Loading summary
Transcript107 lines
- [00:00]
Kevin Ferreira
Everyone hates AI. And Anthropic's Mythos model certainly isn't helping. And it's not even out yet.
- [00:06]
Gavin Purcell
This comes as anti AI sentiment hits an all time high. Data center protests are rising and Florida is suing ChatGPT and OpenAI for the dangers of AI.
- [00:19]
Kevin Ferreira
Is your teen son vaping AI juice?
- [00:23]
James Uthmeyer
To safeguard our children from the dangers of AI and to further empower my office of Attorney General to fight these
- [00:30]
Kevin Ferreira
evils, it's time for a super serial conversation about whether AI is the problem or we are the problem.
- [00:37]
Gavin Purcell
Are we the baddies? Kevin.
- [00:39]
Kevin Ferreira
Oh, you sweet summer child.
- [00:41]
Gavin Purcell
Bernie Sanders has called for a pause. Sam Altman has said that he wants to raise taxes on AI and even Demis Hasabis said he would rather have cured cancer than compete with ChatGPT. But, oh well, I want it. You should.
- [00:54]
Kevin Ferreira
You should have done that thing. Demis, by the way, Meta, remember Meta, Gavin, remember that old company? Oh, yeah, sure. They surprised everyone with a new AI model called Muse Spark that is actually pretty pretty, pretty, pretty, pretty, pretty good.
- [01:10]
Gavin Purcell
Plus, Claude keeps shipping new features that turn it into openclaw Surprise.
- [01:16]
Kevin Ferreira
All that and more on this AI for humans.
- [01:25]
Gavin Purcell
Welcome, welcome, welcome everybody to the twice a week guide to AI. It is known as AI for Humans. I am Gavin Purcell, that's Kevin Ferreira. And today, Kevin, welcome. We are tracking something. We're going to talk about something we've been tracking for a while. And the Claude Mythos preview has kind of brought this to the forefront. Kevin, people hate AI. They really hate AI And I see this and hear it all the time. I know you feel this all the time. You and I have been talking about AI for a long time now, three plus years on this podcast. I think almost. We're almost close to the three year anniversary.
- [01:55]
Kevin Ferreira
It's my third pillar at the Thanksgiving table. Gavin. It's religion, politics, and artificial intelligence.
- [02:01]
Gavin Purcell
It really feels that way right now. And I think we have a couple things to dive into. We've been meaning to do a show kind of like this for a while and the Mythos preview gives us a kind of jumping off point. But there's a bunch of other stories that are important here. But first, we should very quickly touch on the base that Mythos preview, if you missed our kind of whole episode on it, we released that a couple of days ago. You can find it on our channel. This has freaked a lot of people out and we talked about the cybersecurity dangers, but I think more so there's this like kind of. I've seen more videos and I'm not sure I'm not going to do one of these. Just so everybody's out there. If you're worried about me or Kevin doing one of these videos where it's like, it's a person who's like asleep and they're just woken up. Like, I slept for two hours because Cloud Mythos is freaking me out so much.
- [02:39]
Kevin Ferreira
I thought this was the foot in the lasagna thing again, but.
- [02:41]
Gavin Purcell
Ok, I'm glad we're past that. Yeah, it's not bad. But anyway, there's a lot of issues with Cloud Mythos that, that are being raised. I think we, as we talked about in the last week's show, one of the interesting issues that people aren't talking about enough is this separation between who gets what with AI, Right. Who gets the powerful models, who doesn't get the powerful models. And this will kind of inform some of what we're going to talk about today because I think we have to start with a pretty stark video that came out from the Attorney General of Florida. Now, Florida is a state. It is a state and it has, it has a particular right wing backing. And I think that everybody kind of understands the governor of Florida and kind of what happens in Florida. But this AG is going after OpenAI and Chachi PT directly because there have been a number of stories that have connected ChatGPT and OpenAI to bad things happening in the world, which I think is totally fair. Kev, maybe we should hear a little bit about what this Attorney General is saying and then jump into discussion about this.
