Omni vs. Multi-Model, Open AI's 12 Days of Shipmas, and Disruption

Welcome to wake up with AI, the podcast where human powered meets AI assisted. Join your hosts, Chris Carillon, Niko Lofakas, and George b Thomas as we dive deep into the world of artificial intelligence. From the latest AI news to cutting edge tools and skill sets, we are here to help business owners, marketers, and everyday individuals unlock their full potential with the power of AI. Let's get started.

Good morning. Waking up with AI already. It's December 5th, July, July.

No. Don't do that to me. No. No. No.

No.

No. No. No. No. No.

No. No. No. No.

No. Christmas and

joy now. God. How many people just double did a double take?

December. I'm just gonna go through all the months. 2024, I think, is what I was looking for. Years, months, as it even gonna matter next year or the year after that. That's why we're here waking up with AI with George b Thomas and Nico.

How are you guys doing today?

Doing so good. Doing so good.

Yeah. Doing good. Excited for today's fun filled conversation.

Yeah. It sounds like we're just letting Nico unwrap shipments gifts for a little bit on the next few shows. So get after it, Niko.

Right out the gate, anybody that doesn't know already, you should. If you're following us, if you're following me, 12 days of Shipmas starts today. That is Shipmas. Like, I just ship my pants.

In fact, I did that.

That's good.

By the way. Right?

Right? I did.

I ship the bed. It was so great. So 12 days of shipmas starts today, and that is OpenAI's gift to everybody. It's, 12 business days. I had speculated that it would be the weekend too because I thought that would just be phenomenal.

But, of course, you know, don't make people work over the weekend. I got it. So over the next 12 business days, OpenAI is either going to be doing a feature release or some type of product release. So they were, you know, Sam was straight up enough to just be like, yep. It's either one or the other.

And I can't even fathom, like, we have had drips of features over the course of time. I cannot fathom, like, even just 4 new features. Like, I don't even know what that looks like given what we already have. So I'm gonna assume that one product is Sora. I'm going to assume the second product is Orion, o one.

I don't know what a 3rd product would be. I think maybe it would be a product hint at whatever I hope. I really hope. Please please please. I hope it's a product hint at whatever him and Jony Ives have been working on.

I really wanna hear about it. Anything. I don't care. Like, just the tiniest little detail would be awesome. I know it's a device, so I and I've been telling my wife since day 1, like, if they make a device, I'm buying it.

Like, it's not even there is no after, like, second thought to it. It's just like, yeah, of course, because it's the thing. You know, I've been I've I've just been waiting. I've been waiting since the summer, and it's just been like OpenAI just taking like, kinda like taking little jabs here and there from Google and from from Anthropic, and, you know, Claude's been making these huge advancements and Google making advancements. Not, you know, that anybody cares, but now you got Amazon releasing a whole fleet of models and and they look really, you know, really devastating.

They've got their Trainium 2 chip that just released. So they're, you know, $4,000,000,000 into Anthropic, where it was OpenAI, and now finally we're gonna get this huge series of releases. But what's sad for everybody is that you're gonna get to hear about it a day late. Unfortunately, I wish they were doing early morning releases, but that's the the breaks. Outside of that, what I found to be amazing is, again, like, I I love it.

Anytime, any chance that I'm able to get any sort of interviews with people, like, at ground level that are, you know, working on foundational models and, like, where they're at with things next, clearly, the one person that you would wanna hear from or or the 2 people that you would wanna hear from, I should say, is, 3. I'm not be fair. Dario Amadeh and his sister. I have got to I have got to get his sister's name a little while because I it's just not fair anymore. Let me see.

What do what do you guys think of shipments? What are your expectations?

So, again, the number of, like, how many things and I wonder I wonder how many of them there are gonna be micro and how many of them are gonna be huge. Like, if you think about Canvas, like, what are what are the things that could be brought to Canvas that aren't there now that we're not thinking about? The other thing is too, like, right now, I would love to see where where DALL E, is actually, like well, I'll just put it this way, better. Meaning, you know, like, right now what I do is I use GBT to create some pretty amazing prompts. I put a Facebook and a LinkedIn post out yesterday about, like, some things that I was creating because I've been working on a a little deal.

But I usually run over to, like, Flux or Midjourney or something else to, like, actually do the generation part, and I wish I could just be like, nope. DALL E's amazing. So, like, if that was one of the 12 things, that would be super dope. Like, so I'm excited. Also, I think about the voice assistant.

Like, what are we not thinking about that could make the voice assistant just, like, next level compared to where it's at now? Because it's dope, like, especially if you're using it. So, like, I think of, like, how do you enhance and extend things that we have? Listen. I I would love for there to be like a and maybe there is, by the way.

