EP9: Anthropic & Meta's Defense AI, Google’s Jarvis Leak, UMG’s Music AI, & AI Revives Folklore

In this lively episode of "AI with Friends," hosts Marlon Avery, Sekou Doumbouya, and Adrian Green engage in a lively discussion on the intersection of AI and various industries. They explore Anthropic's partnership with Palantir and AWS to enhance U.S. defense capabilities, and Meta's move to offer its AI models to government agencies. The conversation shifts to Google's Project Jarvis and OpenAI's A1 model leaks, questioning whether these are strategic marketing moves. The team also delves into Universal Music Group's collaboration with Clay AI, aiming to integrate AI in music production without overshadowing artists. Additionally, they touch on the ethical implications of AI in labor markets, highlighted by Perplexity's controversial offer to the New York Times. The episode wraps up with insights on AI's role in preserving historical artifacts and its potential impact on the future of streaming and gaming.
----------------------------------------------------
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!

Chapters

• 0:00 Intro
• 10:33 Anthropic and Meta's AI Models for U.S. Government
• 15:47 Palantir's partnership with Anthropic
• 20:09 Government Cloud Transition Challenges
• 33:22 AI and Flight Booking
• 38:28 Universal Music's AI Focus
• 1:02:02 Perplexity CEO Controversy
• 1:09:11 Presentation on Technology and History
• 1:18:57 Language Nuances in Translation
• 1:26:53 Xbox AI Integration

Transcript

Marlon Avery: Might check. 1, 2, 1, 2. What's going on, fellas? 

Sekou Doumbouya: Hey, how's it going? 

Adrian Green: Hey, how's it going, man? Going good over here. 

Marlon Avery: Oh man, Doing excellent, man. How's our weekends, everything so far? 

Sekou Doumbouya: That's good. We surviving? 

Adrian Green: Surviving. Surviving is the word. Yeah, doing good over here. Very busy. How about you? 

Marlon Avery: I'm. I'm partially snowed in, you know, slow down in a weird place. I mean Albuquerque, in Albuquerque, New Mexico and had a conference I was speaking at earlier today and you know, the conference was good. The snow was unexpected, you know, for not only myself but people who live here. It was very unexpected as well. So shout out to Albuquerque. 

Adrian Green: Very cool. 

Marlon Avery: Very cool. 

Adrian Green: Yeah, I've never been out there. Did you see an armadillo in the snow? 

Marlon Avery: Is that, I mean, is that what Albuquerque is known for? And no, I did not. But is that what they're known for? Is it like you know, seeing a moose in Wyoming? 

Adrian Green: Where was Breaking Bad? Where was that taking place? 

Sekou Doumbouya: Is that Albuquerque in Arizona or. Oh, maybe it was Albuquerque. I don't know. Only thing I know about Albuquerque is Bugs Bunny. That's where he always went and was. 

Marlon Avery: Goodbye. Absolutely. 

Adrian Green: You gotta make a left. 

Sekou Doumbouya: I make a left. Albuquerque. That's all I know. 

Marlon Avery: Good. All the way by, man. That's funny, man. Any, any exciting projects you guys working on, you know, throughout the week, you know? 

Adrian Green: Yeah, continued my work with the old replit agents there and came up with an interesting which was to start small, having your replicate agent create something small like a module and then have it create documentation for it and then feed that into another replit agent that you make and say add this to an existing application or add this module to a new maybe CRM backend or something and you just piecemeal build as you go and you know, it's an approach but I've been having fun making a project with that for social media posting and been able to get, you know, kind of front end, back end pretty much actualized, nothing scalable or you know, ready to go yet. I just got a. Actually I just got a notification in the mail that 100% of my replit usage has been used so. 

Sekou Doumbouya: Impressive. Impressive. 

Adrian Green: Yeah. 

Marlon Avery: So with, with that too, I think, I think so many people are trying to figure out this automation around social media, you know. So you know, if you can able to like really put something together there, it's a tons of tons of customers that you'll have access to because I think people understand the importance of it. But just the day to day management of it is what everybody's like, you know, you're trying to avoid it. 

Adrian Green: The big sticking point right now is Instagram because in order to communicate with them still you have to use a private API, which is, you know, there's ways to do it, but it's not the best approach. It's not an official approach. So it just makes a risk of, you know, launching something, you know, to the wider public or whatnot. It makes that a little bit less promising because they could change their API at any moment. So that's, that's the only drawback. This will be used just specifically for, you know, my company and you know, selected others who, you know, no, we're not gonna do sweepstakes. We're not, we're not gonna do it. But anyway, it's been a fun project. 

Marlon Avery: You know, tune in. I mean this, this may become a AI with friends, a private workshop for the community in the future. And so that could be a thing. Seku, what about you man? Building anything interesting stuff there lately? 

Sekou Doumbouya: Not really building, just being a user. I just took a week off and started using some of this time to like, I don't know, I do find doing free thinking with ChatGPT voice pretty, pretty useful. So I just going through my different thoughts, just analyzing them, putting them through. Of course, like you know, you don't have access to you know, 01 preview for that particular feature set. So always take that, you have that extra steps like oh, I wonder what 01 preview would say. So you go through your whole process, you copy it, you stuff it in 01 preview and hopefully you get something interesting. So far it's been, it's been pretty good. 

Marlon Avery: Do you find yourself using, still using for Omni over one preview, like on a day to day basis? 

Sekou Doumbouya: Yeah, I hit it at least once or twice a day. I can get, you know, most things don't require, don't require that. I've actually been having a lot of fun on the. Was it the search feature on on Chat GBT they just released? Which is. Yeah, I'm pretty good. 

Marlon Avery: I would like to point out that I'm the first one that got access to the, on this team. 

Adrian Green: It's up here. 

Marlon Avery: I've had access to chat search for probably like four months. Three. 

Sekou Doumbouya: Four months. Four months. Wow. Wow. 

Adrian Green: Now give us a breakdown like give us a little bit like so chat search like what's that? 

Sekou Doumbouya: Well, I like for me I just use it just for just you know, searching a web. Same way you use search web, Google or actually it Actually is more comparable to Perplexity if you're searching for that. Right. Give you rich results and links to all the different things that connect to your different search. And it's super useful. And this direct competitor to Google. It's way more useful doing a search on OpenAI than it is on Google right now, which I'm sure that can't feel good. 

Marlon Avery: You know what's interesting? I change my main search browser to be search. I changed it and I want to see what my experience would be like. I end up changing it back to Google. Some of the things that became annoying was like for example basic things if I forget how to spell Wednesday, I throw it in Google and it comes back in the spells Wednesday versus I don't want the history of why Wednesday was created. I don't want the or doing too much. Yeah I don't want the origin story because like you know it would, it will basically Google put at the top everything and you know, you know it spell out or if I need to do a quick Google search for like restaurants, you know, it doesn't work well you know on the location aspect which has search and so or search gbt. And so yeah, I ended up switching it back. It became too annoying on my end and so I was kind of have to do double searches a lot. So I just kind of end up switching it back. But now to be able to insert the search within your prompt, I think that works a lot better. Yeah, for sure. 

Sekou Doumbouya: Yeah. Yeah. 

Adrian Green: That's awesome. 

Sekou Doumbouya: That's been my week pretty much. 

Marlon Avery: On my end. I finally touched over to real time API and so I finally got over there, you know and kind of got into those words and stuff if you will and man, I had to real. I had to create boundaries around some of the projects that I was doing and building. So I was building finish up a project for a client and I caught myself adding, I'm not gonna say unnecessary features. Features that I was trying to figure out can it really work? Can I, can it really be built? But it also wasn't, that wasn't underneath the scope of the project. Now they, they would have been ecstatic, you know, but also too would have started to kind of eat into my own budget, you know, some of my end. And so I had to kind of like reel my stuff back in. But it's interesting things and stuff around that as well. There's, there's definitely a lot of like use cases and nuances around that. It's just things you have to figure out. But once it Scales and everything. It has a significant amount of opportunities to go there with real time API. 

Adrian Green: That's cool. 

Sekou Doumbouya: Nice, nice. 

Marlon Avery: Cool. Let's get started here. Anthropic. There are some interesting things over here. Here we go. Anthropic has joined forces with Palantir and AWS to offer Claude AI models to U.S. defense Intelligent Intelligence Agency. This partnership allows Anthropic cloud models to operate within Palantir's impact level level 6 environment which is a high secure system for classified data. This, this collaboration is intended to streamline intelligence analysis and operation efficiencies, helping official process complex data more rapidly make better informed decisions and also too alone as well. Meta also has announced that it will make its llama models available to US government agencies and defense contracts. A notable move for a, you know, kind of concerns, you know, with AI aiding foreign enemies and stuff there. So guys, what's your thoughts? Everything. Meta and, and Palantir now. I'm sorry not Meta. Meta and Anthropic now has time to kind of start to help out the government inner use cases there. Seku man, we'll start with you. 