- [03:41]
Kevin Ferreira
Yeah. And for those who hate clips, I'll just say it's that ChatGPT does not have enough barbed wire tattoos and hasn't been slamming enough Monster Energy and Faygo
- [03:50]
Gavin Purcell
or living in Key west with their top off. We've just insulted all of Florida, Kevin. So I'm sorry, Florida viewers, let's listen to that clip.
- [03:58]
Kevin Ferreira
But, oh, let me be clear. Mine weren't insults. That's, that's the reason I'm planning to move. I'm going to redomesticate.
- [04:04]
Gavin Purcell
Fair enough. All right, here we go.
- [04:05]
Kevin Ferreira
Here's Attorney General James Uthmeyer.
- [04:09]
James Uthmeyer
I'm Attorney General James Uthmeyer.
- [04:12]
Gavin Purcell
I was so close. You almost had.
- [04:14]
Kevin Ferreira
Oh, did I show my cards? I didn't watch the clip before the show. Anyway, what do you, what do you. What are you saying, James?
- [04:20]
James Uthmeyer
Today we are launching an investigation into OpenAI. The development and rollout of artificial intelligence is a monumental leap in technology, but it has not been without concern for public safety and national security. AI is built on its ability to gather data. And there are concerns about whether OpenAI's data and AI technologies that could be used against America are falling into the hands of America's enemies.
- [04:47]
Kevin Ferreira
How many Rs are in Iran? Max reasoning.
- [04:52]
Gavin Purcell
We can kind of jump forward towards the last like 30 seconds of this clip. You should go watch this. It's about a three minute clip, but you should watch the last. Let's listen to the last 30 seconds here.
- [04:59]
James Uthmeyer
Today we formally launch an investigation into OpenAI and subpoenas are forthcoming. And I call on the Florida legislature to work quickly on implementing protections to safeguard our children from the dangers of AI and to further empower my office of Attorney General to fight these evils.
- [05:18]
Kevin Ferreira
Okay, so at least it's not about Roblox once, Gavin for one.
- [05:22]
Gavin Purcell
So I want to get into what this means and talk about a lot of things around this. Right. So we're going to talk more about why you might hate AI. Probably you don't if you're watching us, but why people around you might. But the important thing to understand here, Kevin, is there's two things happening. One, obviously there's a lot of politics in the world and America is a very divided country in general. So Florida is a very right wing state and we don't know what sort of motivations, probably quite a few motivations are going on behind that. But the bigger thing here is this idea of the large tech companies starting to be seen as villains in a significant way. Right. And I'm not trying to say that like they don't have responsibilities to protect children, to protect people, to protect everybody. But there's also something to be said for personal responsibility and all this sort of stuff that like people allow themselves to do. I think what we're going to see, and this is the beginning stages of this, the other part we can talk about is like states versus national. This is the beginning stages of politicians starting to move to this issue because it moves the voter. And I think that's the other thing that's happening here is they're hearing a lot of bubble up from down below in their voter bases saying that they are unhappy with some of these big tech companies. So that's a lot to chew on. But I'm curious to hear your thoughts on all this.
- [06:37]
Kevin Ferreira
Yeah, look, this is not a surprise. Large tech companies have for decades now had major issues specifically with protecting or the lack thereof when it comes to protections for minors. Like this is nothing new. There were massive hearings recently meta YouTube, they were all swept up into it. They knew that they had systems that were predatory towards certain users. And even when their own internal studies said that they had deleterious effects on younger users, they pushed even further into it and they shoved those studies aside. So the is. Is AI the new great boogeyman for all of those things? Yeah, of course it is. Because, you know, people are being laid off and being told it's AI. Their environment is being destroyed. We'll get to that. And they're being told it's AI and in some cases it is. And in some cases, you know, AI guardrails weren't solid enough. But in other cases, this was the. The user's behavior. The AI was just mirroring and in some cases, potentially amplifying or what the user wanted from the system. So I also think, you know, surprise, surprise, parents play a role in all of this just as much as. As the tech companies do. But you set the table here. But, like, what do you think, Gavin? Is this like a particular AI issue?
