. AI doesn't think. All these new big companies that have raised billions of dollars are trying to solve for the fact that AI today does not think.

It just regurgitates. What if though you could actually replicate how you think, how each of you thinks individually and that the way that you think is truly unique. Not just from the unique experiences that everyone's had in their lives, but neurologically. You have a different pathway. You've built different neurological sequences to solve for problems, to answer questions, to like coffee or tea or dogs or cats or the color blue.

So, what if you could do that? Well, we've done this with enough information. We can actually map to the quadrillion space of your mind. and in such a way that we can actually replicate how you think. So in a meeting you can be 100% present. your interests, your concerns, your objections, all of it can be accounted for inside of that discussion.

So that when you do meet in the human form, the vast variety of optionality for any sort of problem can already be researched, can already be u vetted and then ultimately a consensus and a collaborative decision can be made. Now all you need is the humans to agree all of us to be now held accountable for that decision. And that's the big part of what we want to solve for.

So when you you think to yourself, well why can't you know just AI do that? Like well don't doesn't it have memories today? You know like it remembers what I asked it yesterday remember what I asked the day before and can't it pull like preferences and styles and things like this? It can do some basic work maybe 25 to 35% accuracy deriving inferring how you think but it's not enough for cognition and our models now which we build from scratch are above 90%.

In three ways so the first is known recall. So if if you tell it like, "Well, I like dogs more than cats." Then it it'll recall it. Okay. Like, "Yeah, Stuart, do you like dogs or cats?" I like dogs. I mean, I like cats, too, by the way. Not a cat. Um, so that's known recall. Second is likeness. So, how I present myself, how I communicate, how I where, you know, my mission comes through and engagements with people, things of this nature, that comes through.

So, likeness 10 out of 10. The last is unknown recall. So any system like this needs to be able to think like you. So if I've never told it that I've never said the word dog to it or cat to it and I ask it, are you more of a dog or cat guy? It'll answer it. It'll answer it. And that's in the 90th percentile. So it's that unknown recall which is today no one has done.

No one. So, we're super proud about this and we'll be launching this in the next probably three to six weeks for an internal MVP if anybody's open and willing. Um, love to have you on board. But we have proven it time and time again that we can be truly truly unique as individuals and we can do that. We can measure that and prove that in every way, especially vis-à-vis the real human and vis-à-vis the common LLMs that are out there today.

Because where I think all of this is going to go if we can do this truly is we will move away from this hierarchical decision-making system that we have. Right? How does decision how do decisions get made in a company typically? Right? You say, "I've got a sales problem. I'm going to ask my chief revenue officer. I've got a technical problem.

I'm going to talk to my head of engineering. If if I've got a customer problem, I'm going to talk to my head of customer success." And it's going to go down that trail. But maybe the talent that's in that group can't solve the problem definitively. And you need an open mind to take in anybody's ideas. And that's where I believe this hive mind world will start to form.

where you get decisions at compute speed based on the institutional knowledge of your company or team that then can allow you to accelerate, increase productivity and ultimately increase your success rate. So all of that comes together. But many of you, and by the way, how many of you have friends, I won't ask you about yourself, that are scared about your jobs or friends that are scared about your jobs?

Okay, this is fantastic. So, this slide is for you. So, you if you if you know my books, you know I try to debunk myths as as much as humanly possible. And I I actually have a a quarterly webinar that I do podcast called AI Exposed and I expose the limitations of AI at every quarter. And let me share with you four universal myths of AI that are now starting to show up as real.

The first and probably the biggest reason why humans will always need to be in the mix. How can you possibly hold AI accountable for a bad decision? Right? So, Amazon, AWS, did you hear about this one? Right? They laid off all of these developers, thousands of them, allowed AI to to do pulls and commits on their code. For the first time in the history of AWS, they were down for like full six hours, a full that that doesn't happen.

They're like 99.99999% uptime. Okay, they had to hire a whole bunch of humans back. That that that doesn't get the headlines, by the way, but the letting of them go does. Second, you need that experience. This is just knowledge. It's just artificial knowledge. It is not wisdom or experience. It's not all of the scars and the lessons that we've learned from our past.

And that's be and that what happens is that's why you get hallucinations. That's why you get errors and hallucinations are very real. And unless you're checking it, you'll never know. You just think it's the smartest thing ever and you just trust it. Third is the context. And this is anyone know this one? Yeah. Marie Antoinette, right? So talk about being tone-deaf, right?

Like well just let him eat cake. Like stop him from storming the castle. It cannot read the room. it it has no real memory although some of that is getting solved for to a degree and I think it will be solved for in the world of LLMs and then lastly there is no mission AI just takes orders that's all it does it has no spontaneous agency whatsoever so I'll leave you with this thought Desmond Tutu: there comes a point where we need to stop just pulling people out of the river and we need to start going upstream and finding out why they're falling in.

