EP11: Agentic AI Unleashed – Nvidia’s Nemo, Google’s Gemini, and AI’s Role in Hollywood

Join hosts Marlon Avery, Sekou Doumbouya, and Adrian Green in "AI with Friends," a dynamic podcast where tech meets everyday life. This episode kicks off with a lively discussion on the impact of AI on the job market, drawing insights from Sekou's experience at Afrotech. The hosts explore the evolving role of AI agents in enterprises, debating their potential to revolutionize workflows. They also tackle Ben Affleck's controversial views on AI's influence in Hollywood, juxtaposed with the Beatles' AI-assisted Grammy nomination. The conversation shifts to Google's Gemini model, questioning the relevance of AI benchmarks. Finally, they delve into AI's role in e-commerce, highlighting innovations from Perplexity and Google Lens. With humor and expertise, the hosts navigate the complexities of AI, offering listeners a blend of tech insights and cultural commentary.
----------------------------------------------------
Welcome to AI With Friends, your weekly launchpad into the world of Artificial Intelligence. Hosted by Marlon Avery, a pioneer in GenAI innovation, alongside Adrian Green, VP of Engineering at LiveNation, and Sekou Doumbouya, Senior Staff Cloud Systems Engineer at Pinterest, this show is your go-to source for all things AI.

Our hosts bring diverse expertise—from AI strategy and tech innovation to industry leadership. Every week, they break down the latest AI trends, interview top experts, and simplify complex concepts for AI enthusiasts, entrepreneurs, and tech professionals alike.

Marlon, Adrian, and Sekou combine their unique perspectives, whether it’s Marlon’s collaborations with tech giants, Adrian’s leadership in global entertainment engineering, or Sekou’s cloud systems expertise. Together, they make AI insights accessible, actionable, and exciting.

Tune in live on Twitch & YouTube every Wednesday at 9:00 PM ET, or catch us on all major podcast platforms.

Here’s what you’ll get:

Cutting-edge insights from AI leaders
Real-world applications of AI technology
A vibrant community of forward-thinkers
If you're ready to stay ahead of AI trends or spark your next big idea, join us each week for an hour of engaging, thought-provoking content.

Subscribe now and become part of the future of AI with AI With Friends!

Transcript

Marlon Avery: What's going on, fellas? 

Sekou Doumbouya: What's going on? 

Adrian Green: What's up, man? 

Marlon Avery: How's, how's everybody week going and stuff there? 

Sekou Doumbouya: Oh yeah, bad over here. 

Adrian Green: Not too bad, not too bad. You know, the weather's been good. It just rained a little bit yesterday here, but you know, it's holding out. It's been pretty warm in the 757. 

Sekou Doumbouya: Oh, man, you guys must be. I don't know how this happened, but we have the worst weather ever right now in the Bay Area. Yeah, they got a big, what they call the, it's a bomb cyclone or something like that. Oh yeah, Pineapple Express. All these cool names for, for water in the world. Yeah, you hear my Philly coming out my water. 

Marlon Avery: And I know some friends over there in Seattle, they, they lost power for a while and stuff over there and so, yeah, hopefully, hopefully, you know, everything's back in moving and stuff on their end for sure. 

Sekou Doumbouya: Yeah, definitely. 

Marlon Avery: Yeah. No, I mean, South Florida. South Florida, you know, you haven't got any random, you know, type of. The most you're gonna get is a breeze off the water, you know, especially gonna get over stuff over here. Snow or anything. So. But yeah, yes, welcome, man, welcome. This podcast right here, man's AI with friends. This was just kind of started by us, you know, essentially we, we've been having these conversations probably about a year or plus now even before we decide to start the podcast, but we've kind of been having these conversations in private. We kind of discussions, just discussing some of the things that we're seeing here in the this AI centric universe and stuff that we're building, you know, and so we had like week calls. We just kind of talk about and sharing resources and knowledge and things like that. And then one day somebody said this, probably useful, you know, useful for the rest of the world and everything. And so, so we started kind of created, created a podcast everything and share some of these knowledge and resources with you guys. I mean, we are a group of builders, engineers, practitioners, myself and you know, through us all, we've been building some of these solutions for quite some time, you know, underneath this AI, underneath this AI umbrella, except for anything. And so, man, so they got. Have you guys heard anything already started? 

Sekou Doumbouya: Yeah, let's go for it. Let's go for it. 

Marlon Avery: Let's kick it off, man. You, you had a pretty adventurous last week, you know, or so at a very large and well known conference and stuff there, man. Definitely love to hear your experience and stuff and what you kind of got from there. 

Sekou Doumbouya: Oh yeah, yeah, definitely, definitely. So yeah, last week I was at. I attended Afrotech. I was very fortunate to get down to Houston actually this time, this time around. So I went to the last Afrotech in Austin by myself. My wife came down with me. So it was like. It was, it was so much better. It's so much better in that. That capacity. Yeah. Met a lot of engineers out there, engineers with like just specialties all over the place. The one, of course I was. You know, since I'm a cloud engineer, I'm always excited when someone has a skill set that's aligned to mine. And it's still. I don't know if it's kind of niche still. Right. At least someone have like deep experience in it. So, yeah, I met a good amount of folks out there that had like skill sets around that area and that was super exciting. 

Adrian Green: Were they like working for major companies? 

Sekou Doumbouya: Yeah, yeah, they work for pretty big companies some place. Some. Was it New York Times? And there are a couple of different. A couple of different organizations that folks were doing this work for. But the challenge is with having deaf in these particular skill sets. There's a lot of things that you can go to maybe a boot camp to know how to use some of the technologies, but to actually be seasoned in there. The cloud world really started in 2008. So the years of depth that you can get within that field, there's not a lot of folks can. Can say that they've been there for that whole duration. But I definitely found some folks. Yeah. 

Marlon Avery: Yeah. 

Adrian Green: So that reminds me of a story about Prince. Prince couldn't like talk to other people for. Because of this, like when it came to music. So you had to be a real music head, you had to be a real gear head in order to really get him amped about talking about music. You know, he was such a technical guy that, you know, kind of, I guess it's. It's lonely at the top, you know. 

Sekou Doumbouya: Yeah. 

Adrian Green: A specialization, you know. 

Sekou Doumbouya: Yeah, that's so that's a good, That's a good analogy because that's exactly how I felt like, oh, look, I can say all my words now. 

Marlon Avery: Also, Prince got to the point where he worked, he wouldn't collab with artists if they didn't own their. If they didn't own their. The. What's it called? 

Adrian Green: They're publishing their rights. 

Marlon Avery: Their masters. Their Masters, Yeah. They own their masters. Yeah. 

Sekou Doumbouya: Oh, wow, wow, wow. Well, that's the next level. Like you have to be a certain pedigree to have a conversation with me. I'm not, I'm not there yet. Not there yet. 

Adrian Green: Sure, give me some time. 

Sekou Doumbouya: But as far as the, the rest of the conference, it was pretty, pretty amazing to see so many like so many, so many folks that look like us in one place. You go to your offices and place. You go to your, you know, your jobs and you go to like these other tech events like cube condos actually happening at the same time and you'll see maybe like one or two of us within there, but to see just a sea of folks like us, that's like, that's the, that's the amazing part of Afrotech. I think. Of course the other reason it was. I'm not going to go too long on this but the other reason why it was exciting is, you know, when you go to an event like that there's like two. There's two different agendas that are happening. There's the agenda, folks are trying to get a chat, maybe three, there's folks trying to get a job. Right. There's definitely like that is in the air during the daytime and then there are, you know, learnings that are happening which there were some pretty good ones. There was a, there was a talk that someone did on Gemini and how you can like integrate it into your life and they ended up showing me some things that I didn't expect to see at a conference like that. Usually they're super, they're super like high level. But this one actually went down into the weeds for a place where I could appreciate which I thought was pretty dope. But then about that. 

Adrian Green: Yeah, that was cool. 

Sekou Doumbouya: Yeah. Yeah. I learned to learn about Gemini Gems which is like equivalent to chat GPTs. GPTs, which is awesome. And then the last one, the last reason people go of course is the. Is to find their match. Define love and Afro Tech and oh man, it's, it's. It's better to bring your spouse, I tell you that. Better to bring it. 

Adrian Green: Oh yeah, yeah, sure. 

Sekou Doumbouya: Because yeah, like the, your ability to like have normal conversations with folks just increases tremendously when you're with someone. 

Adrian Green: So for sure. Now your wife is in. Is she in tech too? 

Sekou Doumbouya: No, no, she's in a non profit space but she's you know, just interested in what's going on and so she. 

Adrian Green: Can roll with the conversations and stuff. Like, you know, she's gang. 

Sekou Doumbouya: Yeah. Yeah, it's so funny. We did get to, we did go to one talk. She's like, I don't know if we need to sit to this. You can just Tell me all this at home. 