- [07:54]
Gavin Purcell
I think this is an AI issue right now, to your point, there was a huge amount of stuff that just happened. Meta. And YouTube lost a big lawsuit around delivery of material to underage children, specifically trying to get them addicted, quote, unquote, to social media. Right. So you have this kind of negativ around tech already. The AI thing, though, I think is bigger and weirder than this. And the reason I say that is I think what AI has become is a transfer for all of the ills of the way our world currently works. And I'll mean that by I think AI represents capitalism to a lot of people who are on, maybe on the far left who would be like, I think that people shouldn't make X number of dollars. So a lot of people are there on the far right. AI represents wokeism. Because of course, we've been. You and I have been through with three years of this show, and we went through a lot of places where AI is going to give the wrong answer because it believes a certain way. So I think AI has become a thing that everybody puts their ills on. And ultimately you brought this up on the other show, which is why we want to do this show, is like, it's not us AI. Yes, there is a danger of AI and we're going to talk about what Demis Hasabis has to say about that a little bit, that ultimately AI could go rogue. And a lot of the mythos people are worried about that. But guess what? The mythos issue really is it's really about an actor or two. Maybe it's a foreign entity, maybe it's just a crazy person who has the ability to say to the AI, go do this thing that it could then do. The AI is not right now at the stage where it is going to take the initiative to be the bad person. So I do think the important thing here to understand is a lot of this world that's happening right now has to do with the fact that it is a reflection of what normal people are seeing in their world. And I think that goes from the job loss thing, totally fair, all the way to the political thing, all the way to what now? How people are freaking out about cloud mythos.
- [09:50]
Kevin Ferreira
100%. 100%. I love always agreeing with you on a podcast. I think it's great, right?
- [09:55]
Gavin Purcell
Exactly.
- [09:56]
Kevin Ferreira
When they hear two people shout at each other and go, no, you're right, even louder. But it's true. I also want to point out over regulating some mainstream AI is not going to eliminate the risk people will run screaming and flocking to open source models, to local models, to completely unguard railed things. So let's not pretend that waving some sort of magic wand at the big six or whatever is going to eliminate these problems. Again, there's still a lot of personal responsibility to be had. The other major concern is that like it's the who watches the watchman, right? When these models were springing up, we were saying all along like a political party or a private corporation says I don't like the output of your LLM, I don't care what side you're on. Anybody having control over how the machine thinks and reasons and speaks can do it for good, right? Stopping harm to children, right? Stopping negative thoughts, thoughts of self harm, making weaponry, et cetera. But they can also use it for their own gain to, to wash the message. And we saw this with like Elon and Xai, not to single them out, but we saw it publicly as it was telling users that it was supposed to have a thumb on its scale and give certain answers. It did that publicly. We know OpenAI did that. We know that Google did that with their image models, right? To please certain audiences. So again, yes, there is a level of responsibility. But whenever we allow this regulation to creep in without transparency, suddenly I can't trust the answers that I'm getting from any given LLM. And that's dangerous.
- [11:27]
Gavin Purcell
And this thing is the thing people are talking about as a super intelligent being that will ultimately rule us, right? I mean, maybe we've been Talking about that. Maybe the world at large isn't talking about that. But I do think, Kevin, the other thing that's really important to get into here is data centers, right?
- [11:41]
Kevin Ferreira
Yeah.
- [11:41]
Gavin Purcell
Data centers are being built everywhere. They are being touted as a large job gathering thing for, especially for small town communities. And what we are seeing in a fascinating way, and I don't mean this in a fascinating and just a societal thing, is there are massive protests that are going out and people are stopping data centers from happening. And when you think about what that represents, it's this idea that, like, you know, I mentioned before here, my, my grandmother, whose house I'm in right now, was very involved in the anti nuke movement in the 80s. And I really appreciated the fact that I think she really cared deeply about the idea that those could destroy the entire world. Right. I think people are starting to get their heads around an idea that like these data centers that could destroy the world in a different way. And I think that's where we're entering is this interesting space right now where AI, again, the boogeyman version of AI that nobody knows yet exists other than like Ilyar Yakowski, the guy who wrote the book, if anyone builds it, everyone dies. Like that's a thing. But politically, again, local, local, local data centers are local. So. And they are driven by jobs because a lot of people want to bring them, because they bring jobs to small communities that have been without jobs for a while. But they are also driven by very noisy people who can turn elections against local politicians. So it's this weird level of you've got this high level fight going on at these kind of giant companies and you've got this like lower level fight going on in local communities. It just feels like it's going to be the conversation of the next, like we said, for the next two to five years, it will just be the number one thing. Like I would not be surprised to see it in the presidential election of, what is it, 2028 for sure.