I love our detect and respond muscles. They are strong, but we can apply it to the prevention as well and prevent 99% of that which we detect and respond. And with that, I will hopefully inspired you to predict the failure and prevent the fall. If you want more on me, I'm at stuartmcclure.ai. Thank you very much.

Wow.

Wow. That was uh something.

So, I know you guys were really, you know, happy and excited before I got on stage. I don't know what happened. So, this this piece was I I had a couple of things to tee up for Stuart, but I have a feeling there's some questions that the audience wants to ask, too. But maybe just to get this rolling, you've you've innovated here. You've created here.

Here meaning Orange County.

Yeah.

Um what's our faulty latch, you know, in terms of the innovation community here?

Well, I think a couple things. The first is what I heard earlier on the panel which is a lot of great minds here, a lot of great ideation, innovation, entrepreneurship, but not a lot of deep pockets that are willing to risk their money in the venture space. That's why I'm heading to Seattle tonight. I was in New York last week. I was in the Bay Area the week before.

I'll be back in the Bay Area in a couple weeks. I'm going to Boston after that. Like, I love all the investors here. They're wonderful people. They simply don't have the risk appetite or the deep pockets. I mean, they're not raising three billion dollar funds. That's what these people are doing. And they're doing it every six months. They're willing to throw, you know, one, 10, 100 million at a hundred deals a year.

And there's there's no one down here that can do that. It's just a it's a volume thing and it's a risk appetite thing. That's what I've seen my whole career here. Um and I've tried every single company. I've I've done all the rounds.

And part of the reason we bumped into each other on a flight two weeks ago.

Yeah.

We were I was heading to New York. were

Yeah.

And

that's right.

Both looking for capital for you for your company and you know me for some of the companies that we support.

Yeah. I've always said if we had, you know, two or three sort of institutional tier ones come and be birthed from here, tier one investors come and birth there and we would be off and running because the talent is real and the innovation ideas are real. That's the sad state of it. Yeah.

So that's that's one of the faulty latches.

That's one of the faulty latches. The other is and again this is all um you know waves that come and go right so you remember the dot-com wave it was like you're not spending enough I remember I was pitching in the valley for Foundstone and one of the very first investor that we bumped up against said oh you're not spending enough and and we were we were trying to raise like 10 million bucks he's like oh you need to raise a lot more money than that

like you know I'm looking at pro forma you need to triple your marketing spend like why why would I do that I don't need to do that but to get money I needed to do that I needed to tell them that even though it was disingenuous and it so it's these trends that you have to follow unfortunately so the current trend today of course is AI and the second part of that trend is they tend to value uh PhDs over practitioners.

And they tend to value those that are firsttime founders. I'll put it to you that way. Not the old guys like me, unfortunately. And that's the truth. And it's okay. It's just one more challenge and hurdle that we got to jump over. But at the same time, those are real. You want you have to follow the trends, unfortunately, because investors invest on FOMO.

they invest on, you know, hey, their their buddy at such and such is getting in on this deal. Do you want to get in on this deal? And that kind of thing, and they don't want to miss a deal. So, it's it's that's the other side of it. Um, I do think though the talent the the folks that graduate from here, they they they leave quickly because they don't see the potential to, let's say, start with this company, the startup, that that can get fully funded.

and if it doesn't work out, they can walk across the street to another fully funded company. And so they have career resiliency. They don't have that here really. They have it in the Bay Area. I mean, you could literally like walk around the corner in a wei work space and be like, I just got fired. Are you looking for you looking for people?

And they'll be, oh yeah, absolutely. We're hiring right now. You know, so that it's I think it's density, too. and and you know the problem of scarcity.

Yeah. Yeah. So to that point, you're almost moving from underfunded company to underfunded company.

That's right. Exactly. Which is hard. You really have to have high resilience, high adaptability, and high risk taking and probably not a family, you know, or it makes it challenging. So, so those are probably the two I think biggest.

And I will say age is all relative. So to me, Stu, you're young.

Thank you. Thank you.

Um, what what what about before we take a question from the audience or so, what about the the aspect I think you touched on it a little bit, but there is this fear that AI is going to take jobs. Yeah. And

you know, my my sense always is that for every for every one it may eliminate, it's going to create another one someplace else or maybe even a multiplier. Uh, and I love your I love your reversion to the mean. game

and my team's getting blown out by someone else because they're shooting, you know, 15 for 15 from the foul line, you just know maybe the second half's going to be a little bit different that that that that team's going to revert to the mean. and

yeah,

vulnerabilities.