Sekou Doumbouya: Yeah, this is pretty interesting. I think that with Anthropic and. Excuse me, Anthropic and Meta, looking at this there's like, there's a lot of possibilities in like what the U. S government can actually pull from the private sector now with some of these things I think that with Meta and the, let's see Meta's use of lamb giving it to US government that means that you know, their, their models made to be trained on new data and things like that, right? Anthropic. Anthropic doing the same thing is not exactly the same because their models are, you know, they're pre trained to do certain things. So the question is like in order for those models to be useful it requires you to do training for them. Is there a pipeline now to Anthropic on some of our national secrets and things like that in order for them to run it on their gpus? Does the US government have lot like have a giant server farm of GPUs out there that we don't know about? There's a lot of questions that kind of come up come up with that. Right, so the Meta one like okay, that makes sense. Okay, they can train them train models to do very specific things, stuff like that. I can see the purpose around that. On the anthropic side there has to be some pipeline directly to them and I don't know, I'm not sure if a company that Size is ready for that, too. Right. Palantir took a long time to kind of build up their credibility to be able to go and get all these big, lucrative government contracts. Yeah, but, like, yeah, it's interesting. It actually poses a lot of questions on there also. It's more. You know, I'm using the chat GPT on my. On my. On my cell phone right now through Apple Intelligence has a little logo letting me know that, hey, you know, don't trust anything that this AI is providing you like. Yeah, yeah. Imagine what's happening. 

Adrian Green: Hopefully they won't have that logo. Hopefully they don't have that. You don't want that in defense. 

Sekou Doumbouya: Exactly. 

Marlon Avery: Does the government even put a warning label label on their type of partnership, their models? 

Adrian Green: Shoot. Oh, man, I would hope not. I mean, I would. I would hope that everything would be so, you know, everything. I would like that for everything that the government or the military uses to just be accurate and fast. You know, I mean, like, that's it. Yeah. Hopefully they wouldn't need to, you know, nothing. 

Sekou Doumbouya: Nothing popped up ahead, too, of course. Why are we. Why are we, like, trying to recreate every terrible version of Terminator right now? Why? 

Adrian Green: It's like a race. It's a race to do it. 

Sekou Doumbouya: Oh, yeah, let's have the government use AI models that are coming from the corporate world. That's not the plot to any movie at all anyone's heard of, right? 

Adrian Green: Yeah. We'll know we're in trouble if we see the cybernetic arm in the case. Remember that was in the movie somewhere. Like the first one, you know, it was like, in the case. You know, if that pops up on someone's zoom call, you know, we. We're in trouble. 

Marlon Avery: Nobody. Nobody's trying to figure out how to create weapons of mass destruction here. You know, that's a no. 

Sekou Doumbouya: No. We won't let it make decisions on nuclear codes now. 

Adrian Green: Definitely not. 

Marlon Avery: Adrian, what's your thoughts on this? 

Adrian Green: Well, pretty much the same as Sekus, but, you know, a couple more points. One is, it's interesting that, you know, Palm Tier is partnering with Anthropic over OpenAI. And, you know, what is the reasoning behind, you know, that? You know, OpenAI is supposed to have, you know, the most, you know, accurate model that's out there, like the smartest one. So them going with Anthropic, which everyone has been impressed with Anthropic, so. But it could, you know, really speak to something that's really specific. Like what made them more likable than going with Open AI, which just seems like it would be, it would make more sense. It seems like the anthropic side, you know, they're, they focus on things that really specialize, like, you know, really, really specialized tasks. So I think that may have given them the edge and, you know, when they're figuring out what's the best way to, you know, augment what we do, enhance what we do with AI, you know, and Palantir just has like a well of information. I think that was like the first thing that we heard about them was like, oh, man, I just, like they got so much data and they're about intelligence and really having, I don't know, like mission impossible level, like information. You know, they got it, so they got the dossiers. So on the meta side of it, I can't. When I, you know, I heard about them being like, hey, look at us over here too, you know, we got you. Hey, hey, defense people, you know, you can use our model too. You know, the first thing that I thought is just. Same thing I said the other week, man, which was, it's just what's, it's what MET is doing, which is, you know, we're just going to give it away for free to take away some dollars for those people at the top until we're able to get our legs underneath us and really make moves in the metaverse or wherever, you know, this is, this is the tactic for now. It's just Nick, you know, kneecapping people here and there, you know, so that, those are my thoughts. 

Sekou Doumbouya: Yeah, it's a smart move. It's a really smart move. Right? Don't let anyone get enough steam to get a lead on you, you know. 

Adrian Green: Because it's like if I can release something that does it 80, 80 as well. You know, sometimes you, you'll take an RC Cola, you know, we're at the, you know, we're there. 

Marlon Avery: Mr. Pip. 

Adrian Green: Yeah. A doctor perky or something. You know what I mean? 

Marlon Avery: I think one of the interesting, interesting things and stuff of this as well. It said the collaboration is intended to streamline operations and also process complex data into, to make better decisions. It just seems like they're building a high level government rag application. And so that's what it seems like. And so from that, from their standpoint, I find it interesting too as well, that I find interesting as well that they are using, you know, partners that decide to partner with Anthropic, you know, over, you know, you know, other, you know, potential partners and stuff as well. So I mean, is that A signal towards. Is that a signal, signal towards you know, anthropics embedding process and everything stuff works better. You know, is that a signal towards, you know, to be able to build a knowledge foundation and on top of that, you know, agents and thing. Is it better to use anthropic and stuff in these areas? And so, and so I think it's something to figure out everything. I understand why you use aws actually. No, I take that back in the, any in the government standpoint, I would argue there's definitely potential way. There's, there's, there is better ways to use Azure particularly too because they also. Microsoft has more access to GPUs over Amazon. And so I found that interesting as well that you go with Palantir over a competitor and also too you go with AWS over Azure. 

Sekou Doumbouya: Yeah, you know it is once they've made a commitment and sign contracts to aws like changing to go in a different direction and one is, is costly but the, the government, they, you know, they, they integrate, they integrate pretty, pretty deeply into these things. Not just their technology but their security protocols and things like that. So they're slow, they're very slow to get to the cloud. Like being able to shift to another cloud would be just as slow for them. So it's not that I don't think it's an optimal, optimal choice. I think it's just based off of you know, the, the, like the, the amount of technical debt they'd have to like incur in order to, to make a right turn. 

Adrian Green: Yep. I was gonna, I was gonna say that maybe tooling as well because we always talk about the, you know, what's going to sell these at, at the end of the day, it's going to be the tooling. 

Sekou Doumbouya: Yeah, you're right. Because when you think about giving someone a model that's useful, but the infrastructure used to train a model infrastructure used to do inference on the model. Not just like GPUs. Right. Like the actual CI CD pipelines that get you from beginning to the end. That stuff is not just not easy. Like the Kubernetes cluster that's underneath all of that, that needs to get built. That stuff is pretty hard to recreate fast, I'm sure because Anthropic has to do this all the time. This really shows maybe some of the differences. Also on open AI versus Anthropic, they must have a really good tool chain over on the Anthropic side if they're able to like compellingly like do this because like I Said like it's not. Having given them a model is not super useful unless they give them the tools to train it. 

Marlon Avery: Yeah. 

Sekou Doumbouya: And like they have to have GPUs over there for it to be useful. Just giving you a generalized, you know, model is not great. And giving you a model that someone is pre training, you're taking it in, you're using it like that's, it's a really nice place to get like supply chained. Yeah, yeah. Get a supply chain attack. Attack is to you know, attack anthropic. Right. So. 

Marlon Avery: Okay. 

Adrian Green: Yeah, we got, we got you not to give anyone any suggestions. 

Sekou Doumbouya: I'm sorry, we're not helping anyone. 

Adrian Green: We're not helping anyone. 

Sekou Doumbouya: No, do not do these things. Leave our US companies alone. 

Adrian Green: Please. Oh my goodness. Please. 

Marlon Avery: All right, here we go. Recent developments From Google and OpenAI hint at a new era of automations. AI capable with more advanced reasoning and task management. Unfortunately, some of these indications have been leaked as of lately. So Google's Project Jarvis, an AI tool capable of taking over your computers for everyday tasks was briefly exposed on Chrome Web Store. This tool which can make reservations, purchase products, navigate files. Recent Google represents Google's push towards AI driven task automations that stimulate human interactions within digital interfaces. OpenAI meanwhile unintentionally leak access to this power of the O1 model. A reasoning based. A reasoning based AI that can take extra time to think before re responding or enhancing accuracy. Said the new model goes beyond the capabilities of GPT4 Omni handling complex tasks like large like image analysis and data processing. Also too you guys had a question kind of behind the scenes stuff there, you know, are these leaks intentional sales strategies or marketing strategies and stuff? Adrian, we'll start with you. What's your thoughts? 