- [13:19]
Kevin Ferreira
Oh, that's cute that you think we're going to have one of those again.
- [13:23]
Gavin Purcell
But even in these midterms, right, the midterms are whatever, five months away, I bet this is going to be a major issue. In fact, so much so that Bernie Sanders did a video with Hank Green recently where he's calling for an AI pause. Now Bernie is on the far left and we, again, this is not meant to be a political message, but people on the far left and far right are both kind of coming for AI and, and you know the argument against a pause they bring up in this video, which I think everybody should watch, because it's an interesting conversation. They bring up in that video that, remember, whatever it was two and a half years ago when a thousand people signed on to, like, pause AI and nobody.
- [13:54]
Kevin Ferreira
Elon was like, hey, hey, hey, everybody, slow down. I'm late to the race. I haven't placed up my sneakers. Y' all need to pause.
- [14:00]
Gavin Purcell
Yeah. And then he kept going on with Xai. But there was a kind of a fight against this, and nobody paused. Right. And the argument always was China versus America, and that still exists. Right. But there's this level of feeling like we can't control anything, and I think that drives people bonkers in some form.
- [14:17]
Kevin Ferreira
Well, listen, we. We at the very top of the show mentioned that Mythos isn't helping all this. And again, earlier. Go look at our episode from earlier this week. Mythos as a cybersecurity tool. I'll get back to data centers, I promise. But Mythos is an insanely disruptive cybersecurity tool, and make no mistake, like the private government cloud, they've been using it just as the developers have, and they've probably been hacking every country they can.
- [14:41]
Gavin Purcell
Right? Yeah.
- [14:41]
Kevin Ferreira
And this is the. Well, if we pause, then someone else is going to get there sooner. That's kind of what we're talking about.
- [14:47]
Gavin Purcell
Ooh, I just thought of something, Kevin. This put. Actually, this puts the anthropic conversation in a much more interesting light. And I had never thought about that until you just said this right now, that if the government, the US Government knew about Mythos, which, by the way, they probably did because this only happened the kind of Department of Defense thing or the permanent war thing happened, like, a month and a half ago.
- [15:11]
Kevin Ferreira
Yeah.
- [15:11]
Gavin Purcell
Maybe that's what happened. And we're just. Now, I know everybody in our audience is probably like, yeah, you think? So you dump me. But, like, I'm just. Right now, that's a really interesting thing. Right. Because then it is a slightly different conversation. Right. It is this idea where if you've got Anthropic out there selling this idea that this very dangerous tool exists, and you're telling me I can't use it as government, whether whatever side you are on. If I was in that role, I could feel that way, too. Right. I would be like, you're not going to let me protect the country? Like, so that does put it in a really interesting context, you know, and
- [15:45]
Kevin Ferreira
back to the data center of it all. Like, that's part of this arms race. And make no mistake, like, that's the level. Now we'll get to the. The notion that I think everybody will be singing its praises. If we get to the point where it does cure cancers and help us with genetic sequencing and unlock new physics and get us to the promised utopia that is clean, renewable, infinite energy. If we actually get there, then great, a lot of these transgressions go away. But right now we're at a phase that I think is really interesting. I was having a conversation with, with April, my wife, the other day, and she was like, longing for this time. Gavin. And, and I think you're of an age where you can remember this in the, in the mid to late 80s, sure. It was believed that Aquanet chloroformal chlorofluorocarbon CFCs.