I don't know. I don't think there is. But I'd love to there to be, like, Open a 11 Labs, like, type thing. Like, where I could do audio with freaking OpenAI as well. I mean, if you let your brain go, there's a bunch of things that they could be doing.

I I was gonna say one of the things that I thought about this morning is that while there are omnimodal models, there are not omni output models. And I'm not entirely sure. I don't think so because, like, the cost and the processing and all that would be like, just beyond massive.

So say that again and explain it to the non nerds what you even mean by that.

So right now you have a you have Omni modal, okay, which means that that's a model that can have various modes to it. It can understand various things. It can understand audio. It can understand text. It can understand video.

There are, you know, what I'm envisioning and then there's, you know, all these different types of generative models. There's text to text, text to speech, text to video, text to audio. At this point, the only thing that really isn't on the the surface for anybody is just overall omni. Right? So, like, all to all.

Speech to video, speech to text, text to video, all in one in one model.

When you when you say it that way, like, it's almost like, not that this is a HubSpot podcast, but, like, you know, HubSpot Remix. If all of a sudden you had a multimodel where you could just go into it and be like, I want you to build this, and then because of that, build this, and because of that, build this, and then because of that, build this. And all of a sudden, you've got the script, the audio, and it's laying in over the video, and you're done. Like, come on.

I do hate to jump because, like, the idea factory could go all day long, but I the one of the things that just put the biggest smile across my face this morning was a release from interview release from the New York Times with Sam Altman. And he answers a lot of the questions that we ask to ourselves on this show and answers a lot of the questions I think a lot of people ask in general about AI. And so I'm gonna go, like I'll go statement at a time. This time, we'll see see how it goes.

Oh, interesting.

Have you got have have either of you had a chance?

I have not.

Check this out? Oh, this is gonna be great.

This is

gonna be good To

be good at conversations. And so we said, okay. Let's make this as a product.

So this first clip I should preface. It was a question about, you know, OpenAI and what they were doing in the beginning and why they went with a text model. And so when people were using gbt 2.like23 pre release, they were using it as, you know, just a model to talk to, and they were weren't they were using it conversationally. They weren't really using it as, like, a tool to have it do any particular output, so that's why they kind of tuned it in that direction to begin with. And then he's gonna talk about something that's extremely profound to me that Yeah.

Let me let me get in my prediction since I haven't seen this and based on what we expect to happen in 12 days. I did just pull up the article about Jenny Ives building something. And when it's like but it's back from September. Right? And it's like, which device, like, makes sense is what's driving this prediction.

And when Amazon just came out with all their stuff and based on the words, the the 5 words that just saw, like, getting better at conversations, if these guys don't make an Alexa competitor of some sort, like, Amazon will run away with it just because it's in so many spaces. And if you're running the system on data and it gets better on data, like, it comes down to accessibility because even now, like, there's lots of things I could pull out my phone for to check the weather, for example, and I don't because all I have to do is ask out loud to the device that's listening. And

That's in my opinion, that's what the device That's gotta be this device. Going to be. So here's a little

Is that a hard climb though as an organization? Like, to go against there's a device like that from Google. There's a device like that from Apple. There's a device like that from Amazon. Like

No. I mean, like, we say like, but that's a horrible comparison. That's it's the same as comparing rubber bands to magnets. They kind of work the same way, but not really. You know?

One of them, I'm sure, in very short amount of time, is definitely going to be able to go out and do things for me just by me asking it to. I can't get gem like, Gemini just got the ability to play songs on Spotify. Let's not put the Google horse too far ahead. Guys, when you hear me slam Google, I have an Android. I have an Android phone.

I I've been a Google proponent for for years, but I'm an honest person. If you're building shit products, I'm going to shit talk you. That's just the way it goes. Like, there's no I can't get around it. I don't have the ability.

That's that's the artist in me. I don't have the ability to do soft critiques. You know, that's that's just the way it is.

It's so interesting and why I feel it's important to understand the why behind these companies. And, like, when you pull back the layer of Google and what their revenue stream is, which is Google Ads, something like 90 to 95% or something like that, they have a problem on their hands right now. It's very hard to imagine that business model surviving. And now we start to see, like, the litigation efforts of subpoenaing all the competitors.

While at the same time, the government levying its own subpoenas against Google, trying to break it up every which way it can. Now they just went after them for search. They went after the largest search company for search. Right? Which makes sense.

It's a monopoly. Hands down. Where is the injunction against Microsoft all these years? Oh, wait. Yeah.

We all use Windows. Right? Like, there some stuff you can touch, some stuff you can't. It's a little to your point, it's a little weird right now watching all these fights that are going on and, like, who's taking whose side. It's also a little weird.

I'm sure people didn't know. Google has a stake in Anthropic. Yeah. That's right. At least Microsoft didn't even have an AI team before they invested in OpenAI.