Adrian Green: This, this is not my conspiracy. This was brought to us by Seku. So I can't, I cannot take, I cannot take credit, I cannot take credit for this one but he sends, he sends us these articles and I take a look at it and I'm drinking the Kool aid too because it's, it's just the way that it's written. It's just, it's just the way that it's written. I'm looking at the one from Inc.com and the one from Tom's got here and it's just like you know, you know there is this model that was leaked and you know, the people who got their hands on it couldn't stop talking about the amazing capabilities of it. Let's go down the list. You know and it was like, oh, wait a minute, it seems it's more is like a Zynga or whatever kind of link pool that you get at the bottom of a blog now, which is this clickbaity kind of. I forgot what that company, it was like Zynga or something. I would always see like powered by, but it seems like Zoomla or something. But like it seems like one of those clickbaity kind of things. But you know, they're talking about things that really happened here. So it's, it's really thinking, you know, I'm thinking inside job too, a little bit. 

Sekou Doumbouya: Yeah, I'm trying to, I'm trying to imagine here, right? This is Google, tons and tons of like steps to do anything, to like tons of process to do the smallest thing and someone just happens to leak something super interesting to the Chrome Store, which is like folks don't just have production access to do things. You know what I'm saying? Like there's. Yeah, yeah. 

Adrian Green: To the Chrome Store. That was interesting. 

Sekou Doumbouya: Chrome Store, Yeah, like, oops, I didn't know how to use a Chrome Store. Oh wait, I work at Google. No, that's this clearly marketing, 100% marketing. They had some folks who installed this thing and they was like, oh, hey, we got some free telemetry data. All right, let's shut it down. 

Adrian Green: Always get us like that. Oh my God. 

Marlon Avery: So watch this. Past the leaks, what do you guys think of the features of this if it does come to reality? So it says that the Jarvis will have access to make reservations on your behalf, purchase product, navigate files. What do you guys think about the similar features and stuff there? 

Sekou Doumbouya: Sounds interesting. Sounds interesting. I'm not ready for it yet. I'm not ready for the AI to move my cursor or to talk with people on my behalf. Yeah, I'm not ready for it. Like, you know, you. If you hired somebody, right, and they, you know, made mistakes, maybe you know, 20 of the time, right? Like if you knew you couldn't get them, you there's no way to get them to 100. Would you still like work for that person? I don't know. I don't know. Because I got, I still have that person, right? Yeah. Now we'll have the AI in here. It's imperfect self. Like, come on guys, step it up one notch. Gain some trust. But I don't know. 

Marlon Avery: Interesting. So you're saying empathy can be extended to a human being, even if they're making 20% mistakes. But if your AI is making 20% stakes, you're ready to unplug Unacceptable. 

Adrian Green: Unplugged. 

Sekou Doumbouya: I'm just, I'm just saying future robot overlords who are hearing what we're saying now, we're just joking. We would never, never unplug you. But the thing is, we have an AI doing some things, and AI can do things faster. If I can have someone to make 10 calls and do it, you know, 20 bad through the 10 calls, right? Hey, I can make like 100 calls. 200 calls, right? That 20 means a lot more in that case than it does for, you know, on the smaller side of these things, right? So as you, as you like, become more capable, like, the error rate matters so much more because of how much more you're affecting, right? What was it really? Good example. Like, I used to. We used to work at a company that we would do, like, deployments and we would do, like, we would, like, throw up a canary, right? Canary is essentially a. It's based off of the canary in a coal mine. You put a. You have a canary go into a coal mine, and if it doesn't come back up, then there's gas in the, in the coal mine. In this world. We used to say, like, you know, you know, 10 was it 2% of fleet. 2% of the fleet. If we deploy this code and 2% of our fleet breaks, sure to be like, couple small customers that might be upset, good to get, you know, but we'll know that those, those bills are broken and we can roll out things faster and get good signal. Right? But when you have, say, I don't know, 500 servers, 2% is like a whole company. It's like a whole company is 2%. You let that number fail just so you can see if your things work. The math, like, as you. As it gets larger and it gets faster, you kind of have to, like, you can't treat it the same, you know, you can't. 

Adrian Green: You can't. And that's why you can't have that secretary who you've known forever, probably a friend of your families, you know what I mean? She's working for you now. She's a little bit slower, you know, and it's okay when you got 20 clients, you know, but we're scaling up now. Betsy, you know, we're not going to be able to run, you know, you can't write all of these emails, you know. You know, it's tough. 

Sekou Doumbouya: So we're perfect for those, I tell you. 

Adrian Green: Yeah, you can't. We've had all of these people. We're loot. We're running out of people who can install Windows 95, you know, for you to get those to work. 

Marlon Avery: It. I thought, it took me a second there. I thought you was using canary as like a keyword to application. And then when you kept talking, you said, if the canary doesn't come back up, we know there's gas in there. So they're using real birds. 

Sekou Doumbouya: It's a real bird. This is a real thing. Canary in the coal mines, they say, when they say, hey, you guys need to ship some canaries. We're talking about birds dying. So that is what they're talking about. 

Marlon Avery: So that's unfortunately hilarious. It took me a second to process. Hey, all right, cool. So with, with Project Jarvis, I'm gonna give you one application. If it can work well, I'll be extremely excited about Skip Yard. I'm sorry, Skip. Skip Lag. So you guys know what skip lag is? 

Sekou Doumbouya: Mm. 

Adrian Green: Yeah. 

Marlon Avery: Skip Skip Lad was a flight booking alternative and built by a software engineer a couple years ago. And so basically he figured out that if you are in Miami and you're trying to fly to Atlanta, sometimes it was cheaper to book a flight to New York with a layover in Atlanta. You just get up from the layover. 

Adrian Green: Oh. 

Marlon Avery: And so he, he created this whole algorithm and stuff and it worked for quite some time. Like it, it dropped. He found like significantly cheaper, you know, flights. And then I think, I think it was like American or United Airlines ended up suing them. But he won. You know, he won. Which also kind of like, you know, put Skip Lag at a higher brand level. 

Adrian Green: Now, he also had young. He also had young thugs. Lawyer, right? Same guy. 

Marlon Avery: I'm not sure. We have to confirm that on next episode, but with that, with the tool like Project Jarvis, if it has or, you know, if it, if. Well, here we go. If LLAMA creates an open source version of this, one thing I would definitely be excited about is the ability to have an AI agent that is surfing looking for those flights at all time. You know, so, you know, it's like flights are similar, like gas prices. It's like there's typically. It's the price of skyrocket up at night and everything. They're kind of cheap around 4 or 5am, you know, in different areas. Also too, your IP address could have some, you know, some impact to the prices on the into as well. And so if I can have a AI agent and also too, I can tell this agent, you know, if I'm going to a conference or whatever to search, you know, throughout the modifier sources, looking for the cheapest flight that has a window seat, you know, and stuff, whatever it may be. And then once you find it, book it, you know, and book it and stuff with the information. Just kind of simulate confirmation and so something like that. Anything where we're talking about, you know, autonomous, you know, ages, you know, or AI driven task automation is something I'd be extremely excited about. 

Adrian Green: Yeah. And it would drive Ticketmaster crazy, you know, because that's basically like what you're, you're describing is the, the ticket scalping kind of model or the, the reselling market or whatnot. They have bots to just look at, look at the concert and just like constantly try to, you know, figure it out. As soon as they, you know, figure out a way to, you know, bump up that security, it adapts with them. You know, it's like, it's like, you know, so it's a, it's tough because we're gonna move, move into a place where it's going to be AI that's doing all of that, you know, really interactions like that online, like whatever, you know, you know, I'm thinking like a guild of hackers runs the scalping operations that were the ticket resale operations. Now, you know, that's really going to be offset by AI so like being. 

Marlon Avery: A sneaker head all over again. 

Adrian Green: Yeah. 

Sekou Doumbouya: Oh, man. 

Marlon Avery: Looking for flights. 

Adrian Green: Grown folks off grown folks. 

Sekou Doumbouya: Yeah. I just realized that captures are gone. They're gonna be gone. 

Adrian Green: Yeah. 

Sekou Doumbouya: Captures will never be used again once that actually is released because there's no way to like, you know, the captures assume that someone is scraping a site. Right. And protecting against scraping, programmatically scraping it. Right, right. Imagine, imagine having a AI running on your system. A little boss comes up and says, hey, which one of these are cats here? And just fills in that information like there's, what can you do with that? There's no way to stop that. So that technology is, that's, that's completely useless technology now. 

Adrian Green: Yeah. 

Marlon Avery: You know what's interesting was, was, was the technology like caption, was it too early? Because like I, I understand the use case and so forth for it to come out, but, you know, now we're kind of entering into the world of, you know, AI driven task automation. I would argue you probably need it more so now than ever. 