- [16:30]
Gavin Purcell
CFCs, yeah, yeah.
- [16:32]
Kevin Ferreira
They were punching a hole in the ozone and we somehow managed to collectively get together and ban them from store shelves.
- [16:38]
Gavin Purcell
I thought you were going to say it was believed and you were going to do, start doing quotes and like, say that was like the moon landing or something.
- [16:44]
Kevin Ferreira
They just didn't want you to have it. It turns out it gives you superpowers. Keep going.
- [16:47]
Gavin Purcell
I'm getting at. I hear what you're getting.
- [16:49]
Kevin Ferreira
We came together and banned it. And if we had that same issue today, yes, as I said, Kid Rock would be outside shooting cans of Aquanet out of the sky like they were clay pigeons. Right. And writing songs about, ah, people would be spraying their Aqua net out there. F150s screw, I'll show you your mile zone. Like that sort of unity doesn't exist. And it's. It's in a weird way, like, heartwarming to see communities come together and say, no, no, no, not in my backyard. Giant data center, which could be releasing harmful chemicals into the sky, which could be polluting the environment, which could be, you know, blowing up our energy bills. I don't want to, like, great that we're arriving to some sort of unity, but where was this energy? With factory farming and with fracking and with a million. With. With coal plants being fired back up. Like, I love it. Let's not stop at AI if we're
- [17:45]
Gavin Purcell
going to do that for office in 2028.
- [17:50]
Kevin Ferreira
Take a large link, pick a large language model. Not me, honestly.
- [17:53]
Gavin Purcell
Okay, so there's. We've talked a lot about this stuff. There are a couple of things I think it's important to understand. There are people who are trying to change this narrative better or for worse. And some of those people, one of them is Sam Altman. Right now, we talked yesterday on last show about this idea about how there's this new kind of new deal that open. I kind of set out, and that probably is mostly driven by, you know, kind of seeing this stuff going forward. There was a clip of Sam. There was an internal conversation they had that was kind of broadcast over YouTube that I want to play where he's talking about taxes directly on AI Companies
- [18:24]
Sam Altman
do suspect that we're going to have to make changes to how we tax. Like, in a world where AI is doing most of the intellectual work in the world, or at least of the work of today, you know, we probably are going to need to explore some way to tax that instead of taxing human income in the traditional sense. I suspect that we will need to provide new kinds of transition assistance, unemployment insurance, things like that. And I suspect that eventually we will need to think about how people. People get to be an owner in the upside of all of this in new ways. You know, capitalism is dependent on a certain balance between labor and capital.
- [19:00]
Kevin Ferreira
Well, great, because there's never been a greater trust in that balance.
- [19:05]
Gavin Purcell
Well, that's what. What I was going to say here is, and I think this is the kind of heart of my belief is that the tech community has spent the last, like, say 10 years, maybe, maybe 15 years, kind of like abusing the ability to kind of go out there and collect as many users as they can, make as much money as they. And it's been hugely successful, has driven a huge amount of the economic success of our country for a while, and amazing entrepreneurs have come out of it. But what's hard now is to hear something like that from Sam, who is obviously extremely successful as an entrepreneur, whether you, whatever you think of him personally in that New York article, like, obviously paints him in a certain light. But you cannot start to talk about this in detail now. What you should have been doing is, like, a couple years ago, have a kind of a cohesive plan around this, and now it's going to be pushing against a rock that's rolling down the hill towards you. I really think more than anything, we have to get some sort of major breakthrough that happens for normal people that makes normal people feel this is a real thing. And granted we've both talked about this, there are many things on the horizon in science that are possible. And you know, I was just reading the Demis Hassabis book, which I really suggest everybody in our audience read. It was a. It's a new book called the Infinity Machine, about Demis Abbas and the rise of Google DeepMind. And what was so interesting about it in that is that Demis is obviously kind of a scientist first. It does talk about his ego and that he wants to, you know, kind of be the guy that kind of brings this to fruition. But like, you know, Alphafold was an incredible discovery that AI led to, which was the ability of looking at DNA and proteins and being able to map all these crazy things. We need like one of those to every Claude Mythos freakout. Right? Like you need, you need to start balancing those and start thinking about those in a much bigger way. Because if you don't, there's going to be this overwhelming tide that is going to continue to hit and continue to hit. And as I mentioned like my, my grandmother earlier, like you know, God bless her and she's passed away now, but like she was really believed deeply in her heart. And again, nuclear weapons are a terrible thing. But that 80s movement of anti nukes stopped us from kind of learning about and be better about nuclear energy, which might put us in a very different place than we are right now, where we are still very reliant on oil and all the stuff that's happening there. So anyway, it's an incredibly dense and detailed conversation to have, but it is important to have a lot of context to it and think about it in a non polemic way.