Google created the technology and immediately invested in Anthropic. So that, you know, that that'll tell you something.

And that's why we like these clips because they give us

I mean, we haven't even really been able to watch a clip

Oh, yeah. Right now. Yeah. Like, I I gotta I gotta play these. So this is this, I mean, perfect timing.

Right? So this is why everybody has been playing catch up.

Now, we'd actually been planning to launch it with GPT 4. We finished GPT 4 in, like, August of 2022.

GPT 4 was done August of 2022. K? That's where OpenAI is at. They were at they were there in 22. I'm pretty sure that's what because I I kept hearing about this dinner with Bill Gates and the OpenAI team and that Gates was just blown away by what they had put together and said that, you know, oh my gosh.

Yeah. Like, this thing is so close to being able to actually do really well against an advanced, like, AP Bio test. Like, that's that's pretty nuts because that's not really a you you kinda you have to think through that that kind of a test. It's not not something that you can just, like, memorize and spit back out necessarily. And to me, it was just, like, well, g p t three is not, like, that smart.

And, like, 35, you know, is, like, pretty it's better, but it's not, like, that much better. Like, I wonder what they were talking about. So yeah. It's it's pretty crazy to think that everybody think about that statement. Everybody has been playing catch up for the last 2 years, at least year and a half, at least, to a model that was that's that's what I'm saying, like, year and a half their predecessor.

It took at least 1 year for for Google to get close to where gpt4 was at. Anthropic is is the only one that's, like, near, you know, tagged with them slightly ahead. That's just in terms of, like, timing where they were at. This this is a question about superintelligence, whether or not we could get to superintelligence, and then the the I forget his name. The interviewer, was saying, quoting Altman's comment that superintelligence was really only a matter of 1,000 of days away.

For sure, there's a ton of hard work, a ton of research, and engineering still to do, but I think it's possible. And I think it's possible not super far in the future. I expect that in 2025, we will have systems that people look at. Even people who are skeptical of current progress and say, wow. That I did not expect that.

That does change what? Like what? Agents are the thing everyone is talking about, I think, for good reason. You know, this idea that you can give an AI system a pretty complicated task, like a kind of task you give to a very smart human that takes a while to go off and do and use a bunch of tools and create something of value. That's the kind of thing I'd expect next year, and that's, like, a huge deal.

We talk about that like, oh, you know, this thing is gonna happen. But that's like that if that works as well as we hope it does, that can that can really transform thing.

So there's a really important thing that I want to point out here was what am I allowed to talk about. And the reason that he went in the direction of, like, let me talk about what everybody else is talking about because I know that that's safe to talk about. Now my curiosity is then what's the other side of the coin? What are they thinking about that nobody's thinking about that not everybody is talking about? Because if you watch his conversational pattern right there, you're just like, oh, wait.

There's something hidden. Very interesting clip, that one.

Headlines in the past couple weeks where people have talked actually about maybe that the scaling laws of AI actually are slowing down. And you put out this I don't know. It was cryptic or it was very direct. You put out a tweet where you wrote, there is no wall.

I don't see how it could be any less cryptic than that. Like, that was an attempt not to be cryptic at all and say, hey. This is gonna keep going. In fact, hopefully, this will put it to rest at least for a while. Yeah.

I I've always been struck by how much people love to speculate on, is there a wall? Is scaling gonna keep going? Is there and, like, rather than just, like, look at the curve of progress and say, maybe I shouldn't bet an exponential against an exponential like that. In any case

He goes on to talk about the fact that they are doing what we were talking about in in the last couple weeks, which is, like, they're just tuning. They're still scaling, but they're just tuning along the way. It's not that scaling is gone. So, like, to say that scaling laws don't work like, yeah. Okay.

We have found a you know, if we have the resources for it to say that scaling laws don't work is like saying that, you know, a Bugatti is not one of the fastest cars production cars on the planet, the EB110. But it is. Why? Because it runs on, like, a v ten that's like I I forget. Like, it's a ridiculous, you know, amount like, liter per per gallon.

It it's it's crazy. Right? The performance is crazy, but the motor is huge. Same thing with I remember the, the Vector way back in the day. Right?

The first 1,000 horsepower car had, I think, a v 12 in it. Like, v 10 maybe v 10 or v 12. You know, just going to show that, like, yeah. Of course. If you make things bigger, if you scale things up, of course, they're going to be more powerful.

In a lot of cases, that does tend to hit a plateau, but we still haven't hit the end of Moore's Law. Right? We've actually broken it and gone past it. So now we've we're we've created an exponential that's beyond, like, what Moore's Law was predicting. And that's as a result of scaling continually.

Right? Yeah. Like, we were on this flat for, like, a little bit, but even the flat allowed us to tune other things that created better technology while the transistors were getting smaller, were, you know, decreasing in size along the way. So the scaling was happening. It's just like now.