Sekou Doumbouya: I guess if it's, I don't know. It's. The thing is, the thing is they have, these computers have vision. Right. They can see. That is a core thing. It's different from looking at A page being able to see things. Meaning like you know, deducing things based off of what's on your screen. And that is. I think that's the place. That's what, that's what's kind of changed. You always could use like there's like user acceptance testing solver Selenium. Right. Selenium can move a cursor and take a screenshot of the page and then do an action based off of that screenshot, which is technically using some machine learning. Listen very basic things. But with these more advanced systems I can't see how that's going to survive. I feel like we built assuming that the computers would never be able to see. So. 

Marlon Avery: That'S a great point. Yeah, that's a really great point. Having the ability to see. I'm sorry, we gotta be running. 

Adrian Green: We like we should have never taught the computers how to drive. This action movie starts. 

Marlon Avery: Here we go. University Music Group has a created a new partnership with Clay, which is a AI music company. So it said the Clay is going to develop a large music model LMM that respects artist contributions and redefines AI's role in music. Say Clay aims to create an AI driven ecosystem for generating music experiences that complement rather than compete with the artist's original work. I said a partnership aligns with Universal Music Group goal of driving innovation while upholding ethical standards in the music industry. You say Clay's team, including experts from DeepMind, ByteDance, aims to make AI music production a seamless part of daily life, emphasizing cultural relevance and ethical considerations in AI generated content. So the collaboration is the latest Universal Music Gear series. A partnership advancing generative AI technologies in ways that honor human, human work. Let me start with this one. What's, what's the, what's the professional way of saying I told you so? What's, what's the, what's the professional way of saying that? 

Sekou Doumbouya: I'm glad that we all have come to the same conclusion. 

Marlon Avery: So I'm glad that we all came to the same conclusion because you know, I got, I got some heavy pushback on this like a year ago, you know, that we basically will enter into this partnership of this AI, you know, driven ecosystem and stuff. This makes sense. I would say this makes sense. Is also concerned, you know, this makes sense for Universal Music Group to partner, you know, with a. Creating their own LLMM except as well. And so I get it, I get it. I can see how this can work. I can see that you are now hiring a new artist, a new producer and when they kind of have when they have, you know, writer's block and they kind of like, you know, tune in over there to the LLM to kind of like help them write out some of the, you know, the rest of the layers or the hook and stuff what it may be also too, along with some of like even some of the songs. I mean, Adrian, you know, you and I was talking about Timbaland, you know, and he's starting to use was it Sona, you know, as well, sooner. And then we also saw another artist who's kind of like developing, you know, some AI driven content too as well. So again, I, I just see this as a tool. I see this as the ultimate, you know, assistant. And I think now these, these tools and systems are becoming specialized into focus areas like music. There are some, I think the definition of compliment the artist rather to compete is very different per artist. And also too, you know, how do you, how do you distinguish that? What's the definition of that? I mean, because one could say scanning the first two paragraphs of a New York Times article versus copying the entire segment is complimenting it instead of like, you know, competing with it, you know, or you go in or you just hire New York Times writers, you know, to as well. And so I think that definition is going to be unfortunately defined by the music group and not the artists. And I think that's kind of where some concerns and stuff might be. Adrian. 

Adrian Green: Cool. Cool. Yeah. So I work in, you know, adjacent to the music industry, so I found this one particularly interesting. So what Universal Music has gone through this year, you know, has been a lot in the news. There's been a lot of layoffs. There's been a lot of really mergers, a lot of acquisitions, a lot of shifting of the org structure. So a lot of long time people that have been there for, you know, eons, you know, were a part of some of your favorite songs are out, out the door now and they're, you know, Marlon's been saying it too. You know, the shift seems to be, you know, them focusing a lot more on the AI side of stuff with, you know, them suing people, companies like Suno and everything. But meanwhile they're doing business with these, this AI company, Clay, which when you put them next to their recent acquisitions or their recent partnerships, which include Endel, which is a company that's about generating music based on biofeedback, like your heart rate, you know, other great, you know, television. Yep. End up, end up. 

Sekou Doumbouya: Wait, wait a second. Wait, is that the same one that I use for that's the one that I used to go to sleep. Is that the one you talk about? 

Adrian Green: I think it's the same one. I think it's that. I think it's that same one. So they're gonna work with Universal Music to make basically this new. This new model where you'll generate the music based off of the inputs from your body. So almost like a wearable. 

Sekou Doumbouya: Wait, wait, wait, wait. I feel. I feel attacked right now. I had no idea. I thought this is just like a simple app to help you go to sleep with some white noise. 

Marlon Avery: No, it. It is that now you're about to get a music. You better get a music version of that to help you go to sleep. 

Adrian Green: Yeah. 

Sekou Doumbouya: Oh, no. Oh, no. They're gonna record my pulse. Oh, no. 

Adrian Green: It could be your pulse. And it's going to be like, Aretha. You know what I mean? It's gonna be amazing like this. 

Sekou Doumbouya: Oh, man. 

Adrian Green: But so, you know, one. One more thing Marlin's always said is that, like, you know, music labels do not innovate. They don't create. And this is, you know, very much like, okay, well, let's. It's them partnering with the creators of. On the AI side of it. So if you look at the companies that they have, they have Sound Labs, they got Indel, they got Band Lab, right? And they also have, so Sound Labs, Indel, Band Lab, okay? Band Lab has. It's basically Fruity Loops, so Fruity Loops, like company, but they also have a hardware device, right? So I envision a world where Universal Music, Band Lab, they're able to maybe, you know, be a part or be a part of the, you know, whatever music you generate on that device will somehow be licensed through Universal Music or some kind of, you know, money flowing back that way. And I say that because they also have a partnership with this company called Pro Rata, which says that it can listen to an AI generated song and figure out the correct attribution on it. Now, when I brought this up to Seku, he was like, cap, you know, that may be Cap there. So I, you know, I'm, you know, hear more about what Seku has to say about it. But that if that is true, that could be pivotal in, you know, a real, you know, if they're able to pull that off, you know, that would really change the game as far as AI Generated music. You put it through a company like Pro Rata, and they'd be able to figure out, all right, you know, Beyonce gets this tail, Swiss gets, you know, whatever. So it's it's really interesting. They can. They have a couple of different products and another one is a wearable. I'm thinking Universal Music Group could actually, with this, you know, chest of. Or this roster of players could actually release a wearable possibly. 

Marlon Avery: So is attributes are they. Is that a technology. 

Adrian Green: It's like who wrote the song like. Or who is. Who owns the publishing on said song. So sometimes it'll be like, so who wrote it? The writer, producer, composer, performer. You know, if things get split up according to rights with music and everyone agrees to a certain percentage. 

Marlon Avery: So in a technology landscape, this is either a contract or a JSON, if you will. 

Adrian Green: It could be. I mean it could be. And you made the argument. I know where you're going. If this is about the blockchain, it's going to Blockchain. Okay. It's a blockchain. Yeah. 

Marlon Avery: Yes. I, I don't. This is my concern again. If you're going to do these things and you're going to streamline and if you're doing things ethically, there's no reason not to put these things on the blockchain for the artists. There's no reason whatsoever. You know, it's, it's. This is the same category as. There's no reason why we should not be able to vote from our phone. None. 

Adrian Green: Yeah. 

Marlon Avery: You know, and everything. It's like why, why won't you make this change? There are some things behind the scenes that benefit the people who doesn't want to see that change be made? You know, and I, I see music labels in that same category, you know, of doing so. It's particularly too when now we're going to live in a world where the fans will start making, you know, AI AI music as well. 

Sekou Doumbouya: Yeah. I think the music industry and the movie industry, that whole area is super unique for AI because in other places when they're innovation, it's like innovation that happens within industry, but it transfers maybe to another place. Right. But in this case, you have music created by artists who really get paid best based off of their songs being the top songs and being played a lot and all the different things that go with that. Now you have AI which can create a hit song. You're like a BBL drizzy or something and that could be the top song. And now the artists are back of this place where even if they're getting attribution like you look at what's happening from selling CDs, which artists love the idea of cds you can sell for 15, maybe you just had to pay the company that did the distribution of these CDs. But the margins were higher. Streaming where the margins went lower and the amount of winners became smaller within there. Now they're going to get even smaller share because they're going to get a fractional share of a song. Like man, like it's gonna be even less winners inside there that actually make it so that you know, you know, a Beyonce or one of these artists that end up getting added as a digital signature within these songs. Maybe they'll get some money. But like, like how do you not the same. How do you live in the song? Like you know, they. They use your portion of the of that song 10 times versus you know, this other artist that use, you know, 20 times within the song. Like what's that. That model? To me feels like there's so. I don't know. Why would you. 

Marlon Avery: Yeah, so also for. For context as well, here. Here's the breakdown of some of some. How the artists get paid or the amount artists get paid per stream. I don't even know how to like say this. Okay, Pandora is $0. And what is that? $0.00 and 00.13. Yeah. How do you even say that? Spotify is $0.0031Amazon Music is 0.004 40 cents or 40. Whatever it is. Desert $0.011 YouTube Music is $0.002 Apple Music is $0.008 Interest $0.00.01284 and so when you times these times a million streams, you know, it's like it's not a lot anything even for a million streams, you know, and stuff like that. Like you said, the payout amusing record, amusing CDs being so. I mean people could retire essentially, you know. But now to have a million streams and like you said, I think that that is going to get diluted, you know, and stuff over time. 

