Why Smarter AI Still Fails Without Human Trust

Living Your Legacy

Dr. Allen Badeau discusses the critical gap between AI accuracy and human trust, sharing how his experience with intelligence agencies led t

Key takeaways

• Dr. Badeau found that high-accuracy systems are often ignored by operators if the interface lacks transparency and trust.
• JanusAI utilizes 'biomimetic memory' to mimic human brain functions like semantic and archival storage for efficient data retrieval.

Main topics

• AI Trust and Human Interaction
• JanusAI Architecture

Notable quotes

They're trained to tell you what you want to hear.

Conclusion

The episode highlights the necessity of moving toward deterministic, h

Transcript preview

Speaker 3 (0:00) I thought that if I could build a technology that was so smart, that was so accurate, that everybody would just, you know, believe it. And that's not the case, especially in, you know, when working with the CIA and working with some of those other three-letter agencies, those folks, those operatives, they tend to not believe that kind of thing. And that's a problem. And I built this system where that interface was almost completely ignored it was just telling them the answer as opposed to letting them interface with it and so that's when we started to build janus dr Speaker 1 (0:36) alan bedo is a visionary strategic and innovative technology executive and the chief artificial intelligence officer at digitalnet.ai leveraging his expertise he helps organizations integrate cognitive systems to solve complex challenges driving a legacy of technological advancement and modernize intelligence. We Speaker 3 (0:56) have created what we call our biomimetic memory. We created it so it would mimic the human brain. And so semantic memory, your archival memory, which is really your long-term storage. So we've created our memory to behave just like that. And it does it dynamically, so you don't know where the memory is necessarily, but you know the conversation is good. Speaker 2 (1:18) Give us some background so people are like, okay. What is this guy? What makes him qualified to be building systems like this? Yeah, Speaker 3 (1:24) so I have a... It Speaker 4 (1:28) spans the globe like a super high school internet. Elvis, ready. Speaker 3 (1:34) Today, Speaker 4 (1:35) Apple is going to reinvent the phone. It's not over until I win. The Speaker 1 (1:41) Living Your Legacy podcast. For those who live to leave a legacy. Speaker 4 (1:47) That's extraordinary. Say shit. Speaker 4 (2:01) Hey, Speaker 2 (2:02) welcome back everyone to another episode of Living Your Legacy podcast. I'm your host again, Jay Slangen, and thanks for joining us again down here at the sunny Miami Beach studios in the podcast room, where we get the privilege to talk to some of the best entrepreneurs from around the country, the disruptors, and experts from all different kinds of fields, which makes my job so interesting. And today we've got a special guest here. today to talk about the ins and outs of AI. Yes, people, we got an AI guest here today, so let's have a little fun. Why don't you introduce yourself? What's your name? Where are you from? And what do you do? Speaker 3 (2:34) So I'm Alan Bedeau. I am from Key West, and I am the chief AI officer for digitalnet.ai and founder of Harmonic AI. Speaker 2 (2:45) All right. So I guess if LA was Silicon Beach and San Francisco is Silicon Valley. What's Key West? Speaker 3 (2:53) I guess Key West would be South Miami Beach, maybe. Silicon Speaker 2 (2:57) Keys. Silicon Keys. Yeah, sure. AI Keys, something like that. Yeah, that'll work. All right. And what brought you down to Key West? Speaker 3 (3:05) Boy, I wanted to find the farthest place south that I could go and still be in the U.S. Growing up in upstate New York, I hated the cold. I hated the snow. Same thing in the Pittsburgh, D.C. area. Hated the cold. Hated the snow. I don't have to worry about snow in Key West. Speaker 2 (3:22) Yeah, yeah. Perfect, perfect. So, and with AI, you can work remotely and just set up your little studio office. Speaker 3 (3:29) That's right. Yeah, I could build a company down there. I could create all the code that I needed and I could be successful down there. Speaker 2 (3:37) Nice. And so how long have you been down there? Speaker 3 (3:38) Boy, I've been down there a little over six years now. A little over six years. So that's Speaker 2 (3:42) exactly what you did is you built your AI company down there, right? Yes. And you recently sold it. Yes. All from the comfort of the keys. All from the Speaker 3 (3:50) comfort of the keys. Yes. Yes. So Speaker 2 (3:51) it's not just Margaritaville. No, Speaker 3 (3:54) not just Duval Street, although they are some good examples of creating some personalities for our agents that are counterintuitive. So it works out really well. All Speaker 2 (4:05) right. So, yeah, tell us about the AI that you developed and recently sold. Speaker 3 (4:10) Yeah. So we started out from the premise that we wanted to democratize AI. We wanted to give