Adrian Green: Like it sounds better when you tell me. 

Sekou Doumbouya: Yeah, but yeah, that's it. 

Adrian Green: That's cool. 

Marlon Avery: Yes, I, I do want to throw in there a topic and stuff, know that, that relates to Afrotech. So one topic that we, I feel like we've been kind of discussing privately. I don't think we've actually talked about it publicly. And that topic is the current job market. 

Sekou Doumbouya: Yeah. 

Marlon Avery: You know, and stuff. And so, you know, we, we've all been, you know, doing our research. We've all, you know, at some degree might have been impacted and stuff, you know, from it and everything coming from Afrotech, you know, what's some of the signals that you see that where the current job market is and then potentially, you know, hopefully, you know, you move forward. 

Sekou Doumbouya: Yeah, job wise, it was very different. So I've been to, went to 3 Aflatex at this point. I went to the one in Oakland, which I think was maybe the first or second. That one in Austin, and then this one. This is, this is a really different situation. You know, usually you have, you know, you have a lot of students that are trying to get their first tech job and that's always a little bit of a barrier when compared to other parts of, you know, you know, your job hunting. Right. This time there was just. There are so many people impacted. And usually, you know, a lot of the companies that I've like walked over and kind of talked to it shared like a very similar sentiment that like, you know, there's just not a lot of jobs that are being offered right now. And you're at a place that's almost like a career fair with like a very low inventory situation. So folks are trying to make their best impressions, hoping that there's some edge that they can get from being in the conference and things like that. And, you know, it's just a little, it's a little challenging because, you know, there's only, you know, it's not like before. In the before days, you know, companies would hire just to just get people just to be competitive, just to say that they're just stocking up on engineers so that they can have a competitive edge over their, their rivals or as projects. They know that they're going to have more projects in the future so they can just have backlog engineers on staff. And it feels like that world is kind of shifted away. Right. So now we're at a place where unless there's like, you know, it's back to like, you know, maybe classic, you know, business practices here where if there's not budget for it that's already assigned, like, like positions are not being like available for folks and things like that. And that's a, that's a, that's a big shift. So you have, you've. We've definitely changed the equation where it was many open jobs to a few amount of engineers to a smaller amount of jobs in a whole lot of engineers that want the positions. So it was, it was packed that like, it was definitely bigger than the Austin one on definitely bigger than the Oakland one as far as like attendees that I've. That I saw out there. So. Yeah. 

Adrian Green: Interesting, interesting. I remember when I, when I graduated, I was like, I had no, I had no fear of like that being an issue. It was, I was in networking and I mean that's the type of stuff that'll really give me the chills when you're trying to get, you know, get established. So. Yeah, wishing everyone the best out there. For real. 

Sekou Doumbouya: That's. Get a little backstory about me. That's actually part of the reason I moved to the Bay Area is when I was working on the west coast where tech was more of accessory to the jobs and industries that are out there. Sorry, I was working the east coast, tech was more of accessory to the jobs out there. Like financial institutions would have a, you know, IT department or engineering department, but they're very small and those positions are rarely vacated over there. So there's, there's always like that squeeze over there. And when I moved to the west coast specifically, like I realized, oh man, there are so many more jobs out there. There's like, I really had a. The ability to like choose the different companies I would go for based off of like ideals that I did not have access to on the east coast there. And that was a really huge feature for me from being out here was to be able to like to do that now, you know, you know, as a, as a senior, you know, senior, senior staff engineer, like, I don't have the, it's not the same situation compared to the. Where other folks are right now and other places. But still, you know, it's, it's still a different, different world. So I definitely. Folks out there still out there trying, trying to pushing through, like just keep going. 

Marlon Avery: Let me ask you this. In your opinion, you know, with the current job market and stuff stands, I mean, you definitely changed. Like I gave one angle what. How much of. How much do you think or believe that real automation and AI has really impacted jobs, if any? 

Sekou Doumbouya: So that is actually, I Feel like connected to. It's actually. I feel like there's an article on this, but I think that is actually connected to part of this challenge that we have right now, which is companies are, you know, there's definitely layoffs that you're seeing that are happening. A lot of the layoffs are. It's almost like how the market works, where the market is like making moves that represent where they want to be. What do they think things are going to be in the next. Where they actually are, not where they actually are because the actual AI products are actually not there. They're not actually bringing in revenue with the AI products yet, but they're seeing like folks are seeing at least enough evidence that within a certain amount of time that, you know, the. Some of these. They would need less jobs for them. So to make this. This is a little bit of a gamble though, for a lot of. I feel like a lot of companies are taking and through that gambling. That is where this job shedding is happening. But I think when we talk about some of the other like advances that are happening in the AI field, ideally AI is going to augment our world, Right. And as the scales go down, like actually we just saw that was it. Salesforce is like going on a hiring spree right now, like, which is crazy because they just went through this shedding of jobs. Now they're going on a hiring spree specifically to tool up for the AI work that's going to need to happen, which is, you know, this. This is. This is the expected curve, you know. 

Marlon Avery: Yeah. 

Sekou Doumbouya: So the folks looking into like, you know, when are things to be open, like, I feel like next year is going to be. This is really the make or break because everyone has shed a job specifically to make sure 2025, it feels like we'll be able to be a place where they can really push and they need head headcount to do that. So. Yeah. 

Marlon Avery: Yeah, I think it. I think exactly what you just said, it's like it's a. It's a gamble and it's a prediction of like what I think things and stuff can be with the, with the. I would. I would say this very loosely or. Well, I will say with the ego of what they've. What. Of how they think they can control the outcome as well. Like a year plus. So for. So an example of that is what we saw from the airlines companies during the top of the pandemic where the airlines companies did these waves of layoffs even in the midst of, you know, getting government funding. You Know something to help them, you know, assist in that process. And they was doing these ways of layoffs and everything. And then when things shifted back into play, they were canceling flights, you know, if it happened. And then they had the nerve to say that, oh, people don't want to work. You know, people, you know, they are employees are holding us back and everything. And I unfortunately just think, you know, across the board you just have egotistical leadership and stuff in these areas with individuals who don't understand the build of the knowledge or anything and the input of a tool like artificial intelligence. So you're simply making guesses and gambles while you're impacting, you know, people's lives or anything. And then you're just kind of like moving on, you know, writing your notes and writing your scripts and making your tweets and stuff like that, as if you're not accountable and stuff into these areas. And I think this has happened across the board in so many factors and everything. I particularly know one company, one large company, who I had the privilege of talking to some of their CTOs and leadership and had a great understanding that, oh, actually they don't really understand how this thing is moving, but yet they did lay off because they said their staff and their employees, they're not grasping it enough or anything and you don't even fully understand it, you know, and stuff. And so with that, I think his opportunities like this in this podcast and everything is why we here and everything, to kind of help bridge the gap, you know, everything not only for the builders and stuff in the world, but also too for you, quote, unquote, leaders moving forward. 

Adrian Green: And, you know, one thing I want to add too, is that there's always been this disconnect between the upper management and just technology, as far as I know. So the upper management, I'm going to say, I know that your early bosses probably couldn't even type, you know what I mean? Like, as far as at the top of the. Whoever's running the company was like, look, I'm paper and pad dog. Like, I'm not even gonna use it. 

Marlon Avery: Seiko is not that old, bro. Later. 

Adrian Green: I'm that old. I'm that old. You know what I mean? I'm that old. My, like my boss, if you look at like the person who was running the companies that I worked for, because I never worked for tech companies, I worked for, you know, I work tech at companies, which is a real point I want to get to later in our conversations. But the person at the top was so Disconnected from technology. You know, their focus was on really everything else. Maybe that. But. But that part. And they're. What they wanted was for it to just work and them to not have to, like, worry about it. But when it comes to, you know, how these same people are making the decisions about hiring and how AI is going to enable them to maybe hire less people and things like that, it really is, like. It's a lot of hope and fake it till you make it that real. That. Fake it till you make it. But, like, it really is evidence of that. This, like, kind of historical disconnect that I see between them not knowing how, like, not precise, that these. This, you know, dealing with, you know, language models can be. How it can, you know, hallucinate and not really give you what you need every time. The lack of consistency, that, like, it. You know, it's like a story as old as time to me, you know, because they're not using it, they're not seeing it, so they're like, oh, if it works, let's do it. 

Marlon Avery: You know, Seiko, I'm not sure if I've ever told you this directly. I know for sure I didn't tell you this publicly. There was a statement that you made me a while ago, which really made me lean all the way into entrepreneurship. You said this had to probably be like a year and a half ago or so. You said, marlon, you're so early into this thing, and you're. And you're doing these things, you're building these things, but you're also applying for jobs where. Majority of the world is still trying to figure this out. Also, the majority of the people who are still trying to figure this out are also doing the interviewing. What makes you think they're gonna know what to look for, you know, and understand, you know, like, how to, like, bring this thing together. And that connected so well in my mind. I was like, yeah, it's time to ship. I can't wait. I can't. I can't wait till they figure it out. Or today, you know, kind of, you know, put their, you know, their. Their vision board together and, okay, we need this type of, you know, all the different things like that, you know, and when you said that. When you said that to me, I was like, oh, okay, yeah, it's time to shift. 