Sekou Doumbouya: Yeah, that's why, that's why artists end up going on tour so much. That's where they actually can actually control the amount of money that they make and pull in large amounts of cash. It's interesting. But yeah, I don't know. The incentives to be artists is. 

Marlon Avery: I just thought of something. That payout goes to the record label. 

Sekou Doumbouya: Then I guess another fraction. 

Adrian Green: That's not funny. 

Marlon Avery: All right, whatever. So that payout goes to the record labor and then the artist gets a percentage of that. 

Adrian Green: Some artists have direct deals. Like so we, you know, we deal with artists with direct deals. So whatever percentage, you know, we'll take out of that, you Know, our, you know, kind of a working percentage there. But some artists don't go through a record label. They have direct deals, you know. 

Marlon Avery: Yeah. 

Adrian Green: And just do. Just do distribution. 

Sekou Doumbouya: Wow. 

Adrian Green: But yeah, and because that's. That, That's. That's the state of the record, you know, the music industry, it's like, you know, you would be using a record label, you know, so excited about signing that record, you know, deal, but that advance you got to pay back. And, you know, it's not like they'll. They're not really all the time rushing to put out your music either. 

Marlon Avery: You know, Adrian, you've been in the. The technology world of entertainment for quite some time and stuff now. You've seen these worlds advanced. In your opinion, in today's social media society, what are the benefits, if any, to sign with a regular versus being independent? 

Adrian Green: It depends. I mean, sometimes you need the money, like a rec.1, the one thing that record label can offer you is two things, that is cash advances. And they can offer you radio play. And radio play is still big, still matters, and it's hard to get by. 

Marlon Avery: Yourself in this streaming world. Radio is still a factor. 

Adrian Green: Radio still out there. It still matters, you know. 

Sekou Doumbouya: Wow. 

Adrian Green: That's still a factor. 

Sekou Doumbouya: Where does I heart radio sit in this? I guess. 

Adrian Green: I guess between both worlds, you know. 

Sekou Doumbouya: Yeah. 

Marlon Avery: Yeah. They also do concerts too, as well. Interesting. How would you compare. If you can, how would you compare the revelant, the, The. The relevance of radio today compared to TV commercials? 

Adrian Green: Radio way over TV commercials. I mean, I don't, I don't look at my day. If I've looked at a commercial during my day, I had a bad day, you know, my day wasn't so great. All right, I skip them, okay? I'm skipping the commercials. 

Marlon Avery: I'm not. 

Adrian Green: I don't. You know, that's just me, you know, I. But if I'm listening to the radio, you know, which I still do, I'm still listening to the radio, you know, I don't. 

Marlon Avery: That's why I'm asking this question. Yeah, that's what I'm seeing. Like, I, I haven't listened to. If I listen to the radio, that means the Internet. My Tesla has went out. 

Adrian Green: I got local radio stations that I grew up on. Like, so I've seen the DJs come and go, and it's a, It's a fun. It's been a fun little kind of listen for me for, like, so long. And I'll. 

Marlon Avery: I'll take it a step further. A friend of mine's got into the Tesla and he turned on the radio application. And I didn't know that existed. And I had the car for like a year, you know, because it was just like either it was either podcast or, like streaming, you know, you got. 

Adrian Green: A tape deck in there, too. You got a tape deck in there, too. You don't even know about, you know, just. 

Marlon Avery: Just ask. 

Adrian Green: You never. 

Sekou Doumbouya: Just realizing I've never used the radio on my car either, on my Tesla. I've never. 

Adrian Green: Oh, man, I've never thought of that. Down here. I'm really on the dirt roads. 

Sekou Doumbouya: Yes, I do miss. I do miss talk radio. I'm sure that's still a thing, Ricky Smiley in the morning. Yeah, yeah, I'm sure. I'm sure that's still huge thing. I have not listened in a long time. A lot of folks are going to series XM at some point, right? They become a big player of that middle ground. Not quite radio, but something else. Yeah, yeah, yeah. 

Marlon Avery: So interesting. 

Adrian Green: You know, it ramped up, though, during those Columbia Record. Columbia House. Remember those? 

Sekou Doumbouya: Oh, yeah, yeah, yeah, yeah. 

Adrian Green: When that was going, dude, you know, the record labels were getting paid, man. Like, they had to. They had to. They had to be like a big grift, man. Like, everyone was laugh. Was wrapped up in a scheme with the Columbia Record House. You know, your friend would get it. He'd be like, yo, I got the new everything. Like, you know, get, get like me. Then a couple of months later, you know what I mean? The. The. It changes. You realize you owe a bunch of money more and you, you know, it was a bunch of, you know, it was confusing. 

Marlon Avery: I'm. I'm really stuck on this Seiko. What was last time you listed the. 

Sekou Doumbouya: Rodeo, the last time I listened to the radio? I don't know. It's been more than a year, probably. It's probably more than a year. Yeah. Oh, you know, wait a second, Wait a second. I have listened to the radio, but it's in my wife's car. 

Marlon Avery: That's not intentional. That doesn't count. Well, no one's. 

Adrian Green: No one's getting excited about turning on the radio, Marley. At the same time, it's not like I'm trying to catch, like, the Phantom. 

Marlon Avery: No, I understand that and everything, but it's like, to even hear that the music industry is still getting paid, you know, from radio plays, it's like amazing Internet to me, because also, too. I just also know. I know, like, networks are still getting paid from commercials, but it's like, you know, what are the percentages of people who are watching this. And so like I got to my hotel yesterday and just tired from just, you know, traveling, whatever. And I just kind of plopped on the couch here, turned on the tv looking for a game or something. And I lied to you, not I going through the guy. I saw fresh prints first. All right, cool. I'm gonna put, put that on for a second. And so it was on a commercial and I was like, okay, I'm gonna keep scrolling. And then I think I saw like Big Bang Theory. I said, all right, cool. I'm gonna put that on for a strong six to seven minutes. Not one piece of content play. It was all commercials. Yeah, I didn't turn it off. 

Adrian Green: It's crazy. Yeah, I would rather read. I'll read a book. 

Marlon Avery: I would rather watch paint dry than just sit there and watch commercials like all day. 

Adrian Green: Just like all different volumes, you know? You know. 

Sekou Doumbouya: You know what I think? It's not, it's not really. It's not just folks who looking at commercial all these place getting paid. It's the people like me who just like, I will pay anything and not to see a commercial. I will get YouTube Premium. So like, oh, oh, they're adding commercials to Prime. I guess I'm buying Prime's video upgrade now because I will not watch a commercial. 

Adrian Green: Prime, Prime. 

Sekou Doumbouya: Now I gotta get Prime Prime, prime, prime plus. Yeah. 

Marlon Avery: I'm the same way. I. I refuse. Yeah, I refuse. He watches even our own, even on our AI with friends YouTube channel. So sometimes when I'm like making edits, whatever stuff on there and that channel stays like, you know, kind of like, you know, it's. The profile stays in there. And now I'm like some hours ago, about a day or go by or whatever and I need to go YouTube something and immediately as pick pop up, I'm like ill like change this quickly. Just like, yeah, please hurry up out of here. So yes, I'm the same way. I repeat, I. I have additional question too. I'm gonna throw it in and stuff at the end and stuff there. I'm interested. You gotta say. So here we go. Coming back to Perplexity. Perplexity CEO has sparked some controversy. This is, this is funny. Not funny. Spark some controversy because he has offered to replace the New York Times employees who are currently in strike and stuff right now. All right, so here we go. Recently found himself at the center of a controversy after offering his company services to support New York Times during a tech staff strike. The New York Tech New York Times members who handles the essential tech functions are demanding fair wages and other benefits, it says. Although later clarified, the Perex offer was provided to do technical infrastructure rather than replacing staff. The incident has drawn attention to ethical tensions between AI companies and labor rights, especially as AI has increasingly stepped into traditional human dominated roles. Seiku, this. 

Sekou Doumbouya: Yeah, yeah, I'm not gonna say, I'm. 

Marlon Avery: Not gonna say this, I'm not gonna say this sounds like the ports kind of like all over again but I think it has some similarities. But go ahead, what you think? 

Sekou Doumbouya: Yeah, they definitely, it definitely has a little bit of that, that feel into it. Like, you know folks, I don't, I feel like I really, I don't like the idea of AI using AI threats as a way to like discipline folks. Right. As a way to like, you know, get folks in line. 

Marlon Avery: That's hilarious. Go ahead. 

Sekou Doumbouya: Yeah, it's like if you don't do this, I'll get an AI to do it for. 