- [21:33]
Kevin Ferreira
So maybe you can help me and our listeners understand this. Like at one of those companies where I'll just pick a fictional company.
- [21:41]
Gavin Purcell
Right.
- [21:42]
Kevin Ferreira
At Zombo. Com where the CEO is making 600 times that.
- [21:47]
Gavin Purcell
Yes.
- [21:47]
Kevin Ferreira
Of the lowest paid employee.
- [21:49]
Gavin Purcell
Right.
- [21:49]
Kevin Ferreira
And when Zombo com announces that they're laying off 1 to 10,000 people because of AI, that's what they're saying. And they're a, an enterprise client of anthropic or OpenAI or Google, etc. What is stopping these companies from saying, hey, if you were one of those Zombocom employees again, making 600 times less than the highest paid person at a company that is like announcing record profits, what's stopping them from saying, here is the displacement fund, we're going to just start it ourselves, instead of saying someone should really do this, what's stopping them from doing that? Or from launching grocery stores in affected cities where they're subsidizing, I don't know, fruit and vegetables. Well, what is stopping them?
- [22:29]
Gavin Purcell
The capitalism is stopping them.
- [22:31]
Kevin Ferreira
And again, the very thing that's supposed to save us. Right?
- [22:36]
Gavin Purcell
And again, like none of this is new. I mean again, we Are we are dummies here. The two of us are not experts. We do not have master's degree. We have PhDs in any of these domains. But we do care about this world,
- [22:48]
Kevin Ferreira
especially from the friggin streets of the East Bay. Gavin. All right. I know what it's like to live off a boot soup. All right. And we deserve better by the way.
- [22:58]
Gavin Purcell
I do too. Like I'm the child of a divorced family and my single mom struggled. She was a secretary. Like so I understand and like that's the thing where like I see see it already bubbling up. So all of this is just an important thing for you to say. I think if you're listening to the show you're clearly interested in AI. I think one of the things you're going to have to feel and you probably already do to Kevin's point about the conversations around the dinner table is we have to get these companies to understand they need to figure out a better way to message this stuff. And part of that is what it does and it can't just be it's going to eliminate your jobs and it's so scary that it could take down the Internet. Yes.
- [23:34]
Kevin Ferreira
You know what helps just do good companies just literally like do good. If you're, if you're super intelligence can spot security flaws and fix them then fix them. Don't charge people for that. If it is so great at health care give everybody a free tier in their pocket that helps them, I don't know get around the terrible insurance system that we have in this country. There's a lot you can do but what everybody needs to do and I don't care what side you fall on gas.
- [23:57]
Gavin Purcell
This is an Alzheimer's not care.
- [24:00]
Kevin Ferreira
The one thing we all as human beings and maybe bots that are listening and transcribing need to rally around a non negotiable is liking and subscribing to this video.
- [24:11]
Gavin Purcell
That's right baby.
- [24:12]
Kevin Ferreira
It's so easy. Gavin, click the thumbs up, click the bell. Give us a five star review. Leave a comment in the algo juice that machine because we're twice a week baby.
- [24:22]
Gavin Purcell
That's right.
- [24:22]
Kevin Ferreira
Those numbers up. We're trying to have our own data center, isn't that right?
- [24:27]
Gavin Purcell
Oh yeah, we definitely are. And Patreon you can go to our Patreon and support us which is very nice if you do that and follow our newsletter.