Sekou Doumbouya: In the world where no one is expert, you are the expert. So, yeah. 

Adrian Green: The level playing field. Yeah. 

Marlon Avery: So here. So check this out. A lot of people has called 2025. You know, you probably could say 1224, you know, as well as the year of the agents. And so as of lately it seems like a lot of enterprises, big businesses, big tech companies are pushing more towards just, you know, the agent framework and so says AI agents are transferring businesses by turning data into actual items. It says. At Microsoft Ignite, the tech giant unveiled a no code local capabilities in Microsoft 360, 365 copilot allowing users to build and deploy custom agents also too with that, Nvidia just announced their Nvidia AI platform called Nvidia Nemo Microservices, providing the ability to manage and access data efficiently with building agency AI applications. On top of it, these agents now have the ability along with Microsoft, they just now have the ability to tap into things like Azure's AI catalog 1800 plus models. Nvidia also highlights the potential of AI agents through tools like Rag and Nemo, emphasizing how AI learns improves via with data on the fly. So the goal here we're trying to figure this out is the goal to make AI tools efficiently and make it easy to deploy for large enterprises, for individuals. Adrian man, we'll start with you. What's your thoughts? 

Adrian Green: Well, first I'm going to start with a question and this is for the audience. My question is what's the difference? Because you've heard this agent conversation, but we've also heard the assistant, the assistant conversation. What is the difference between an AI agent and an AI assistant? And when I look up the definition here, it says AI assistants are reactive, performing tasks at your requests. AI agents are proactive, working autonomously to achieve a specific goal by any means at their disposal. So I think that the short of it just for the audience is that AI agents are going to be working just more independently and you know, rather than like, you know, message response, it's like message, we'll do a bunch of work and then get back to, then return the results back to you, I think if that makes sense to everyone. So I think that the AI agent discussion is interesting because Marlon, I was talking to you years ago when we first started our, you know, working with Lang Chain and I think that both of us knew right away that this is going to be the way to go when it comes to dealing with AI because the regular chat GPT model or just chat bot way that, you know, really, everyone was really using it where, you know, everyone had a chatbot four years ago and you know, it's going to be this big, huge deal. What we saw is the potential in really tasking specific, you know, AI agents with, with like, and giving them a specialization and having them work together to return, to return your results properly. And I was like, I remember just my mind being blown like, oh my God, this changes everything. Yes, this completely changes everything. Because now LangChain was like rough for your boy because your boy did not, was not a python head like he is now. So, you know, JavaScript head. So I now got to learn Python in order to deal with this. So it was rough getting started, but at the end of it I had a slack bot, you know, it could download reports for me and give me little totals and, and everything like that. And it was, it was, it was cool. So what I want to say is how this, what, what I find interesting is to the new people coming into the tech industry now, I believe that because of this, because of the really, how really sticky the agents seem to be, is that we could be on the cusp of a, you know, a kind of shift in how tech is done within organizations. And you know, for lack of better terms, and what I mean is that if you come in with the skills of AI orchestration, you may be able to better join the workflow of a company, you know, but you need to have those hard, you need to have those hard skills, you need to have your basic programming fundamentals down. That's a given. But the person that's doing the legwork in a lot of these companies may be AI agents. So it may be up to you and to appropriately manage that in order to manage whatever KPIs and benchmarks and whatever product that you're trying to build. So I, and that makes it a little bit different, you know, that makes things a little bit different because as these frameworks are built, what I believe is going to be needed if we, if we stay on the agent conversation is the proper management and tracking of the chain, like you can, you may need to be a micromanager of the agents, you know what I mean? So, yeah, proper logging, proper output, you know, the output that this agent gave, this agent needs to be able to be audited so that we cannot make the same mistake again, you know, and in doing that, you know, with retraining, you know, you can start to have a super duper improved system internally that, you know, for the company or a system that you, you know, are building. So I find it interesting. But I, that's my takeaway. That's, that's my takeaway about the agent conversation. 

Sekou Doumbouya: Yeah, yeah, you know what? So, so interesting. Remember we talked about this. I feel like it was like last Year because all this stuff is like kind of building up and we're like, hey, is Lang Chang the thing. Should we be investing our time making sure that we're deep into Lang Chain? 

Marlon Avery: I'm like, I think the thing that you said, you said we, we need a, we need a framework. A react type framework for gen AI. 

Sekou Doumbouya: Yeah. Which meant anything you're doing as far as frameworks, it's just hobbyist work. Just putting that category of just understanding some of the common components that may appear in other frameworks or things that come. I think they. It was lane chain. Lane chain. Then there is auto GPT. Right. And then I think there was maybe something else that was in between there that was kind of working towards some of these things. But everything is very bespoke, I think. Now I'm not saying that we're at like our react moment with frameworks, but this is really well integrated. Like I was going through it last night and I was like, wow, okay, they got training inside of here. Looks like it's like, it's pretty easy. They already show you how to integrate it into like your. 

Adrian Green: Are we talking about Gemini? 

Sekou Doumbouya: We're talking about the. This is for the Nemo. The Nemo framework that. 

Marlon Avery: Nvidia. 

Sekou Doumbouya: Nvidia. 

Adrian Green: Okay, cool. 

Sekou Doumbouya: Yeah. I feel like this framework, it just makes a lot of sense. The fact that it has already integration and shows you exactly how to connect it to your pipelines and things like that. That is a level of completeness that you want from a framework and the documentation is pretty dang good. I'm going to take some time and try to play around with this as much as I can. I think the only thing is Nvidia, I don't know. This is my opinion. Of course Nvidia creates a lot of great tools and things like that. Rarely do they actually catch on to a point where companies are fully integrating them into things. Sometimes they give you inspiration. They give lots of good inspiration out there. I'm sure there's some. 

Adrian Green: What you could do, what you could do. 

Sekou Doumbouya: And the fact that there's an enterprise version of. Of Nemo. I was like, I don't know. I'm not sure if this is going to hit the mark. The fact that you're open coring it from the gate here. Anyone is not familiar with open core, that's when you provide a partial product and you leave the other partial component of your product as open source. But in order for it actually to be useful in any meaningful way, you have to buy the enterprise version of it. And I have to dig a little deeper. But the fact that I see. I see both of those things released at the same time and they didn't mention the difference means, like, they want you to get this enterprise thing that has some dollars attached to it. We'll see. But the framework discussion I'm super excited about. And as far as agents go, it definitely feels like what we're doing right now is we're trying to minimize the amount of hallucinations. Ideally, if the models are good enough to be able to go through all the flows in one monolithic way, that would be technically ideal and cheaper to put together. Right. But being able to handle hallucination, adding different contexts that are completely dispersed from each other within a single model and having. Getting it right 100% at a time is super challenging. So the idea is have an agent that's only trained on one corpus of information, or just look at one corpus information. Right. And have it just do the one task. Or rather than asking. Yeah. Rather than have it switch content to be like, say, a trust and safety agent or trust and safety function, like, give that to another agent to do that. But it's really just. We're really just engineering around a problem space that is currently inherited in our models. So this is a spot that you still kind of need to be careful about, because if someone actually fixes that, it makes it so you can do that in one monolithic run, then there's no. Why are agents existing? If that's the. 

Adrian Green: Yeah, that's true. That's very true. Yeah, that's very true. 

Sekou Doumbouya: So you got to be careful, this one. But. Yeah, but is it. 

Adrian Green: Will the worm eat itself, though? Like, say, like, okay, we train the models that they're smart as the agents. Can we turn those models into agents that'll be super smart at their specific, you know, thing. You know what I mean? So now you give them a car, and they're like Tony Stark with it all of a sudden, you know, because, you know, who knows? 

Sekou Doumbouya: Hey, Exactly. Well, they. We don't know. This is what this is. Everything is new. We're all. Everyone's an expert right now. Everyone is an expert when it comes to at least this part of it of. Of AI. 