Adrian Green: It's very mob like. 

Sekou Doumbouya: Yeah. So like I see that's where we're going. It's like it's not the, it's not the, going to be the last version of that that we hear. But like I think, I think what actually is probably going to start coming out like some of these AI systems are good, some are not that great. We have yet to see true products that actually can generate actual revenue that are AI like all the things we're seeing. Everything is an expense on companies balance sheets right now that has AI on it for that potential for the one day it will become useful. And man, I don't know, I think, I think the guy, I think he needs to take a little bit of humble pie and take a, take a, take a step back a little bit. But especially it's perplexity. Perplexity actually like of the AI companies that are out there, they have a, you know, was it Sonos is I think is the name of their model. Right. It doesn't rank high compared to any of the other things that are out there. Like of all the folks to talk. 

Adrian Green: Sonar. 

Sekou Doumbouya: Sonar, yeah, like of all the companies to talk. It's probably not the one to talk about talk, you know, like, yeah, come on, you got to be at least like you got to be up there with like a top model, something that's hitting some high ranks to be saying things like that. So. 

Marlon Avery: Yeah, yo, I, I can't get this out of my head. You said it's real mafia like, you know, I can just, I can just see, I can just see CEOs waiting, you know, to pull that car when it's gotten, like, too many, you know, employee complaints or like, whatever it may be, and the CEO is just like creeping up to the screen like, oh, yeah. 

Sekou Doumbouya: Oh. 

Marlon Avery: Just waiting to throw it down. 

Adrian Green: Oh, my God, that's funny. Yeah. 

Sekou Doumbouya: Oh, man. You remember, that was it. There's a South park episode that is very, very infamous where there's these guys that, the folks that are saying that people are taking their jobs, Right? I can't wait for a South park episode on this. They have to do it. I have to. Yeah, yeah. 

Marlon Avery: Adrian, what's your thoughts? 

Adrian Green: Oh, man, it's. I think that the, the humor is not lost on it to me, because we, we covered earlier back in October. New York Times suing Perplexity. New York Times sends AI startup Perplexity cease and desist notice over content use. So, you know, the. They sued them over their, you know, mid level model. They're. They're basically saying that, you know, you used our content in order to train your models. They got sued. So, you know, I see the humor is not lost in perplexity. Then after them, you know, really, you know, perceivably not giving a fair wage to their tech team there, then saying back and being like, well, you know, I would love to generate articles for you if you wanted to. I got plenty of them over here. I would love to help. You know, it seems like you. You got a lawsuit against us right now. My hands are tied. My hands are tied. But I would love to help. I would love to help you. You know, so it's, it's very funny to me. It seems like, you know, just, you know, a good kind of humorous banter, but because it's like, you know, they're not like the top model out there. So it's a. That's what I think. I think that the humor. The humor is definitely there. It's not lost on me. 

Marlon Avery: This is like. This is like sending flowers to the person that you cheated on after they get cheated on again and saying like. 

Adrian Green: Hey, that's a Drake move. That's a Drake move. Take care. Take care. 

Sekou Doumbouya: But on top of that, it's. It's AI. So we're gonna be serving you the same thing that you did, the thing we took when you were gonna serve it back to you. 

Marlon Avery: Oh, yeah. I. I'm not sure. I'm not sure if this is, like, I'm not sure if this is an AI with Friends thing that we find humor in these type of things, you know, this. It may be, but it's like, it's. It's interesting, you know, it's. It's like. I don't know if you're trying to like Band aid the situation by kind of making it offer, you know, and also too. Who to say. Who to say is that he. He's not the. Where. They're not the only one who made the offer. You know, all the rest of the office could have just been private, you know, and so. But I do find it interesting, you know, two things. I just find two things interesting. I find the, the. I find the. The initial offer interesting and I find the no response from New York Times interesting as well. It's not, it's not New York Times that created like, create the, you know, the, the surge and stuff of this article. It was people who saw that this, that this was made to New York Times that is like surging that this is a problem. Anything. But also too New York Times didn't. They didn't make a response as well. So I find that as well interesting. You know, so currently right now, you know, currently right now, perplexity is left on red, you know, or, you know, they put privacy on the DMs. 

Sekou Doumbouya: Yeah, we'll see. We'll see what, we'll see what comes out of this. Right? But you know, silence is a statement too, right? 

Marlon Avery: That part. 

Sekou Doumbouya: So, yeah. 

Marlon Avery: Very cool. So here guys, I'm gonna lead this one and stuff here. As you guys know, I am. I'm. It's currently in New Mexico, Albuquerque. And I had the privilege of speaking at a. Had the privilege of speaking at American Folklore Society. And so if you don't know folklore, individuals are all focused around, you know, keeping and preserving things like, you know, historic artifacts, communities, languages, you know, and, you know, a lot of these things. And so I also have the privilege of a client of mine and a partner of mine is the famous Viscoy Museum, you know, down in Miami. And so if you haven't seen Visaya Museum and Gardens, it is a beautiful, beautiful place right there also in the water too as well. And it's initially started by individual by the name of James Deering, which is his house. And he was kind of building Garner and so they turned it into, you know, a museum as well. So, you know, today I. I had some interesting discussion. My presentation today was basically how do you take individuals like historians and partner with today's technologists and technologist individuals and create a, you know, a better or unique experience. And so the example of that I'M working with this guy museum and the. The. The head of technology, Jeff Gwynn, reached out to me some. Some months ago and said, hey, you know, Baskaya. We acquired a bunch of letters, historic artifacts, from the founder of James Earrings. And these letters are from, like, the late 1800s or 1900s, essentially. You know, these are a lot of way. These are text messages. Some of these, you know, these. These are, you know, late 1800s. So a lot of these are text messages. These are communications, you know, back and forth. These are deals, purchases. You know, stuff happened back and forth by text message. 

Adrian Green: You mean like letter writing? Like letters. 

Marlon Avery: Yeah, I'm so. Yeah, yeah. Letter and everything. 

Adrian Green: But just like 1800, they had the sidekick. They had the sidekick. 

Marlon Avery: Yeah. So it's a Dave's, or it'll be a text message to you sending somebody, you know, but these are either typed up or written in cursor. And so, yeah. So, you know, how do you take those type of artifacts and digitize them in a way, but also to prepare and to be embedded into a LLM to have a conversational, you know, communication, you know, like, agent and stuff with it that. That can talk to and where visitors can have a, you know, unique experience with it. So I broke that down, you know, today of, like, how we built that in partnership with the Scott Museum. But along with that, you know, I just had some interesting conversations, some things I probably would never, you know, thought of. And I met an individual by name, but. Okay. And he said, you know, he is right now trying to figure out how to translate and how to store, you know, and build a model around African languages. And he said, you know, right now there's just like, you know, Google doesn't have access or they haven't built these, you know, these models or these languages. He said, you know, he can't. He. He can find them, you know, wildly across, you know, multiple different, you know, sectors. And so the whole process kind of got me thinking, and I just really thought it was unique, you know, as well. And so also figured that I'll bring you to the front stage here, you know, as a topic of, like, you know, what do you guys think? Like, you know, are we. I'm not gonna say we looking over, but, like, what are some other ways that you guys are seeing, like, how AI can be a pivotal role into. Into history? You know, kind of creating a different, you know, or unique experience, stuff moving forward. 

Sekou Doumbouya: Yeah. Oh, man, this is such an interesting topic. I think. I think one of the things we've actually. We've talked about before is about how this AI movement is like. It is actually. It's really based off of English. English Language Transformers is based off of compiling English language. That is the basis to do any conversion. Right. That's happening now. So being able to do training, being able to do things with other languages to make. So you can have the same capability. I feel like that's a. That's an important step. Right. Because otherwise we are just. We're just, you know, centering everything more upon, like, you know, a Western approach or Western view of the world. You know, a lot of the languages in. In Africa for. For a large portion of time were not. They're not written down. So we have a very rich oral tradition, but a lot of. A lot of languages that never had characters associated to them. 

Marlon Avery: Interesting. 

Sekou Doumbouya: So, you know, years, I think, was it years and years ago, like, my. Actually my father worked on specifically trying to get a language which served as, like. Served as a way to express the different dialects and tones that come from that, that. That the different languages make, specifically the Mandinkin languages, what they make into an actual set of characters that could be entered into Unicode. Because if you get things into Unicode now, you have a place for translation. You can make a keyboard off of that. 

Marlon Avery: Can you explain Unicode? 

Sekou Doumbouya: Unicode. Unicode standard is essentially of binary or binary conversions to a set of characters. So each one of these characters represent, you know, represent a placement on a keyboard or, you know, essentially when you look at a set of characters that are out there, generally Unicode is what was actually being used to create that and also helps with language conversions and things like that. Because an A on your keyboard could actually match another character that's on another keyboard, essentially. Yep. 

