Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
InflectionAI Raises $1.3B | Blockchain | Unity’s ChatGPT* #AITownHall
Recorded: June 29, 2023 Duration: 1:28:01Space Recording
Short Summary
Inflection AI secured a major $1.3 billion fundraise with strategic partners like Microsoft and Nvidia, highlighting significant investment trends in AI. Unity launched its AI platform, Muse, marking a technological innovation in game development by integrating advanced AI tools for animation and texture creation. The discussion also touched on industry trends where companies are moving to favorable jurisdictions, drawing parallels with the crypto sector.
Full Transcription
All right, everybody.
We're getting everyone up.
We got a big day today.
Some big developments this morning and lots of other stuff going on.
So the world of AI just never stops.
All right.
So, yeah, I mean, the main thing is this morning, Infliction AI basically, you know,
announced a pretty monstrous raise, $1.3 billion raised.
That's not the valuation.
That's the raise.
The valuation itself is about $4 billion.
So, yeah.
You know, before, I'd say a few days ago, we were talking about how, you know, 1.3 billion was the nice, obviously, exit and nice valuation.
And now we're talking about fundraisers and that size.
So some pretty incredible things.
Maybe before we dive into that and we'll get to Unity as well later, they release some incredible news.
Yeah, I mean, I don't know, John, what's some of your favorite stuff happening in AI right now?
That's maybe unrelated or related to some of these headliners.
One of the things I find I find interesting is the rate of change in of itself.
So it's, and this is more personal.
I'm trying to prepare a talk in a month, and it's impossible because everything is changing.
So the rate of change is really interesting.
Recursive learning is real interesting for me now.
It's this, I'm calling it sort of this infinite spiral of learning, and I'm trying to relate it to, you know,
The hallucination construct, it seems like AI has been tagged with a lot of psychopathologies,
even though hallucinations is the wrong term.
Hinton said that it was more of a confabulation.
So for me, it's very, very interesting.
I've been studying a lot of this sort of recursive learning and this sort of spiral of innovation that is quite interesting.
Yeah, definitely. I mean, we're there's so many topics and I've been chatting with lots of folks about things that are going to be interesting.
You know, AGI, you and I were going back and forth on this. I mean, any initial thoughts before we dive into the heart of the matter on this?
Well, I'm very impressed with the inflection.
So they have a cluster of 22,000 H100 GPU.
So that's, I'm very happy to see that.
That means that AI is accelerating.
And so this is a old founder from DeepMind.
So they were three funders.
So Mustafa left...
very early. Deep mind, it was not kind of aligned with them, I think.
And so I'm very happy to see that just kind of one person can raise $1.3 billion.
So it's very promising.
Yeah, it's actually interesting, right?
I mean, basically it's like one, you mentioned the word one person, right?
So, I mean, some of these companies, like Mistral AI, famously, four-week-old company had three co-founders, right?
Two from Deep Mind, as well as one from Meta's AI.
So basically, it's like these small teams.
I mean...
You know, I'll talk a little bit about what this funding round means.
And then, you know, I'd love to dive into the kind of kind of the heart of this is happening.
But to give folks an overview, just so you all know, inflection AI raised $1.3 billion.
The investors included, I'm looking at this report here.
So, you know, this literally just broke.
So still looking at the details of the funding round.
I think most people in AI know about inflection AI.
But basically led by Microsoft, Reid Hoffman, Bill Gates, Eric Schmidt.
The new investor is Nvidia, which is.
interesting, right?
So it looks like this was a trunched investment because they mentioned that some
insider was talking about some tronching.
So total, total raised about $1.5.25 billion.
four billion dollar evaluations.
But for those who don't know, I mean,
it's hard to know about the structure of fund raises.
And it's funny, I don't want to turn this from an AI space
to a finance space, but maybe a little bit of that.
So basically the cash they raise is 1.5 to 525 billion total,
right, over some period of time.
And tronching can occur if investors decide to put in money
at various stages, right?
So maybe at various milestones, et cetera.
Honestly, it's a good thing for investors,
not so great for if you're the founder or the person
on the receiving side of the investment.
So basically the valuation of inflection is $4 billion, right?
So that's what the value is.
You know, you start to wonder and worry, well, you know, like large exits, like the
Databricks acquisition was, you know, in the neighborhood of this fundraise.
So like who's left to like acquire these companies, right?
It's only going to be the big folks.
So, you know, raising big money sounds great.
You know, it makes for nice tech crunch headlines.
But as an entrepreneur, now you have to pay that back and then some.
Right? Because investors are looking for return on investment. So fundraisers are literally just the beginning, not the end. And as you mentioned, AGI, CEO Mustafa Silemon was a co-founder of Google owned AI Lap Deep Mind. So, you know, Infliction 1, they announced this recently. It's an AI powered model. It's a foundation model. They say that they being the company says it's the best foundation model in its compute class.