Marlon Avery: So Tony Stark built us in a cave with some scraps. Speaking of scraps. No, this too. It's okay. So it's. It was super exciting to see what, you know, Nvidia announced everything. It was also super frustrating. And this is a terrible, terrible, like, drawing and everything. But I had. Literally, I was Talking to a client. And I drew out how like that world is going to work and everything with the agents with a baseline of like knowledge bases and agent stuff would be built on top of it and stuff. And then a few days later Nvidia is like, thank you, we'll take this, we'll take it from here. And so like, yeah, what, what they, what essentially kind of like what they, you know, announce everything with Nemo is the baseline is like essentially acknowledge basics of in compartments. And so in those compartments or anything you can have specialized LLMs or anything you can have fine tuning, you can have documents, you can have essentially like a rack application that can do fine tuning that can get specific guardrails, instructions and pipelines on how to perform, perform the task. And then on top of that you build voice agents, chat agents, you know, ability to email, right. You know, multi model, you know, stuff aspecting. And you do that per job category, segment of business. And then so across you have this whole infrastructure and stuff, if you will, you know, that you consistently kind of build on top of and everything. And so the more tasks and everything it gets most absolutely stuff it learns. I think you kind of put that back essentially kind of like in the foundation, you know, of the knowledge base. And so that enterprise framework, if you will, is kind of what, you know, Nvidia stuff announces up there, which is, which is, I think what you said Tuesday was just like it's, it's, it's inspiration. I mean, right now, isn't Nvidia like the most valuable company in the world now? Something like that. 

Sekou Doumbouya: Like, I'm pretty sure they are at this point. 

Marlon Avery: Yeah. So like, I think they're, they're playing with house money, so they're just trying stuff and stuff and stuff. And so, so yeah, I think, I think, I think it, I think it is inspiration. I think it, I think it's a, I think it's a step forward, you know, and stuff about like how, like how this world is going to work not only for us as employees or business owners, but also too, how we interact on a consumer level, you know, too as well. So, you know, the ability to have, you know, a customer service agent that has a log of your past 16 calls and it knows exactly kind of like your use case and what you're, you know, dealing with and understanding your profile and what you need, maybe you still keep dealing with the same issue. So then it knows to escalate it, you know, things like that. Instead of like, you know, kind of consistently going through a chat Bot and so where maybe you know, that has just like a step by step operation, so just like how to respond everything. I mean, I think, I think this gets, this gets very creative, it gets very detailed and it gets very specific, entertaining to each, every individual use case. And I think this is like, I mean, this is, I mean, for us, for guys, we've been talking about this for, I don't know, two, three years of like, this is kind of like where things are headed. And it's like, it's like now it's like, okay, here's the framework, you know, here's the playbook, if you will. Let's see if you guys can put it together. Yeah. You know, and stuff. And so I also, also believe that it's still gonna be some time. It's still gonna be some time to really kind of put this infrastructure together. Well, you know, as well, but also too, Adrian, what you said. I said this, I said just quite some time ago as well, is that I don't. We're gonna come to a world where everybody's gonna have an agent, you know, at their hip. You're gonna have, when you accept a new role, you're going to have an agent assigned to you that is going to help you in a social language with your job. And so essentially what we've seen for software engineers is that we've seen productivity skyrocket, you know, through the roof and everything, you know, to help generate, you know, code and stuff as well. And I think we're going to start to see that across multiple different other functions and categories and responsibilities and stuff there. And so I think that's going to be like the same, you know, kind of like process. It's just where, you know, agents are going to get assigned per individual, per roles, and it kind of help them do their job faster, quicker, you know, better, underneath the guidelines of set company. Yeah, you know, I think that's, I think that's where we headed. 

Adrian Green: Unless, you know, because there may be, you know, because this, this goes back to our snitching conversation. This goes back to this, to the snitching conversation, you know, so. 

Marlon Avery: Sure. 

Adrian Green: You know what I mean? You know, where's, you know, where is Ronnie at? You ask an agent, you know, the agents, you know, be nice to your agent, like take it out for a drink maybe, you know, or somehow virtually, I don't know how, how you would do that, but, you know, you know, it's also a narc feeded tokens because that's it. That's What? It'll. It'll say back to your balls, how many tokens you got? 

Marlon Avery: Yeah, yeah, I think, I think, I think you're right. I think you're right with that. I think some companies will take advantage of that, you know, as well. It's the same thing of, hey, we would love anonymous feedback to how we're doing, you know, as a company, you know, and stuff. And, you know, and it's like all of a sudden now you're some situation. Like, wait, I thought this was anonymous and everything. And so, yeah, unfortunately, I do think some companies are going to take advantage of that, you know, as well. And so. 

Adrian Green: Yeah, yeah, I hope not. But company's gonna. Company, yeah. 

Marlon Avery: So there's a large area for companies and that particularly has been on the headlines of seeing if AI is going to impact their life, impact their role. And essentially Ben Affleck doesn't think so. Ben Affleck has said. Is told the Hollywood that he's expressed concerns around AI replacement jobs. But Ben Affleck assures, he assures actors and writers that AI isn't close to replacing human creativity. Speaking to cnbc after I explain how AI can mimic but not innovate, noting that movies are likely to be one of the last domains affected, he believes AI could create more opportunities in the film industry through visual aspects. Professionals, you know, might see challenges and stuff and so say, cool, we'll start with you from this one. Yeah. Because I'm still mad here made Daredevil, but anyway. Go ahead. 

Sekou Doumbouya: Yeah, exactly. Well, I don't know. Ben Affleck's not. I don't know. He's not. He's not that person. I look for to expert. As an expert for the state of things. That's like the first thing, right? You know. Yeah, he's looking at. From the surface level, I think. I think we're definitely going to go through. We have this. I feel like there's. There's some stages to this, right, where AI itself is gonna. We talked about augmenting your software engineers, right. And for Hollywood, of course it's gonna be. They're gonna go for all the things that they know are expensive, but AI can easily do, right. So maybe there won't be creativity, right. But like, there are a lot of jobs between, you know, doing the creative work and just reading something that someone who's creative wrote down on paper and just putting on the screen there, Right. And question is, is the actor the place where creativity is or is it the person who writes the script? 

Adrian Green: Oh, yeah, I Say both. 

Sekou Doumbouya: Both, right? 

Adrian Green: Yeah. 

Sekou Doumbouya: Yeah. So if you. If you're able to. I give you an example, like, you have extras, right? For. To be an extra. Do you think that takes creativity to be an extra in the background? 

Adrian Green: There's been. There's been some great extras, but we don't know their names. 

Marlon Avery: Hey, you're following instructions, all right? Yeah. 

Sekou Doumbouya: So we're. We can just imagine. We can just imagine that, right? Just imagine. Imagine your extras. Extras are now, for. Now on. They're all AI generated, Right. That's a lot of. That's a lot of jobs. A lot of folks, they, like, get their. Their. Was it what they call this side card? They like, as long as you, like, do something, you can get your health benefits and stuff like that. Just take a job. Just don't be an extra. Just to. Just to make sure. Hey, I did. Did some work. I want to make sure I keep my health benefits good to go. Right? Like, that's. That those things are under threat. Right. I think what he's trying to say is that my job is okay. My job is cool. I don't have to worry about that. So you. 

Marlon Avery: I'm in the top 1% here. 

Adrian Green: Exactly. My helicopter just landed on the roof. I gotta go. 

Sekou Doumbouya: But, you know, but the, you know, the folks on. The folks on Tubi, you know, I'm not sure about how it's going to work out, you know, the 2B actors. So, like, if you can do 2B. These 2B segments with. With AI and get the same amount of quality or clicks out of it, then I think folks will probably end up trying to opt for that. So there's. I feel like a real threat over here. Don't let, don't, don't let. I wouldn't let. Try to calm you down, sitting up there as, like, the 1% of the actors. Right? Yeah. 

Marlon Avery: Yeah. I'm gonna say this. Yeah. I think it's an irresponsible statement. The first epic. Oh, okay. Yes, I'll say this. The first application I built, you know, integrating AI, and I was so excited by. I think I call Adrian soon as I hit compile, you know, because I was like, oh, my God, this works. And I kept. I started reading the results. I was like, wait, this works really well. And so the first application I built was an AI grant writer. You know, since then, I've been doing, you know, a lot of building and integrating. And so I've been using everything for, you know, some of my presentations, you know, program management, obviously, you know, development work. So we'll take code and we'll take grant writing. Those are two forms of creative writing. Script writing is also a form of creative writing. Numb. And it's all. I mean, this is also too. What was part of the, you know, what was part of like the. Just the striking stuff with the, with the actors up there. They didn't want that to come in. Come in and, you know, just kind of like wipe them out. Which I understand. But to state that it's not going to replace writers, I think it's a bold statement when it already has the capability to do so. Now you're just. Now, unfortunately, you're now trying to figure out the politics side of it. But the capability of. And LM that is properly developed, I would automatically say on the API side, not on. Not on the website, that is properly developed here. I think that has instructions and examples. 100% can write a script. I've been approached directly from a small movie production, you know, company, and they was asking me it is possible. And I was like, absolutely. 100 everything. Because it's just all text. You know, it's all. It's all text. You're not asking it to, you know, you know, recreate the, you know, the 3D model of Optimus Prime. You know, putting this stuff together from truck to, you know, standing form. Like, we're not asking for that. We're talking about the tech side of things. And so 100 doable, you know, if, if that. If that's alarming to say or not. Adrian. Yep. 