Adrian Green: And I remember this back in the day, because they were. You had to hold down the ALT key and then press the button combination, and then he lifted up and it's like, oh, there's my accent. O with the accent mark on there. 

Sekou Doumbouya: Exactly, exactly. That's a special Unicode code that you're. That you're kind of entering in to be able to access those keys. So imagine Transformers, which are based off of UTF8, right? Characters, which are. They start out as the English, like an English character, but they have conversions and things like that. But to actually get the nuances in natural language processing requires you to actually look at lots of languages and go through the process of seeing how a sentence form and those. And those different languages. Because, you know, just because you say, I go to the store, and our language is not this exact you're like ordering for another language when they say the exact same thing, right? 

Adrian Green: Yeah. 

Sekou Doumbouya: So for the most part, you know, all these AI things that were all the AI tools and that we're looking at, they are definitely 100. Like English is becoming even more a dominant language this process. 

Adrian Green: The only other one that I can think of is Mistral. Mistral out in France right now. They got the what. English, French and German are what they're specializing in now, but again with dialects and stuff. This article that I'm reading now from TechCrunch talks about how, you know, using their. They want to get into the content moderation game. All right, so content moderation across different languages, everything like that. As they're doing that, they're finding, you know, bias towards some African American vernacular English AAVE being labeled as more aggressive than it really is by definition. So there's so many with language, with language, it's so many, so many different nuances. But, you know, something like that is definitely to be aware of. 

Marlon Avery: Let me ask you this with, with the character being such an important role in language. I got a friend of mine who he, He's Cuban and his first language was Spanish. You know, he speaks, he speaks fluent Spanish. He speaks fluent English and everything. But he's also told me that he cannot read Spanish. Like he can't, you know, he can't read Spanish. He can only speak it. Which is interesting too as well, because now we're in a world where voice models are so powerful. Do you think that we can simply live in a world currently now where a. You can talk to a voice model and have it translate in back into a different language without the character base level stuff of it? Without. Or do you still need. You. 

Sekou Doumbouya: You still need the processing, right? So was it. There's a, like, it would be good example, right. And some languages, if you're talking to your significant other, you say, you say I love you. Right? One. One example. Right. And another language it would be you are love. They mean the same thing, but different ordering. But you have to actually done training on those nuances to do a good conversion. That's why a lot of translators actually don't work that well because they miss the actual nuances that are out there. There's no, there's no real perfect translator out there. That's. That really exists. That's why there's whole local localization teams that are every company. They're responsible for making sure that you actually get accurate translations as you're going from, you know, user to User. That's out there. 

Adrian Green: Yeah. Pacquiao wasn't saying any of that stuff. That translator was just playing around the whole time. 

Sekou Doumbouya: Making up your own sign language. 

Adrian Green: Yeah. 

Marlon Avery: Didn't some translator get in trouble for that with it? Like, oh, she said she could translate it and she was just, like, making up things and somebody. 

Sekou Doumbouya: She was just doing something and it was not the actual thing. It was not. It, like, I think it was like, for council or something. Or something. 

Marlon Avery: Talk about human hallucinations. Just making up stuff and stuff over there. Exactly. 

Sekou Doumbouya: My question is the audacity. We should use this word. It's like one of our words to get up, put your shirt on in the morning, drive your car to this event, and it's like. And you are going to fake sign. I bet what happened is they're like. 

Adrian Green: Fully committing to it. Fully committing to it, too. 

Marlon Avery: Yeah. 

Sekou Doumbouya: They're like, does anyone know how to sign? Oh, I think. I think. I think Janice knows how to sign. Right. 

Adrian Green: She's like, I got this. 

Sekou Doumbouya: I got this. It's easy. I see it on TV all the time. 

Marlon Avery: And also to the. To add to the audacity of it is to do this underneath the umbrella of the government. 

Adrian Green: Yeah, yeah. 

Sekou Doumbouya: It wouldn't be police officers. 

Adrian Green: That's the main. Yeah, Right there. Right there. You don't think anyone's gonna see this later? You don't think anyone's gonna bring this up? No. 

Marlon Avery: Some. Somebody was recording or recorded like, 15 seconds of her. Sent it to a friend of theirs that they know they can speak sign and got a text message back and say, that person is absolutely saying nothing. It's just. So, yeah. Just like you said, have the audacity to do that. The boldness to do that. Yeah, that's hilarious. 

Adrian Green: Yeah. 

Marlon Avery: Here. I know, Seiko, I know you had one too, as well. I'm gonna throw out a bonus topic as well. We're currently. Right now, we're streaming live and we're streaming live to, you know, multiple platforms. You know, YouTube, Twitch, and so on, so on. I. I'm also a believer and agreer that the simply streaming and the time that, you know, some of your favorite and top streamers will eventually replace tv. And, you know, you will definitely, you will have your live streamings from your favorite, you know, streamers, artists, you know, even TV shows. And then two, you still have your streaming services, which are movies, things like that. Yeah. As well. And so eventually, you know, we will have, you know, hundreds and, you know, here's, you know, thousands of people who understand that basically every Wednesday night a hour Friends comes on and it's like almost like a guide how we used to watch, you know, Fresh France Martin because we knew what time was coming on. You know, you wish you rushing home to, you know, watch the Cosby show, like, whatever it may be. And so I think a lot of is going to come back, you know, in the streaming, you know, streaming. So that the question is, what role do you think AI will play in this streaming, you know, universe that eventually, you know, we'll live in. 

Sekou Doumbouya: Oh, we won't be there. Simply we won't be. There'll be. Our AI, Our AI avatar will be in there. 

Adrian Green: Exactly. 

Sekou Doumbouya: Taking on the topics that we feed it. 

Adrian Green: Yeah. Crushing it, you know, crushing it. 

Sekou Doumbouya: You know. 

Adrian Green: I'm looking at the notebook, that Notebook ll thing. I'm like, crush that. You crush that. I had to redo. I was like, you know, redo the. Redo this podcast. Just focus, focus more on this specific thing. 

Sekou Doumbouya: Oh yeah, they just added that in. Yeah, you can like tell it to, to like, oh, I want to hear more about this subject and the podcast. You do? Yeah. So goodbye. I'm looking forward to my fleet of AI based podcast shows I'm gonna send out there. 

Adrian Green: Yeah, yeah. 

Marlon Avery: I said this guy, I said a couple months ago at this conference, I said we're gonna live in a world where it's gonna be a fully AI generated podcast that's gonna be talking about your favorite sports teams, you know, talk about your favorite players, you know, as well, and then just like the every day of what's happening to them. And you'll just subscribe to, you know, Lakers AI podcast with like getting your everyday updates from players getting traded to, you know, front office things to even like the finances of what's going on with the Lakers. You know, this new merch is available, you know, all these things. And so I definitely believe, you know, we'll definitely get there. So for sure. 

Adrian Green: Yep. Yep. And who knows? I mean, no one would have known. Who would have known that the interface to end all would be a chat window. You know, you can get so much done in the chat window. You know, it's really, you know, it's, it's impressive when I think about it. You know, I would not have thought that it would have been that that interface would have been the one that's revolutionary. 

Sekou Doumbouya: Yeah, yeah, I thought it'd be the one. The one that wins is a chat window. But you know what it's like, Isn't that like the most basic app that A lot of us started on. 

Marlon Avery: Yeah. I was about to say that our original experiences were. 

Sekou Doumbouya: What was it? Depending on the age. 

Marlon Avery: Hello. 

Sekou Doumbouya: My first was hello world. It's a. Was it. I'm trying to think of the chat systems. Icq, that was one of the ICQ. 

Adrian Green: I had aol aim, aol Instagram. 

Sekou Doumbouya: Yep. 

Adrian Green: My first one. 

Sekou Doumbouya: Yeah, it's a messenger. That was, that was a big thing. That's a lot of, a lot of folks that was like their first true interaction with technology. Right. Look at us here, we're starting all over again. 

Adrian Green: Is it like that though? It is, it is like that. Wow. Now that I think about it, yeah. Millennials. Boy, I tell you. Look at us go. 

Marlon Avery: Seiko. I know you had a bonus topic there too, as well. What you got? 

Sekou Doumbouya: Oh, where's my bonus topic? You gotta give me a hint. 

Marlon Avery: Xbox. 

Sekou Doumbouya: Oh yeah, yeah, yeah. So this is a, this is a new one. This actually, this is, this is actually like just breaking that Xbox. AI is now becoming a thing. Microsoft is now pushing out updates to Xbox to integrate AI into them. That is once you, once you pull up the article. 

Marlon Avery: Can you, can you elaborate and stuff here? Like what is, what is it going to do? 