So one of the places that's getting really really hit hard right now is developers, software engineers. I don't know if you've seen that, but um which you know that doesn't really encourage people to graduate with that CS degree anymore, but I think the energies are misaligned there. I I don't believe that that is true. I I I believe that what it means is you're going to have to adapt as a software engineer.

And there was a there's a great um there's a great YouTuber guy that I have never subscribed to a single YouTube channel in the history of YouTube. and my involvement on YouTube, but I subscribed to this guy. Um, and he had a great point and he's a developer, okay? And he talks about the AI industry and he said, you know, look, every, you know, phase of life and every single sort of phase of your career, you have to attune to what is really needed.

Now, right now, AI slop is valued. Has everybody heard of that term AI slop? So, it's sort of the derogatory representation of the actual output of any AI. It's just sloppy. Super sloppy. Us as humans have to like wait, what what is it trying to say? And like be able to pull out the right words at the right places and be like, oh yeah, okay, that's what it's trying to say.

I get it. And he said, look, you have to embrace the slop. and at every job interview because there was a there was another YouTuber he was commenting on that basically lost his job and he went on all of these interviews and just keeps you know getting passed that there's no no jobs out there and he he shared a lot of the details of that process and so this other person Mo is his name was commenting on it and he said look buddy I wouldn't have

hired you either like you you've got to create a better energy you know when you're interviewing like you have to solve for problems they didn't even know they had in the interview like these are things that are going to separate you from the pack and one of the harsh realities is for a software developer is you have to embrace AI and AI coding.

So if you walk into an interview and you and his advice was real simple. is like, "Look, you're going to Google for a job interview. Um, vibe code a replacement for Google." I'm like, "Yeah, that's right." If somebody came to me and was like, "Stu, I've I've rebuilt all of it using Vibe coding." I'd say, "Sh, show me. This is fantastic. This really shows me that you have a mind, an innovative mind that can take any tool that'll help you achieve the the ultimate goal and use it.

Now, here's the other side of the coin, though. So, not just, hey, okay, built a replacement, you know, with Vibe coding, let's say, but now you have to explain the code to me. And that's not easy. So, I would say, okay, great. Now, show me the code. pull up, you know, cursor, whatever, you know, VS code you got and now walk me through the code structure, the front end, the back end, if there's any middleware, and explain the functions inside of it because if you can't explain the AI slop now, you become completely dependent on AI to fix it.

So, and we can't have that. I can't be spending, you know, tens of thousands of dollars every single week because people are leveraging AI to build code that has multiple errors that require it to go back and to fix each error, which then creates more errors. Then you have to go back and feed that back in. I mean, it is a token generation hurricane.

It is an absolute class one, you know, like tyrannical typhoon of tokens. So, you have to be smart. You have to know it. You have to know how to manage AI to help you code quickly and it can code very quickly, but it's not always the best. So, anyway, th those are the realities I think of of the world. that

right

that's Exactly. So part of what we're doing at WethosAI is we go into a company and we identify your data and workflows so that we can replicate the institutional knowledge of your departments and functions so that you can start to begin to scale and augment whatever function that might be audit tax compliance whatever consulting because I don't I don't believe no one's going to rely on AI to just make decisions blindly.

It's too error-prone. There's just no way. Maybe some of the new world models will come out to get it down to the 0.1 or 0.01 on hallucination, but until then, I mean, we are there's no way. You have to have humans in the mix to take full responsibility. And if you believe ultimately that there'll be the world, the future worlds will be AI to AI.

So, an AI will sell to another AI that'll buy from that AI and go down the mix if that's what you really believe, and many people do, then you absolutely need the unique representation of human thought inside that mix. Or you will simply have nothing but average. I mean, how does Coke compete with Pepsi if they're following the same marketing plan, genius marketing plan that came out of, you know, OpenAI or Gemini or any of these places?

They're not different. They're all the same slop. So, just put it in perspective a little bit. Weather the storm a little bit, adapt, and and be AI savvy.

Great advice. How about the audience? Anybody have something here? Go with you right here. uh

one

it's not like a question it's just a it's an observation which I have been uh work because we are working on a capability testing on a real life project for our standard uh uh for our standard um we are doing the compliant city project so we have done the capability testing on AIs what exist and it is what exists it can be helpful for us uh in our you know project and all that.

So there's lot of basics have been covered by these AIs if you look at it and then at that point I just observed that these graduates which you are getting out of these universities the experience they used to get in first few years we don't need them we already have AI is providing that you know expertise and knowledge so the biggest I feel is that we have to standard um raise the standard of our education make sure these graduates when they graduate they are equipped with the advanced level of knowledge of AI, not the basic because AI is already giving us basics.