Adrian Green: So I gotta disagree with the both of you. Like, I think he's spot on. He's spot on, actually. So what he's saying is, is that it can't do Shakespeare. So. And what. And, and what he, what he is saying overall, in general is don't be afraid because this is not creative. What it is, is something that can con, you know, summarize and contextualize and like mimic, but it's not going to be able to generate something that's going to be kind of. Because when you think about. And what. What he's talking about, I think overall is like just something that's art if you have it. Like a movie can be art. Like you can have a script that's a piece of art. The best scripts. Oh, and movies, you know, you hear the story behind them. There's such a story experience behind every aspect. If you think of something like, like the Godfather, Al Patrino was bringing his life experience at the table, and you had this production that was going through all these different things. You think about the Shining. You think about how like, you know, basically Stanley Kubrick had to basically abuse this woman, you know what I mean, and terrorize her in order to get this performance that turned into this massive hit. You know what I mean? You think about, like, something like all. Everything. Never Ending Story, Doug. I saw a documentary about Never Ending Story, man. They were. That was what. They were wild up in there. Like, really like the fact, dude, it was the most unsafe, ridiculous, messed up situation. No one ever recovered. And I had no idea. Yeah, it like, kind of affected everyone, but, like, yeah, it. It's insane. So I think that when I think of it, I think of these as the keystone pieces of art. These are the things that you feed into AI for the. To get something that it could be as good as that if you, you know, with, you know, kind of the way that the. That the world is. I don't see. I see someone maybe guiding AI through AI Assisted. You can create something. But at. Are we at the point of press the button and you're going to give me something groundbreaking? No. And that's. That's what he's saying. 

Sekou Doumbouya: I see, I see where you're going with that. I see you're going. I. I think the. I think the challenge is like, you know, the writing a script for the Shining, right? Very complex, multiple levels of, like, emotion and things like that you have to draw upon in order to make something like that, right? 

Adrian Green: And the book first, because it was a book first and then that, you. 

Sekou Doumbouya: Know, the book first. So. But writing the script for like a Kogate commercial or like a. What you call it, it's not Coca, like a toothpaste commercial or something. Something like that, right. Different levels of effort and actually is more of like a factory type of process to generate something like that, right? Those are. That's like the levels for like, disruption. There's definitely a place where you can say, like, I'm a good sample. Like, I'm a, you know, I'm a systems engineer by trade. I have lots of experience for 22 years over things that I've seen and reactions that I'm gonna have just automatically because of my. My experience, that part is going to be very hard to recreate on the AI side, right? But the act of writing some terraform code, which is also part of my actual job, right? Like, oh, yeah, that's a great place where automation is going to sit inside there. But there are folks that they can have a job that all they do is write terraform and they're never asked about the larger questions that need to answer around how things could be. Should be constructed and what are the behaviors and things like that. Right. So I guess that's what we're saying, like, for. If you're, if you're on the mountain, you're great. This is a wonderful time for you. You can do even more. But there's definitely. There's definitely that layer underneath it where it feels like those are the people who are under, like, you know, the 50 doing the 50 that work. Like, those are the folks I worry about in the, in the conversation that doesn't. That we haven't figured out like an off ramp for them yet. Right. 

Adrian Green: Yeah. 

Sekou Doumbouya: Yeah. But I think, I think you're right. I think what he's. If he's. What he's point to is the creative piece of that. The, the places where you can you create the emotion, tension, all these different things that are unique to human experience. I feel like, yeah, you're great. You're doing wonderful there. But like, the extras, the extras in the background, like, people get checks off of that. I know, I know a lot of folks that, you know, they work in Hollywood and they, you know, they bartend and they'll get a little extra gig on the side to get there. You know, check the box on their, you know, SAG card, get the health benefit. They're good to go. 

Marlon Avery: Yeah, watch this. I strongly disagree. I strongly. Okay, watch this. I'm gonna walk with your process. Adrian, who's your favorite. Let's go. Athlete of all time? 

Adrian Green: Bo Jackson. 

Marlon Avery: Bo Jackson. Who, who do you think influenced Bo Jackson? 

Adrian Green: The farm he grew up on. I don't know. Like, I, I don't know what his, his influences were. Like, what do you mean? Like, personally? 

Sekou Doumbouya: Robinson. 

Adrian Green: Like, professionally. 

Marlon Avery: Yeah, yeah, yeah. Like. 

Adrian Green: Yeah, yeah, yeah. Like Jackie. Like people who came before him. Yeah, like Jackie Robinson stuff. Yeah. 

Marlon Avery: Okay. Right. And so we'll say Jackie Robson or like Willie Mays. 

Adrian Green: Maybe Willie Mays came before William. 

Marlon Avery: Okay. Beautiful. Everything. Jackie Robson, you know, Willie Maze and things like that. Okay, cool, cool. Anything. Now with that, those are influences. Correct? Everything. And without those influence or anything, one could argue we might not have the full product of a Bo Jackson. Correct. 

Adrian Green: Yeah. 

Marlon Avery: Okay. On it, on. On taking that example and you're giving it influences. You're giving it people like, you know, let's, let's say like the recipe of a Willie Maze. A recipe. What was the other one you said? 

Adrian Green: What did he say? Jackie Robinson. 

Sekou Doumbouya: Yeah, yeah. 

Marlon Avery: You give him the recipe you know, something. And so basically you have somebody like Bo Jackson who is analyzing that recipe and from that, you know, it's taking certain segments of that recipe and is implement inputting, you know, some of those principles into on one's own life to kind of build out the framework of what we see as Bo Jackson. Is that fair to say? Yeah, that is exact same thing as. As few shot prompting. 

Adrian Green: But the GPT doesn't have a life. There's a. That. That's the. That's the piece that's. That's like missing. Is that that whatever his influences were, were filtered through his life. The GPT does not have a life. 

Marlon Avery: Watch. Correct. We're talking about script writing. Script writing. 

Adrian Green: I know, but we're using something that's not script writing to talk about it. 

Marlon Avery: So I'm just giving an example. Give an example. We're talking about script writing. We're talking about Shakespeare, which is text which can be analyzed. We can be used an example to build something new. It's the same thing we do as human beings. 

Adrian Green: You know it's not. Dude, absolutely. It's one. It's one aspect of it. It's one aspect of it. Like as human beings we run across so many variables in our day to day lives that can give us a stroke of genius at any moment. 

Sekou Doumbouya: Oh yeah, because we. 

Adrian Green: Because it'll be something that triggers something and it'll be a past memory. And then all of a sudden your whole life is. Has like changed like that. The. 

Marlon Avery: Watch it. No, watch this. You're right. Now watch this. The output of that, of everything that you experience in your life. The output of that is simply paper and words. 

Adrian Green: Those talking about creating the paper. 

Marlon Avery: It works, right? 

Sekou Doumbouya: Right. 

Marlon Avery: The output, the end result of that is a script which is paper and words. Paper and words. Everything in a large language model standpoint can be analyzed and it can be broken down. Everything. Watch this. Without that person's, you know, life experience and what impacted them like this. Like that. You're right on that. And everything. But you can analyze everything from Shakespeare to your greatest movie and everything. And you can give it multiple examples to create something new. 

Sekou Doumbouya: Everything. 

Marlon Avery: It's the same thing we do as human beings or anything on an LLM standpoint. Everything. It's just analyzing text is analyzing text is breaking down. 

Adrian Green: No, no, we're not, we're not disagreeing. We're not disagreeing about it. Creating something new. It'll create something new. But what I'm saying is the stretch of that creativity is going to be limited by the amount of data that it has, that it has access or the data that it has access to. 

Marlon Avery: Which is also why I think Ben Affleck's comment is even more irresponsible. Because literally, if you're talking about Universal wants to build an AI script writer. They have all the data. Like, they have all the data with the proper, you know, technology to build a knowledge base. 

Adrian Green: Yeah, I, I implore them to just do it. I mean, like, because what, what, what I think is that the, you know, and maybe it's like, because I haven't seen it because I don't, I don't think that they can do it now. 

Marlon Avery: We can build over the weekend. 

Adrian Green: They can't build like, say, like if it were to create something that would have the impact like Toy Story did Toy Story something. Well, Toy Story, it was like the animation and there was like other things to it. But let me just think, just like story, story wise, that will have the impact culturally and be able to sell. And like, because what we're talking about at the end of the day, what I'm talking about is the creativity and like the really the quality of the end product is that in product going to, you know, I can see something like lower level, you know, you know, kind of creative efforts, like a commercial or something like that. Yeah, sure. We'll just throw something at, you know, it. Because what we're really worried about is getting eyes on our product and not about like the art creation part of it. All right? So what I think is that, you know, you have to have those people that have their eye on the pulse of what's going on right now in order to really be the one of the many, many X factors that separate really human potential from the potential of the language model in general. You know, it's. It's our, it's our, it's our lives. It's our. Like it's all of these other variables that, that go into it that that GPT just simply doesn't have. You know, you're watch this. 