our AI. AI agents' personalities so that they could learn better, they could use tools better, they could, you know, be created by any user. So they would be able to, you know, really apply it to any situation that they were trying to solve. And so when we put all of that together and used, you know, the non-chat GPT methods of, you know, solving these types of problems, then that's how we got Janus AI. Speaker 2 (4:44) Why'd you name him Janus? Janus, like Janus Joplin or? Speaker 3 (4:47) No, no, no. If you look at the Greek god of Janus, it was able to look into the past and predict the future from the past. So it was that two-faced god that could do that. And so that's how we came up with Janus AI. We used the past. to be able to better predict the future. Speaker 2 (5:05) All right. And the best predictor of future behavior is past behavior. That's exactly right. All right. Okay. That's very cool, Janice. So it's like a Greek guy. Speaker 3 (5:14) That's right. Speaker 2 (5:15) Okay. And so we'll get into the chat GPT models in a bit. But yeah, AI agents is something that's... Kind of on the verge of the kind of AI chatter right now because, you know, they just launched Grokbot, right? Right, right. And it's supposed to be like the easy AI agents, right? And before AI agents was quite complicated, right? And they're getting easier and easier to develop different teams, right? Right, right. That's kind of the general use. If people think of AI agents, you know, and they think of like, you know, someone that has... is launching a business, they'll be like, okay, what do I need? If I was going to hire people, I'd need an accounting team. I'd need probably a marketing team. I'd need my chief operating officer and just the various different roles of my company. And now they're kind of filling those roles with AI agents. Yeah. Talk to us a little bit about that. Speaker 3 (6:12) Yeah. So agents, if you think about it, they're not new because the problems have already... been, you know, we've tried to solve them in a lot of different ways with, you know, RPA, robotic process automation, you know, 10 years ago to, you know, really trying to break down AI problems and, you know, different types of problems so they could be used in different ways. But the AI agents term really started to take hold about two years ago. And it was about, you know, just trying to apply at the time. you know, these large language models to solve specific types of problems and have them pretend to be your accountant, your CFO, your CTO, whatever that was going to be. Speaker 2 (6:55) Well, in the modern world, people kind of think of Jarvis from Iron Speaker 3 (6:59) Man. Speaker 2 (6:59) That's exactly right, yeah. Right? Jarvis, do this, do that, do this, right? It's like when he's working on his cars in his studio, or his garage, I mean, right? Jarvis, he knows about that, right? Jarvis, he knows all about the systems of his robots. So that's kind of like where we get our mindset of AI. And actually, I saw something really cool. Some guy was developing, he had all his different agents, right? And he had this kind of like 8-bit beach. Yeah, yeah. That's Speaker 3 (7:35) tough. Speaker 2 (7:36) That's tough. Okay. So 8-bit beach. And, you know, some of his AI agents were out surfing, right? Some were kind of having a picnic. Some were playing beach volleyball, right? And then he called for a meeting. And then, you know, the people surfing, they came in. And everyone, the volleyball people, they stopped playing. And they all... gather around this picnic table, right? And then each of them gave him a rundown of the day. Right. You know, the marketing, the CEO and all that, you know. And it's kind of cool to see them, you know, during their work period, but they're kind of just hanging out. Yeah. But they're getting creative with AI agents now. They are. Yeah. Have you seen anything cool in the agent world that's kind of interesting to you? Speaker 3 (8:20) Oh, we see them in a lot of different government. programs that we've we've started to support we see them in some financial industries where you would traditionally think that humans uh were the only ones to be able to do these sort of things but we also see them in some creative type roles where you're asking it to do uh you know those same sort of things marketing creativity type type activities and you know some more successful than others. But one thing I think what's fun using your analogy with Jarvis is people also forget the other robot that was in there that would extinguish the fire after the fire had already, you know, happened with Tony Stark and, and that whole thing. So you've got your really smart agents like Jarvis. And then every once in a while you get the B team. And they're not performing the way that you want them to perform. And Speaker 2 (9:18) that goes into Janus and what you created with Janus, right? That's right. And it's really, really interesting. And it makes me think about AI a lot of how general people use them and how I use them as well, right?