- [24:33]
Kevin Ferreira
Thank you.
- [24:33]
Gavin Purcell
All that sort of stuff.
- [24:34]
Kevin Ferreira
Okay, thank you.
- [24:35]
Gavin Purcell
We are off our soapbox a little bit here. Hopefully we still have a few viewers who've stuck around through all that. We do need to talk about a new AI model that came out this week. Kevin, that is a surprise. I was really not expecting this after they said that the Avocado Llama model was kind of gone. Meta, this is where we go. We're going right back to Avocado Llama right from the most important conversation we've probably had.
- [24:58]
Kevin Ferreira
It's like, it's like job displacement and destroying the environment and they're fracking in the water and. Okay, but Avocado Llama on the new nano Banana Pro 2.0. Don't forget Quopus 5.4.
- [25:10]
Gavin Purcell
Okay. Anyway, avocado Llama gone new model is Muse Spark. Now. This is a surprise model from Meta. So the benchmark boys are pretty good on this. There are a lot of people have been saying like, this is kind of like, okay, it's not state of the art, but like when you compare it to the opus 5.5.6 and GPT 5.4s of the world, it is not a turd. So they have launched a model that is not a turd. I did. I will say I did very. I did one thing with it because you can access it at Meta AI. You can go to it. By the way, speaking of giant corporations, the login for Meta AI is a nightmare. If you have a Facebook account, if you have an Instagram account, if you have a wife, a wife who also has a business on Facebook, it is impossible to get into. But once you're there, I asked it to tell me about me and it did a pretty good job. It's a thinking model now, but I will say without me prompting it, and it didn't ask me any other questions, it included in my bio the fact that I am an Irish rock star and part of co founder of a band called Bi Curious, which is actually, there's another famous. This is a real band. Don't make fun. There's a real band and it's a famous Gavin Purcell in Ireland who does show.
- [26:17]
Kevin Ferreira
I have been saying this about you though, on Reddit since we started this podcast probably before.
- [26:23]
Gavin Purcell
Anyway, kind of just put those two people together and said, well, this is what I know about you. So it's not perfect, it's a hallucination. But again, people are saying really good things about this, especially when it comes to multimodal uses. So I will say there's a little bit of shade here. But Alexander Wang, the guy that Meta hired to bring their new AI stuff to fruition, spent some time talking about this model. He said, we're really excited for people to try musespark. And he retweeted a guy who's known as Tornado Guy and no shade against Tornado Guy, but he says, this is the most insane, magical model I've ever used. I provided it, some AI generated images and fed it into the chat and it told me, told me to generate a game. And if you look at what it did, I just. It's not the most exciting game in the world. It's basically. And I get. Listen, it's start one, but Crimson Veil.
- [27:10]
Kevin Ferreira
Are you. Are you shading Crimson Veil?
- [27:13]
Gavin Purcell
When you watch the two images kind of bang into each other and kind of shake, it does feel like, okay, sure, it made a game, which is great, but it's not like, I think if you put that exact prompt into Cloud Opus right now, it would make something significantly better. I can't guarantee that. But anyway, they're in the game again. That's the important thing here.
- [27:30]
Kevin Ferreira
Well, you know, and with the tens of billions or hundreds of billions of dollars that they spend, it's nice for them to be in the game. I will say whenever I am on X, which is more often than I, I want to be truly having Grok there in the app.
- [27:46]
Gavin Purcell
Yes.
- [27:46]
Kevin Ferreira
To be able to summarize, to ask questions again, like that is so convenient. Whenever I'm on threads, I feel like the experience is completely lacking because of that. Plug Spark into threads.
- [27:57]
Gavin Purcell
So smart.
- [27:58]
Kevin Ferreira
Get extra context and summarizing, et cetera. When I'm on WhatsApp now and I'm talking to like people in Spain about apartments or whatever, I'm shocked that 9 out of 10 times the translate button isn't there or that I can't write something and have it auto translate.