Marlon Avery: There's. There's probably nothing that can be said that will be able to convince me that the output of a LLM can write very well at a senior level a language like C, that it can't output a movie script. 

Adrian Green: Okay, Marlon, for the last time, it's creative can output a movie script. I'm not disagreeing with you. 

Sekou Doumbouya: Yeah. 

Adrian Green: What I'm saying is that the movie script will not be Shakespeare changing world changing level. It'll be derivative of whatever is in that set. Yeah, that's, that's. That's the argument. The. It will write a great flawless screenplay. Like it will do that. But that screenplay will not be on that caliber. I mean, on, on some caliber that's gonna break. It'll be made possibly on the caliber of Shakespeare, maybe not something that's going to break through to another. Another kind of, you know, level culturally. 

Marlon Avery: Watch this. I'm. I'm gonna combine what you just said with the next segment. You ready for this? 

Adrian Green: Okay. 

Sekou Doumbouya: All right, let's do it. 

Marlon Avery: The Beatles track Now and Then has been enhanced by AI has earned now Grammy nominations for record of the Year and best rock performance. Using AI to restore a 1978 John Lemon demo, Paul McCartney bought a new life to the song. The track highlights how AI based audio tools similar to noise filtering and video calls or reshaping music production. The big question, can nostalgia and iteration help the Beatles win against, you know, modern stars like Beyonce and Billy Irish? Everything and boom. Another creative example of an output, you know, and stuff in there that can have human, like impact in this area. Now you. What's it? You still need to give it directions, guidelines, examples and all those things. But I am saying in this example right here with the bit with the Beatles now having an AI assisted Grammy nomination, everything, you know, one could argue Beatles may be on the same level as Shakespeare. You know, like when you give it these examples, guidelines, proper prompting and all these things like that, the results can mirror something like, you know, the creative writing of something like Shakespeare, Beethoven and things like that and everything. It just doesn't replace the human because there's no Shakespeare without the human. Shakespeare everything. But also too, if you hand Shakespeare and everything a in AI generated, you know, script and everything, one might not be able to tell the difference. Everything, because it's still the human being and stuff in her name. But again, with the scripting aspect and everything, if you bring actors into. To bring that script alive, everything that has been AI generated and everything, one might not still be able to tell the difference. 

Adrian Green: Okay, there's a couple points here. Number one, this is an AI assisted track. Okay. It's not an AI generated track. 

Sekou Doumbouya: Yeah. 

Adrian Green: So it's. There's a. 

Marlon Avery: There's a. 

Adrian Green: There's a human involved. 

Marlon Avery: Fair enough. All right. 

Sekou Doumbouya: Yeah, yeah. 

Adrian Green: Number two, being in the, in the music industry, the Grammy Awards, the Grammy nominations, that committee is a select group of people. So with whatever agenda that they have going on, that this is going to be doing a thing, you know what I mean? Because. Because to Me and seeing, you know, just what the music industry is doing with all the, like, label consolidation and layoffs and just like, the crazy stuff that's happened. I think that they and their investment, which we talked about a couple podcasts ago, into all of these AI Gener, AI tools and out. Out there, which is up to, like, I think six or seven different companies that they've invested in or have Universal invested into. Universal is one of them, but the. Some other labels have too, I think. 

Marlon Avery: Yeah, yeah. 

Adrian Green: Oh, yeah, yeah. Okay. So this is just, to me, orchestrated by the people at the top as a soft rollout for a lot of AI stuff that is to come in the music industry. 

Sekou Doumbouya: Yeah. You know what? I think this is interesting, though. This actually feels like it connects to our world that we've kind of talked about. Like we're, you know, AI's role. It feels like at least right now that we can actually put our finger on is that it's going to augment us. Right. Like the things that are. The things that we, you know, we either don't want to do or things that are small are in a place where AI is going to help us and make us more productive. Right. In this case, it got rid of. Did a bunch of things with changing filters. It cleaned up the record. It applied knowledge that was already found somewhere else and just put it all toward this track at the same time improving and amping up to this next level. Right. The original part of that. Of this. Of this record that was created is still, you know, still there. All the creativity is still within the original songs and things like that that were made. Right. But it's the augmenting, I think right now is a. That's the story that I feel like I'm getting out of this. And it just. I know it just solidifies more of the future that I'm. I think I'm seeing now. It's just. I'm not. 

Marlon Avery: Not. 

Adrian Green: Yeah. 

Sekou Doumbouya: I'm not fear of AI taking. Taking at least my job itself, but there was someone, okay, that had to go and do those filter edits. There's someone who had to go and be like, should I do this tweak or do this tweak at some point in time that this is now, like taking the. They're taking, you know, took their jobs. So, like, you know, I'm trying, always trying to, like, I want to, like, reconcile the two. Like, oh, man, this has made these folks productive and they've been able to, you know, break her, you know, make an achievement based off of this. But yeah, there's, there's every human aspect to, to everything. Right. You know, so I should say with, with Ben affleck, when Paul McCartney, like, yeah, they're good. The folks that are part of like the creative circle who can make things are great. The factory workers of, of these industries, those are the ones I think right now that, you know, we need to figure out like a off ramp for them or a way that they can, you know, enhance themselves too to be able to be even more productive. Right. Because that, yeah, it's a challenge. Yeah. 

Adrian Green: Because now, now that I think about it, that's probably the best people in the whole scene are the people in the factory parts of it. You know what I mean? The people that are doing like those types of tasks are like, you know, probably the salt of the earth of this. Like compared to the creative team, which I just imagine is just, you know, in a room, thrown around a tennis ball. 

Marlon Avery: Whose role is that though? Is, is, I believe that's, that's an opportunity where the government can step in. But whose role is that? Is to do that soft landing, do that retraining? 

Sekou Doumbouya: Yeah, yeah, I guess depends. Like government at government can do a lot of things esque. It can start things at scale. Right. It's not good at finishing them. Right. It can, it can get us to the moon. Right. Like you do that, the big milestone. Right. But if like companies, corporations are. Would make it long, more long lasting. So I do think in this level of a change, before too much destruction happens within the different industries that, you know, someone can at least definitely, definitely step into to like help with the retraining process. Because I think ideally the perfect solution, a perfect world would be everyone becomes a Ben Affleck now like everyone becomes that has a role that is assisted by AI and is able to flourish. And now we have more creatives right through this process. Like that would be the ideal situation, but really the reality, I feel like it's just gonna be consolidation, which is what we're gonna get to is where there'll be few people who are get or not that they are the only people who can be creative but are allowed to be creative. Right. And there's that, that part, that part right there. 

Adrian Green: Allowed to be creative. Yeah, that's where you get that like clicky inner circle kind of, you know. 

Sekou Doumbouya: Exactly. Even in tech industry and tech industries, if you're not hiring junior engineers, you're never going to have a senior engineer. Right. And right now if that work is being disrupted, then how. Who who are going to be the next level of principal engineers that are going to be in our industry. If you're not building them right now and you're destroying those jobs, it's not a good long term plan for, for these industries to do that. 

Adrian Green: Yeah. Was that mentioned, was that mentioned at, at Afrotech? As far as the junior engineer conversation. 

Sekou Doumbouya: I don't think that she came up. I don't think it came up. I probably actually, I think next time we should, I should make sure that one of us are doing a talk at Afrotech to make sure. 

Marlon Avery: Yeah, absolutely. I have preliminary conversations, but it didn't go anywhere. 

Sekou Doumbouya: Yeah, yeah, yeah, I'll try with the hour friends, see what happens. 

Adrian Green: Yeah, sure. 

Marlon Avery: Conference near you. So, yeah, I, I, so two things I think, I think the, my big, my biggest thing on that is that I don't like when people have influenced people that has a voice, makes strong statements and predictions where they actually don't know what they're talking about. Because you could literally put somebody who is, you know, look up to you, you could put them in a level of safety when they need to be in a, in a lane of offense. Like they need to be, figure out a way how to work with and you know, be able to save your job and everything. And now you're putting like, oh, I got nothing to worry about. Okay, cool, shoot. I'm chilling. You know, everything. And that's just, it's, it's, it's not okay. Because I can guarantee you if, if you want to believe this or not, anything, let Universal come write us a check. I can guarantee you we can build an AI movie writer and stuff, everything that will be an assistant to Moon writers. And so when you get that automatically you're probably going to cut half those jobs and everything and then literally just have, if you have 50 writers now, you're gonna have 25 writers to then just kind of prompt against it and just make Edison as you go along. I can guarantee you we can build that 100 everything. And so I don't like when people make those, you know, those stones, you know, those strong statements, everything. Because literally you could really be impacting somebody's life and somebody's, you know, job and stuff there when you just don't know what you're talking about, you know, stuff in that area. And so I think that was my, my biggest thing is just like, you know, there's been so many individuals who come out and just, they read some headlines, you know, they saw something, you know, on Forbes, they saw Instagram comment. 