Sekou Doumbouya: Well, imagine. Yeah, imagine having a nice little. Oh, so imagine having a, a chat box in on your Xbox where you can talk to copilot and exit different questions about things. Well now the question is, you know, we've gone from marketing things to like business folks, adults, things like that you're putting on the Xbox. These are the same people who will cuss at you playing any, any of the video games online. Right? Say, they say the most terrible thing and you are gonna put this in their hands. Right. Like I feel like this, this is something that I think you do need to like have some type of age controls on some of these things. Yeah, right. Just to like, I don't know, just. I feel like there's nothing good gonna come out of giving teenagers access to an AI powered natural language chat box that they can use for, to conversate with. Like, what type of terrible things do you think are going to happen in these, these rooms? Yeah, right. 

Adrian Green: Well, luckily, well, hopefully they keep it to the like very minimum. Like a very minimum offering at first. Like because looking at the release notes, it looks like it's suited towards the search first. So like it gives an example. Cozy farming games with fishing or FPS games with zombies. So you know, maybe it'll be less out of the, out of the. What's it called? The lobby. It won't be in the lobby. You know, it'll be on the search side. Because I think that what this, what this could do actually is give people, you know, a way to, you know, just like a really easy entry point for some non gamers for them to be very specific about what they want. So like, you know, you got a non gamer person who wants to get up there and you know, cozy farming game with fishing does sound like someone who doesn't game that heard about some other game. And you know, it was this one cozy farm game with fishing. You know, it'll just do a better job at finding that. 

Marlon Avery: So are we, are we talking about the ability to tell your Xbox AI to update my shooting settings in Call of Duty so I can be more accurate? 

Sekou Doumbouya: Not that aimbot, the AIM bot. Oh no, yeah, not quite. Not, not yet. Not yet. 

Adrian Green: My foes. 

Sekou Doumbouya: I feel like this is, this is, this is how really the Skynet storyline goes, right? It gets on Xbox, then it takes over the US government because they over using the AI models. Someone's gonna take me, someone's gonna take. 

Adrian Green: Me serious here, you know, and let's also add in that, you know, this is Microsoft. So they're gonna have Three Mile island back in this maybe, you know why? What they're gonna have the juice to, you know, beefed up this natural language search approach. You know, they're gonna have the juice for it. So. 

Marlon Avery: Maybe it's just my dislike towards Xbox. I will, I will give some grace here and say Microsoft has done, has done some really great things, you know, in the past decade or so, you know, and only I could think about Microsoft at this time. It's just like, remember three rings of death? I can see that coming back, coming back at the evil prompt that doesn't work. The red ring of death, you know. 

Adrian Green: Oh my God, if I got my screen to death. Yeah, Blue screen to death. Yeah, yeah. 

Sekou Doumbouya: You know what I am excited about, Elise, when it comes to the gaming world and AI. AI, was it AI power, npcs. I think that's going to be something interesting, which is also think about it, what's the difference in AI powered NPC and a chat bot, right? It's just a different interface. Right. So but just like I'm imagining, you know, I used to play the game, it's called Skyrim, right. Going and walk to a person and you get your little quest and things like that. They always say the same thing right now these guys are actually like having interesting dialogues. They have something to say to you that is unique I feel like that is a level of immersion that, that's, that, see, that's the AI product. Someone's gonna pay for. They're gonna pay for that one. 

Marlon Avery: That what you said could possibly be the hold up of GT6. 

Sekou Doumbouya: Oh, yeah. I bet you GT6 is gonna have AI. They're gonna have to. They could, yeah. 

Marlon Avery: Integrated, integrating, you know, LLMs to NPCs that have like just an additional conversations and it changes every single time. You know, it's all, it's all based off, you know, it all is based off the IDs of materials and things enclosed, stuff that you have on the car, stuff that you're driving on, you know, and so it creates different responses, you know, so right there, I mean, that could be the hold up GT6. 

Adrian Green: And I'm gonna also throw another one in the mix here, which was this, this guy will write. And I think he did. Did he do freaking God of War. But anyway, he released this game called spore back in 2008. And it was. You start off as a single celled organism and you slowly evolve into like an interplanetary creature by, you know, going through, like, evolutions. You get two legs, you kind of start as like a little thing, you know, you. You literally evolve into a whole. A whole thing, you know, and this would be perfect for the NPC and the, you know, if you were to put AI to this. This was made back in 08, I think if you were to put AI to that and you have maybe an approach that's generative along the way, that's adapting to however it is that you're, you know, what you're, you know, how you're behaving in this environment. I don't know what happened to that game. It kind of came and it kind of came and went, but I just saw a Reddit thread that said it's still playable, so. 

Sekou Doumbouya: Really? 

Adrian Green: Still? Yeah, still worth it. 

Sekou Doumbouya: Wow. 

Marlon Avery: Let me, let me ask about this. This is definitely an ego question. How does it feel to just throw out some of these use cases? And even at the, even if it's at a basic skeleton level, you have an idea or you know, how to build it. Even, even, even the. I could, I, in my head, I could definitely figure out and know how to build an LLM in a GT6 environment for the NPC that all focuses back to the character, the mission, the city, you know, what's going on and stuff around them. Like, I, I for sure have a skeleton aspect of how to, like, create that. 

Sekou Doumbouya: Yeah. 

Adrian Green: And it's just like with any game it's going to be in layers. Like it's going to need. There's so like I was just talking to my son about this today. It's like game development is so fascinating to me because you have to be a specialist in so many different areas. I mean, or a game company has to be a specialist in so many different areas, they got to be a specialist in real time communication and being number one. If you're talking about a multiplayer online game, you know, rendering, optimization, so many, like it's, it's so many, you know, huge details that are involved in that and you know, but yeah, what, what you just laid out right there, like it does feel good. It feels good because it is, you know, kind of when I saw the magic of technology as a little kid, I wanted to be able to be involved. Like, and I think that some people feel like that about music. You know, they want to pick up the guitar too. And me, I wanted to, you know, I wanted to get behind the keyboard. Just to be honest with you, with. 

Marlon Avery: That being said, with it being said too. Rocks are games. If you want to figure this out, we are, we, we can build it for you. You know, you know, us three right here, you know, we can build it for you. We will gladly take 2025 off to make sure that this gets shipped at the end, you know, Q4, 2025, you know, we can, we can build this for you. Highlighters. 

Adrian Green: No biggie. 

Sekou Doumbouya: All right. All right. 

Marlon Avery: Well guys, you know man, we've had another great discussion and stuff here, man. We've discussed here Anthropic, you know, in Palantir. I'm sorry, Anthropic partnering with Palantir aws, you know, for on the government defense side along with Meta also too, you know, making their. Some of their models available for government entity as well. And then we talked about some of the leaks like things like Project Jarvis from Google which can control your computer to do certain actions and stuff as well, along with how it lines up with the leak of open the eyes A1 model. And so it looked like this kind of become a marketing trend, if you will. We talked about Universal Music Group and them their unique partnership with Clay AI in the music world. And so you said the whole focus there is for this new partnership to not compete with the artist but compliment the artists. And so I guess in coming years we'll figure out what that means. And then also to complexity offers a interesting, makes an interesting offer to New York Times and says like, hey, you know, as your staff is over there striking, you know, our models, smart servers can help you, you know, in some of the. In these weeds as well. And then finally, we also talked about just kind of the role of what we're seeing with historical artifacts along with folklore. And then we added some bonus topics in there for fun too, as well. And so, man, this is. This has been great. Man, this has been great, guys. How you feel about it? 

Sekou Doumbouya: Great. Good, good. Awesome episode. 

Adrian Green: Great episode. 

Marlon Avery: Well, guys, definitely, if you want to follow. Want to follow me and stuff, there I am. I am Marlon Avery in all platforms along with the podcast, which is I Am AI With Friends podcast on all platforms as well. Adrian, thank you. 

Adrian Green: Yeah, good. Hey. 

Sekou Doumbouya: Yep, you can follow me. It's cool. The Wise on Twitch and Tick Tock. 

Adrian Green: Adrian and I, I'm gonna get my, you know, more of a unified name, but I'm Green Lantern on Tick Tock and I'm Infamous Adrian on Twitter. That's where you can find me on X. Sorry. Yeah, I know, right? I'm gonna get it together, huh? What? 

Marlon Avery: First of all, there's no way for you to get Green Lantern on all platforms. 

Adrian Green: Oh, Green underscore Lantern. Sorry. And it's with, like an R in. It's missing an E there at the end. So, yeah, it's kind of. Kind of thrown together. I didn't. You know, honestly, Tik Tok really stuck around, didn't it? Like, I didn't think it was gonna really be anything, you know? 

Marlon Avery: You know, they're making dances now. They're making dances. What was it called? What was it called? Was. It was. It wasn't it musically, before they got acquired by Bite Dance? It was musically and then change to Tik Tok. To Tik Tok. Well, it looks like that that band is still kind of like, being extended. So we'll see here. Well, guys, man, this has been a great episode, man. So until next time, tune in next week. We'll be available ends of every week and everything. Wednesday, 9pm Live stream, and then also too, you can listen to this episode on all your favorite podcast platforms. Until next time, we are out. 

Adrian Green: Peace. 

Sekou Doumbouya: Peace.