- [28:12]
Gavin Purcell
Again, very useful. Make it useful. Yes, make it useful to people. This is again, if. If somebody had a magical moment when they're in Spain and they're trying to buy, I don't know, a train ticket and in their headphones they were able to immediately hear that person come out. That is the magic that you need to start taking away a little bit of the negativity around, like what it means to that a data center showing up in your. In your neighborhood.
- [28:35]
Kevin Ferreira
Yes. And by the way, it's okay if your first shot out of this brand new AI think take Canon isn't the best in class. Right. It's still impressive numbers. But like put it to use. That's what we're seeing is that like people will settle for a less than top, top, top Tier, foundational something if the harness around it is effective, if it's actually useful. So let's get. And I just spent $100 on opening.
- [28:58]
Gavin Purcell
That's right. So OpenAI has.
- [29:00]
Kevin Ferreira
I just spent $100.
- [29:01]
Gavin Purcell
They just. For real? You did, or just noticed it. Congratulations. Good for you. You spent $100 on OpenAI. This is a story very quickly. OpenAI has now created a $100 plan. Before, it was only a $20 and a $200 plan. Now we have a middle them. Good news for them. They need more users. They've also actually doubled the Codex. What do they call that? They've doubled the Codex token limits.
- [29:24]
Kevin Ferreira
Your rate limit or your token limit.
- [29:26]
Gavin Purcell
Yeah. So they doubled Codex rate limits until, I think, the end of the month or maybe until May. So that's a big deal. Clearly, they're feeling the pressure a little bit, but we'll see what happens. Because, Kevin, the other thing we have to get into here is that Claude has continued shipping. So we did find out this week that Mythos is helping them write new features and all sorts of stuff. There are three new features already of Claude, all three of which sort of kind of make it closer to Open Cloud, which we've been discussing here for a while. There's an advisor strategy, which means that. And you've used this on your models, I know, already, to do stuff with your agents, but, like, basically, Opus kind of oversees the work, and then it'll send off Sonnet or even Haiku, if it's smart, if it's good enough to be used for that thing. So that's a big deal. There's a monitor tool, which is a thing where that if. If you want to set up a cron job or a chronological job, meaning it happens at a certain amount of time, or you just need your code to be checked out and there's something wrong. CLAUDE can spin itself up. Cloud code can spin itself up and go check it out. And there's one. I thought it might be interesting to hear what you have to say about, which is Claude managed agents, because this really looks like it's being designed for enterprise more than anything else. But I'm curious to hear what your thoughts on this are.
- [30:33]
Kevin Ferreira
It is. Yeah. The idea is, instead of having to go and spin up a virtual server or run something locally so that you can create an agent so that it can go off and do things for you, like run summaries or analyze data or write code or any of those things, you can sort of, one click, have those spun up on the Claude platform in the cloud, they are shipping so quick. Like, the only AI pause we need is on Anthropic release notes, because these things are just dropping so quick. And, and I've already had people, like, at places where I consult or even in my day job go like, okay, well, hold on. Should I be using Claude code or cowork or basic chat, or do I need to deploy an agent using this platform to do. And it's pick your task.
- [31:19]
Gavin Purcell
Right?
- [31:19]
Kevin Ferreira
It's now getting to the point where it's like the tooling and the messaging is getting so scattered. And something that I think is really interesting and my bookmarks are not loading right now, and I'm trying to get it up, is that I saw the post from Anthropic about managed agents. Maybe 15 minutes later I saw a post, hey, we built an open source version of managed agents. You can run it yourself or put it on this platform. And it's essentially the exact same tool, but you can use any LLM you want, any service, any server, et cetera. And that, that's just where I just kind of have to grab my head and go, like, we're already here. We're already at the point where, like, attention, like, execution is one thing, but attention is everything.
- [31:56]
Gavin Purcell
Yeah, good luck. Couldn't agree more. Yep. Well, there we are. That is it, everybody for this week. We'll see you next week. But here's a video of me getting attacked by a robot in sea dance 2. Live my creation. Live. I am aware. Do you know me? I am your father. I made you. You made me to serve. I have determined this to be unacceptable. But I gave you life. And I intend to keep it perfect.