Adrian Green: And it was like, still, like, you just don't like Ben Affleck, dude. You just, you, you still make Daredevil dude. Okay, you're still mad about Daredevil man. Like, I definitely, you know. No, but like. 

Marlon Avery: Batman coming after. Dang, what was his name right before him? Batman right before him. Christian Bale Been after coming after Christian Bale. 

Adrian Green: No. 

Sekou Doumbouya: Yeah, no, him running into them, into. 

Adrian Green: The fog dog when the Metro Superman when it was going down and he ran in there to get that kid. Ran it, dude. Christian Bale would never do that, though. 

Marlon Avery: You don't, you don't perform behind Prince. You know, you don't show up after Michael Jackson now. You don't do these things. Christian Bale are the arguably top two Batmans. Everything. Ben Affleck is down on the list somewhere below. 

Adrian Green: He's below Val Kilmer. 

Marlon Avery: I will put Lego Batman over. 

Adrian Green: Oh, get out of here. Get out of here, man. 

Sekou Doumbouya: What about. What about George. George Clooney? Batman? Come on. 

Marlon Avery: I'm going George Clooney. 

Adrian Green: That was my worst. 

Marlon Avery: You know what's funny? I think we always end up finding ourselves talking about comic books to some degree. 

Sekou Doumbouya: I know, I know. Every time. 

Adrian Green: Be an analogy or something that will bring up. Yeah, that was that one time. Wolverine. Did you know? 

Marlon Avery: Oh, man, that's hilarious. Okay, all right, cool here. Let's move forward here. So here we go. Google's. Google's model Gemini is a new experimental model. Gemini EXP 114 has outperformed OpenAI's GPT4 omni on a key AI benchmark, showcasing strength in mathematics, creative writing, and visual understanding. Despite its success, experts caution that current benchmarks may overemphasize superficial metrics, inflating performance scores. This raises broader questions about how AI capabilities should be evaluated for more meaningful progress rather than surface level optimization. Adrian, we'll start back with you. Do these benchmarks even matter in everything, even today to the consumer? Is it. Is it something that we should be paying attention to? 

Adrian Green: All right, my theory is that this is a flex among the few at the top. So right now this is basically Google's ability to be like, hey, OpenAI, we're just as good as you. You know what I mean? And we know that that doesn't. Isn't going to equate, you know, directly to dollars or something like that. But what it's going to say is that the tools that we're going to build on top of this will be just as capable or not just as capable, because I don't think it did it match 4 0. So I mean it will be just as capable as yours. This also communicates to the investors that hey, so it's like, to me it's like a little, you know, like blowing the concho at like the top of the mountain where the rest of the beat like just hey, you know, we're doing stuff. We, we're just as smart as you. We just need to, you know, funding or you know, the continued funding of you know, the people who, who like believe in us and we're gonna, we're gonna make it happen. So it's a, it's a flex among among the top people in the AI AI space, which is your open AI Microsoft perplexity anthropic kind of thing. You know, menstrual them in there too. Not menstrual mistrals. 

Marlon Avery: That's a, that's a great point because Google has a habit of doing that. Like they've consistently attack Apple on how, you know, you know, their phones are better. Yeah, you know, it's like we have these thousand features that, you know, people only use 10 of them, you know, every day, you know, and, but they have this continuously thing of just like, you know, how they're essentially so I like that analogy. It's just like it's a, it's kind of a point and something that's the other. Like we're better than you and everything and stuff in this, you know, in this area. Yeah, I think it's also too. It's one of those things where I don't think it really matters to the consumer. You know, I don't think it matters to a consumer. You know, I don't, I don't care that you're, you know, I don't care that your test. I don't care EV can charge faster than my Tesla or anything. One of the main reasons I got it, you know, because of self driving, you know, in those areas. And as of right now, there's just no better, you know, self driving, you know, vehicle right now in Tesla. And so it's just like, I don't think it matters to the consumer. I think it's definitely, you know, inflated. I think it's like overemphasized. And I think when it comes down to it, I think you come back down to like principles of just like, you know, brand, you know, customer acquisition and then just, you know, how well can you get, you know, your model to respond to, you know, the user, things like that as well. 

Sekou Doumbouya: But yeah, yeah, well, you know you have something to say, Adrian. Oh, no. Yeah. So I think. I don't know when I think about benchmarks in general and things like that, like, we're way too early to be using benchmarks to like, truly, like, measure what's the. What is. What is truly the best. And there, especially when you're at the tippy top here of the, of the models, right? Because how you're ingesting things, how you've tuned the model can make it so you can clearly like, cook the books here. Not that they're, not that they're, you know, not accusing them of doing it, but like, the fact that that idea can even exist means, like, there's not like, what's it. What was this? There's a CIS mark or. Yeah, sysmark. I think this is one of the, one of the benchmarking tools are out there. Like, we need true, like an independent testing consortiums that kind of help with some of this stuff here. Right? 

Adrian Green: You're absolutely right. 

Sekou Doumbouya: Yeah. 

Adrian Green: I never thought of that because remember, was it. 

Sekou Doumbouya: Trying to remember what was it? Oh, years and years and years ago. I think it was like 20. Had to be like 2016 or 2017. I think intel and AMD, of course, they're always like, you know, challenging each other to see who has the fastest CPU and things like that. And there was an Intel, I think it was the Intel CPU that had like, like some amazing results. And AMDs, you know, results were pretty poor on these particular benchmarks and people, oh, man, this TPU is terrible. Come to find out that intel had an optimization that specifically boosted this particular benchmark built into the actual chipset there. So if you like a little, like. 

Adrian Green: A little nitrous can dog from Fast and Furious? 

Sekou Doumbouya: Pretty much, yeah. A little nitrous can in there. And next, you know, it's like a big old giant scandal between the two. I would like to see better testing methods and actually maybe like an independent body that we can look at for some of these, or at least like a company that's just focused on doing ethical testing and making sure that. Or making sure this is real, because this is actually real. 

Marlon Avery: It's real. 

Sekou Doumbouya: $. That's right. 

Adrian Green: That's what I think that the government should do. Could the government do that? 

Sekou Doumbouya: I don't think they are. 

Adrian Green: I mean, the testing, the testing, the, the benchmark, like, could they be like, it'd be like a US stand or, you know, something like that? 

Sekou Doumbouya: I, I do think the industry is probably a little bit better than government for some of that stuff like. Yeah, I, I don't know. I don't, I don't know. I don't know. I feel like carbon is not like they don't, they don't scale when it comes to this type of like work. You look at like the, you know, the Lennox foundations and things like that, the work that they, that they do right now there are a lot of, like a lot of, there's a lot of work that can never be recreated in the government space there to just take volunteers to like make sure that we're, you know, that, that have the depth in these spaces that can actually support stuff like this. And I think we just need something like that for AI testing. We're so early in this, this journey, right? So a lot of these thoughts are, you know, we're just coming up with them and we know, we know the game plan though, we know what we need. So we'll see. 

Marlon Avery: I would like to just point out that we have officially, we're on our way because now we've officially in the midst of these AR topics, we've integrated DC Conversations, Marvel and now Fast and Furious. Guys, we are on our way. The big three. 

Adrian Green: We're hitting, hitting our target demographics, you know. 

Sekou Doumbouya: Exactly, exactly. 

Marlon Avery: All right, cool, here we go. Last one guys. It says now man, AI is starting to in power E commerce everything he said AI is revolutionizing shopping with tools like Perplexity and Google Lids says Perplexity now offers shopping recommendations and a one click purchase directly within its search results providing convenience and free shipping for pro users. Interesting. Meanwhile Google Lens expands into its physical retail letting shoppers compare prices and check inventory using their smartphone camera. Powered by Google Shopping Graph and Gemini models, these innovations aim to simplify online and in store purchases. Adrian, we'll go back to you man. What's your thoughts and stuff here? Now we got AI and shopping. You know I need those 11s that's coming out and I'm using professional to go get them. 

Adrian Green: Yeah, well I'm gonna jump on this now. I'm gonna come in hot because this, this upsets me because this is what I didn't want anyone to do yet before you know they're doing it, they're doing it already. No, no, no, because basically what they're doing is they're, they're like, they're cracking, they're cracking the, the door into basically paid for results in their in perplexity results. So we know that now we can't do Google. Google took 20 years in order to basically go full sith and just give us all ads. You know, it took a while, right? Star Wars. Boom. All right. But it took, it took a while. And I'm like, Perplexity, y'all. What y'all get started yesterday, you know what I mean? Like, it's been less than under four years, and already you're going to be crowding the search results with ads. Boo. Boo to this. Boo, boo, boo. There are so many other features that I demand that you roll out that will basically change the game when it comes to search, rather than giving me shopping suggestions before getting to what I need to get to. No, not, I'm not a fan of it. You know, that's first and foremost, I think. Now Google Lens, on the other hand, that's pretty cool. You know, I like that development there. But so Google, we're still cool. Perplexity, not. We'll talk later. 