So that's I wanted to really talk to you know because Orange County is really active in this thing. So I think we have to raise that education standard to a level. So when these graduates came out they're already at advanced level to be hired immediately because even with company like we don't need basics uh you know expertise.

That's true. Until you get the bill.

Yeah.

Yeah. Yeah, I don't we'll have to prove it out a little bit because for every one question you ask of AI, it traditionally triggers at least nine more questions. And so it's not a 1x, it's like a 9x the cost of what you think it will be and it catches you off guard. In fact, a lot of people are getting the bills and starting to slow slow down their rollouts of AI because of this.

So there is a point of diminishing returns, I think. But you're you're absolutely right. I agree with you with regard to the educational system and infusing AI, the knowledge of it, the the leveraging of it come with not just hey I know how to open a ChatGPT app but really know how to leverage it how to uh reduce the hallucinations reduce the errors and there are there are many ways to do that today it it's not easy but you can and so bringing all that to bear to has to be taught.

It has to be assisted and I think they will. I think educational systems will naturally have to step up because you know going to school a big part of it is to get a job uh over here in the middle.

Sure.

I'm so inspired by you. Um, my name is Head Day and I'm in AI marketing.

is

because I'm a a geek as well and I vibe code and I and I create systems with AI. Um, when you do know the foundations of start to finish and how to reverse engineer what you vibe code is, are you seeing that the people that apply the science could actually suffer from burnout? Because I think you can keep going and going and going and when you're deep into it innovating.

Um, what are you seeing that discussed as opposed to losing your job?

Yes. Yeah, I am. Absolutely. Because the most senior folks are the ones that are bearing the burden of understanding. The juniors, you sort of expect them not to understand. Um, but the seniors have to take responsibility for the juniors and so they tend to get burned out. I have multiple folks like that on my team. And I constantly have to tell them like leave this office.

Like this is ridiculous. Like I get there at 8, you're here. I leave at 6, you're you're here. like you need to have something else going on. But they get into these rabbit holes of of AI vibe coding etc. and it's very difficult to to break out. So I do believe there is going to be a falling off stage at some point here soon where the value of AI is not worth the risk to the individual to the team etc etc.

I don't know where that balance is going to be yet. And I don't know when, but there there will be a balancing. I I you can sort of feel it out there if you if you watch enough of the of the news and sort of the zeitgeist uh discussions, you can start to see it starting to happen now. Uh people are not not as as highly valuing like take take for example Opus 4.7 which was the latest Claude that came out.

Um and even Mythos, I don't know if you've heard of Mythos, right? all the hype around that, which is silliness, right? It's all marketing garbage. But, um, in in both of those cases, uh, it it looks like it's starting to regress in terms of, uh, the errors are starting to increase, hallucination is starting to increase, the quality is starting to decrease.

So, we've said in the industry for many years now that it will hit a ceiling and not be able to get any better. It just and you'll need to rely on these marketing angles like oh this model is so so scary that we can't release it to the public because it'll be able to hack every every you know website and company computer under the sun. Look, the current AIs can already do that.

Not that's not a reason to hold that back. There's other reasons why. I think they're starting to see that there's a limit now to what you can do with text and that's because that's what they've trained on just trillions and trillions of words. So yeah, I'm I'm I don't know. I'm I'm hopeful and we're going to make it all through this quite well actually in my opinion as humans.

But we do have to make sure that we are savvy enough to be able to say things just like that. like can you you know credibly go to somebody that thinks they know AI and say that's not how I understand it are you open to a counter opinion on that and then no no that's not you know I look at all this stuff look at all this material look at all the testing look at all this stuff so anyway that I think could be absolutely part of the education which is you know hey let's use first principles here's what's the problem like let's go to the original source of the problem and let's build back up and be able to show that in empower that critical thinking.

I mean, it's one thing. So, I was a psych degree major. I was a philosophy major and I was a comp sci and I had no idea. I mean, I thought it was schizophrenic, but I I had no idea what I was going to use all that for. Today, it looks looks like a pretty cool combo for me to be able to communicate to other humans on the challenges of technology and AI.

And so, I I've been I've been happy about that. But like, yeah. Again, critical thinking, power that muscle to me, that is the most valuable thing here, and none of us do that.

So, anyway, we we've got to move on here. We're a little bit over time right now, and uh we're going to have one one speaker come up and join me for 15 minutes. Uh I'll introduce that in a minute. Uh but I really wanted to thank Stu. I mean, you you've been an inspiration to a lot of the entrepreneurs here. It's a model to follow. It demonstrates that you could build uh a very highly successful company here in Orange County.

Uh you got to go find the money someplace else as Stu said, but but it's it's doable. It it you know it's absolutely doable and Stu's been a great friend to Octane, to the community, and I want to thank you.

Thank you. Appreciate it. I appreciate it.