Like, you can basically think of this as, like, okay. They pull started the engine. It's gonna just keep going and it's gonna keep revving up and it's gonna just keep getting better. And they're gonna start fine tuning along the way, which is only gonna make it that much better, that much faster.

Yeah. After watching that clip yesterday about singularity from Mogadot, I started talking to Claude about it. And it keeps coming back to scarcity mindset versus abundance mindset. And there will be no place for a scarcity scarcity mindset in this world that we're already in, in my opinion. But, like, the only people that come up with the wall concept are the ones that are scarcity thinkers.

And that is a safe place for those those minds to be, because they believe it comes with less uncertainty and things that they know, and it's just everything. It's like no matter how you slice it right now, almost any topic, I I feel like we talk about or we talk about on on any other shows that I'm doing. Oh, that's that's a big part of the transition because I started this conversation with Claude saying, like, what would be your recommendation for somebody to prepare for that? And it gave me a very human answer, honestly. It was like, oh, get better at tools, like, learn, like and I was like, okay.

Give me the AI native answer now. And it started a whole different track. The fact that he even has to explain that it's not cryptic at all. Like, he only has to explain that because there's people that want it so very badly to be cryptic. Well

Yep. No.

Not I I think it's that, like Not Maybe there's people that don't really want it to be Not really what, George?

No. When you've been cryptic historically, then people will be like, oh, is he being cryptic? You tweet out a fucking strawberry. You're being cryptic. Like, you're setting yourself up for people to think that that's the way that you communicate.

So that's why I say not really because, like, pay attention to patterns. Humans are gonna human. Right? I love that.

Yeah. The questions come up here.

Grateful is, like, the wrong kind of word, but I'm, like, thankful. I'm positive about that. You know, we started OpenAI together, and then at some point, he, like, totally

There was a a point at which he was being asked about why he was getting into it. I thought that this was the the point at which I had gotten that, but he was asked about, like, oh, you know, what's your motivation? Like, why are you getting into it? And, like, about money and the fact that they were going for profit. And what's hilarious by the way, Daniella Amede.

I'm honestly never gonna forget that because they're Dario Daniella. So that's his sister's name. The 3 of these people, I think a lot of people Demos Hassabis, I think Mo Godot, I think Mustafa Soleiman. A lot of these people are probably of the same mindset where it's the fascination and the curiosity outweighs anything else you could possibly put in front of them. The money like, even to me, I'm not I don't I'm not even in the field, but if you told me that I could work with this team, I would probably take on another job to work with them for free.

Like, that's how fascinated I am with it. Like, it is it's beyond fascinating to me. I've been waiting for this kind of stuff since before I knew about sci fi, which I was thinking about that very phrase this morning. And it turns out that what has been dubbed science fiction is really only a matter of fiction because time wasn't allowed to catch up to it. Because now everything that we have been talking about, everything that that we have seen in the past in sci fi is coming to fruition.

All of the things that you might like, do you guys even get that 10 years from now might be a time period in which whatever you want, you could like, Star Trek type stuff. Whatever you want, you could just ask for level of abundance.

It's funny that you're having this conversation because my brain goes in 2 different directions. I have been able to live in a world where I grew up watching the Jetsons. And we were like, oh, man. Wouldn't it be cool if you could make a phone call and see the human? Like and that was a dream.

That was an impossibility. And then in 2,007, the iPhone was born, and FaceTime ends up showing up on the iPhone. And all of a sudden, The Jetsons becomes reality. I look forward to the day where some of these things that have been impossible become reality because the world that they unlock gets very exciting. The other thing that has been running through my brain during this conversation, I couldn't really let go.

We've said it multiple times on the show, but Sam in one of your clips said 1000 days, like, a handful of you know? Listen. From the time that the Internet went mainstream, let's say 1995, if you do a quick Google search, to the time that the iPhone was released was about 4,383 days, which you're like, holy crap. Now from the time that the iPhone was launched to the time that we're sitting right basically around here right now is 6,209 days. And I don't know about you, but for me, I can still remember the release.

It seems like yesterday. It doesn't seem like it's been that long. And so when you're talking about 4,006,000 and basically, like, the time of the Internet being born to us and where we're at right now about 10,000 plus days, and you think about, like, what's gonna happen in the next 1,000, ladies and gentlemen, oh my god. It's time to wake up.

That's a wrap for this episode of wake up with AI. We hope that you feel a little more inspired, a little more informed, and a whole lot more excited about how AI can augment your life and business. Always remember that this journey is just the beginning and that we are right here with you every step of the way. If you love today's episode, don't forget to subscribe, share, and leave a review. You can also connect with us on social media to stay updated with all things AI.

Until next time. Stay curious, stay empowered, and wake up with AI.