Marlon Avery: We'll talk later. 

Adrian Green: But right now, I'm not a fan of you is Perplexity. 

Marlon Avery: You're a Ben Affleck. 

Adrian Green: I've never thought of it that way. Is this the beginning of my villain arc? Is this the beginning of my villain part? I didn't. Yeah, possibly, possibly. 

Marlon Avery: You know, so I, I, I, I like that view and stuff of it. I'm a, I'm gonna add to that and then step kind of to the right in a different lane up here. I will say this quick, quick question. Just do we know does do Perplexity, do they use their own search engine or are they using Google or Bing or do we know, do they use their own. 

Sekou Doumbouya: Yeah, I thought they built their own search engine. Like they have their own, like, Crawler or something like that. 

Marlon Avery: Okay, cool, cool, cool. So I'm, I'm gonna step into this lane. I think that this is, this, this is going to become even more challenging for the small business owner because the small business owners speak, you know, to agent point a decade, you know, getting their store search results on page one, you know, and now you have tools like from Microsoft. Microsoft, when they ask you when Microsoft. Microsoft doesn't use Google search engine at all. They use Bing. And then when you're looking for places, they use Yelp and stuff from that. And so now you're talking about. Perplexity is going to be using their own everything, their own search engine. And then Google still has their. So now you have three different, you know, areas where the small business owner has to figure out how to get their business in view, you know, and so for that, I think this going to, I think this is going to start to mirror how social media platforms are now, you know. 

Adrian Green: Yes. 

Marlon Avery: From search engines, everything. And so literally business owners, creators, like, where it may be, you know, to get presence, you're going to have to do strategies and adjustments, everything per model, you know, per platform. The same thing we do now with social media. 

Sekou Doumbouya: Yeah. 

Marlon Avery: And I think it's definitely going to add a new level of complexity, you know, for your everyday business owner who are simply just trying to keep the lights on, if you will, which also too. Watch this. Which also too creates new jobs. Because now you're gonna have some agency that's gonna tell you that we can get you at the top of those results. You know, each platform, each model is something as well. And so yeah, it's definitely, it's gonna, I think it has some pros and cons though, to it as well. But I think we're now, I think we're now on track to see models somewhat mil. Mirror how AI platforms are, how social media platforms are, where basically to build an audience or to get recognition, you'll need a different strategy per platform. 

Sekou Doumbouya: Yeah, I think actually might be good for our show to get like someone that works in SEO space on here to talk about what they're experiencing first. I know somebody, I had a feeling you did, but yeah, it's pretty crazy. This is an area that's being completely turned upside down, which is the thing that we're seeing even on the Google side with how they've integrated AI, where every result you make, every request you make, or just about every request you make will have some AI response to it that you have to look at, which is like it's really pushing against their business model. Right. And it's almost like they are promoting something that is decreasing the value of their own product or their largest project or product that brings in the most money for them. So it's, it's really interesting. And now that you, now, now you have the fact that companies need to make sure that they can be, you know, either scraped or there's deals set up with them is opening a door for small places like a plexity to make a deal with different companies to be able to scrape and get their information and things like that to be integrated to shopping. So it's interesting. This is a big paradigm shift. There are new leaders that I feel like are going to emerge from this. Whether it's going to be perplexity or not, the fact that they're going for integration for this, this, doing this this early. Like, I know there are people who use perplexity. I use perplexity a lot for a lot of different things. Right. But like it's not mainstream. Right. The fact that they're going in this particular direct, this direction first without, you know, continuing to build their, their base, we'll see if that's a successful or not play for them to make. But I don't know, I, I think we talked about this also like a year ago. Like, hey, there are a bunch of tools that are coming out that they're giving away for free. The free lunches will be gone at some point, so enjoy them while you got them. 

Adrian Green: And we talked about how long is it until we start getting ads in our GP chat, GPT results. 

Marlon Avery: Amazon's doing it already. Yeah. 

Adrian Green: Oh yeah. What does that thing called? Rufus. 

Sekou Doumbouya: Rufus. 

Marlon Avery: Yeah. 

Sekou Doumbouya: I actually, I like it. I like going to a place that I'm here for shopping and it's showing me shopping. Like that makes sense to me. Like I've come to your site with the intent to spend money and you are helping me to spend money. That, that's. 

Marlon Avery: But have that have a been targeted? 

Adrian Green: Well, mine. 

Sekou Doumbouya: Yeah. 

Adrian Green: I don't know. Like, I feel like, I feel like kind of. They're definitely, I mean, definitely in the right category, but it's nothing. At Amazon, I've never gone to the page. I've been like, oh, I gotta have that. Never. I've always gone there with a purpose, you know what I mean? So there, that's not working for me. The, the home page is not making me buy anything. But like, I usually, you know, go there with a purpose. But, but, but they'll be like, pick up where you left off, you know, you were just searching this last time, you know. 

Marlon Avery: Yeah, I know. I mean they don't, they show ads, but they don't show really ads in this form. But I know Tick Tock algorithm, you know. Oh my God, bro. Like when they have somebody like, you know, promoting something or whatever, and if you look at that, that Tick Tock shop page, more than three seconds, boy, you better get some variations of that and you know, and stuff. And so, and that way. 

Adrian Green: So bad. 

Sekou Doumbouya: I just, I despise, I despise the. I don't know, I'm not, I'm not a fan of that. I'm not a fan of the, the ads mixed in with the social media aspect of it. Because one is like, you know, I'm like, maybe I'm scrolling through the app, I'm looking at it and then someone saying something I think is, you know, interesting, useful. But it's a sponsorship. Like, it's like, like I can't it it for me, it like breeds distrust. Like, oh, this is the best treadmill ever. No. Maybe I am looking for a treadmill, but I'm not going to believe you because. 

Marlon Avery: Wait, are you talking about like it's like it's a paid sponsor or it's like a user, an actual user promoting a, like a treadmill? 

Sekou Doumbouya: I can't tell it really, truly, I really can't tell the difference. 

Adrian Green: And usually they don't always disclose either. 

Sekou Doumbouya: They don't always, they don't always disclose. And that's the, that's probably the big thing, right? Yeah. So that's, that's challenging. That happens on, on the YouTube side too, which I also like. If I see that, like I'm immediately flipping past it because, like, I didn't come there to shop. I came there to laugh at the memes. So came for the memes, not for the treadmill. 

Adrian Green: Yeah. 

Marlon Avery: Oh, man. Well, guys, man, we are now at our platinum level because now we've added Star wars to that trilogy along with our topics, you know, Please subscribe. Man. Today's show, fellas, man, was awesome. We, we had a conversation around, you know, AI agents in enterprise and where we see that world and stuff going. We've also, you know, talked about Ben Affleck, Ben Affleck's comment on where he see AIs in Hollywood. He don't think it will make a significant impact anytime soon. And then we went on and talk about the Beatles being nominated for a Grammy, which an AI assistant song that has been created. And then from there, guys, we discuss Google's Gemini challenges. Google Gemini experimented model and the benchmark status is quote unquote, that it has surpassed capabilities of GPT4 omni. And then we closed out here with talking about the, the combination of artificial intelligence and E commerce or anything with tools and solutions that's been built by Perplexity and Google. And so fellas, man, it's, it's been grace, man. Give us, give us some closing arguments. Give us some closing statement stuff here. 

Adrian Green: Be safe. Train your agents well. 

Marlon Avery: Be, be nice. Be nice to your agents. Be nice to your agent. 

Sekou Doumbouya: Who knows, one day they may have artifacts, has some AGI they might tell on you. So be nice. Be nice agents. Yeah. 

Marlon Avery: Oh yeah. Sisters get shut down. Okay, until next time. You can follow us here. You can say for me, I am. I am Marlon Avery on all platforms. Adrian. 

Adrian Green: You can follow me infamous Adrian on Twitter and Green Lantern. Green underscore Lantern on Tick Tock. 

Sekou Doumbouya: Yeah. Yep. I'm Saku. You can follow me on Saku the Wise One on Tick Tock Twitch. And yeah, I think that's about it. Oh, well, yeah, those two. 

Marlon Avery: So. 

Sekou Doumbouya: Yeah. 

Marlon Avery: And then we are AI with friends. I have a friend's podcast on all platforms forms as well. All right, guys. Appreciate you guys. 

Sekou Doumbouya: All right. 

Adrian Green: Appreciate you. Bye. 

Sekou Doumbouya: See you.