So it basically says it beats out.
So GPT35 is in its compute class, according to the company.
You know, chinchilla, palm, llama are all in its compute class.
But it doesn't count GPT4, which requires a lot more compute as part of its compute class.
So just I tweeted up, put it in the nest.
They have some nice things.
They have a technical paper on inflection one.
It's worth looking through if you're interested in this stuff.
But Josh, maybe I'll go to you.
What do you think about this fundraise?
Yeah, I think this fundraise represents, you know, a lot of what Databricks did, which is that these guys understand that this represents a huge portion of the future labor market.
These AI agents replacing jobs of, you know, people in accounting, financing and so much more.
or what this represents things like data bricks and inflection AI, these giant acquisition numbers,
these big fundraisers, represents the portion of the future labor market that these platforms
are going to be essentially powering.
And so it's very interesting when you look at all the big players coming in.
It's essentially like owning railroads or owning, you know, actual roads.
It's the way in which these AI agents are going to be able to perform.
So it's very interesting.
This is really the future, the labor market.
And yeah, that's my thoughts.
Super interesting. AGI. What are your thoughts?
Yes. If you look at the value for the work, the value that's created by all of the
human that are working using a computer, it's something like 120 trillion a year.
So if you look at it that way, those amount of $1.3 billion to secure a part, a fraction of that
market. It's kind of...
It's like a very discounted body.
did we lose you?
I'm still there.
Okay, so you think, so you think, so one question, maybe both to you, Josh, like, so basically it's like, okay, we're talking about these large markets, right?
But like,
You're talking Josh compared it to the new railroads.
I mean, do we think these are all justified?
So basically, if you think the market opportunity, the exit opportunity is there, right?
Like trillions of dollars are going to get disrupted.
If you believe that, then you say, well, actually, these are undervalued companies, right?
But I think most, you know, normal folks like me look at these, you know, we'll look at this.
You know, folks from the street will look at this and say, hey, you know, scratching their heads.
What's, you know, what's going on, right?
Like, is there, you know, is this the sign of a bubble?
So what are your thoughts, you know, either Josh or AGI?
Yeah.
Yeah, I just want to punch in. I do not think this is a sign of a bubble. I definitely hear what you're saying. UIC for sure. But no, I think that this is foundational growth for the economy. I think what we're seeing is completely new tools that need to be powered by large sets of data, by a lot of energy. And so what you're seeing is actually a foundational buildup of what are going to be new curves and new markets.
Now, these AI agents, I think it's very interesting when we start looking at the way that they're going to replace jobs.
And obviously in America, and I don't want to deviate too much, but in America, we have the privilege of talking about things like universal basic income.
and other parts of the world where, you know, 100,000 students graduate, but there's only 10,000 jobs.
How does, you know, AI agents and data bricks and inflection AI in the future labor market affect the third world?
I think it's a very interesting part of the conversation as well.
So just wanted to answer some of that of what you're talking about, EYC, and then touch on.
You know, I think it's, I think we should discuss a little bit about what this means for,
you know, people. And, you know, although folks will be replaced by AI, there's definitely a new
curve of creation. But what does that look like in terms of making sure people have the basic
needs met and that they're spending money in the economy? I think this is all really important.
Yeah, definitely. And AJ, I'll go to you next and welcome up, Brian. But I wanted to kind of frame this,
right? So a lot of us who have been in startup land have looked, have been a VC. You know, we've seen
large fundraisers. But I think what's, what's kind of,
different here is we're seeing very large fundraisers, which, you know, have happened in the past,
but for very early companies, meaning very new companies, with very small amounts of people,
some cases it could be, you know, you know, like just a couple people, right?
So the question I ask is, why, why, right?
So, for example, you know, GPT-35 reportedly took somewhere between $5 to $10 million to train, right?
you know, the compute costs for NPT, the latest one was, you know, Mosaic's latest was somewhere in the neighborhood of 200,000, took about nine and a half days, zero human input, right?
So the cost of these models go down, compute costs are going up, GPU, you know, we're actually compute bound for the first time in a while, right?
So, you know, GPUs are scarce.
They're, you know, it's still hard to find the ones you need to do this.
But at the same time, you know, the cost to do models because the caching and stuff is going down.
So why does a small company that just started need $1.525 billion?
Right?
Like, I'm curious if somebody could address that.
Yeah.
Well, I think people are starting to see that scale is really, scale matter.
And that you want to, it's an opportunity of 120 trillion a year to replace human workers that are using a computer to do, to produce some valuable work.
So if you look at it like that, $1.3 billion is very a small amount.
So you want to have the largest amount of money as possible.
You want to secure the GPUs.
You want to train the biggest model as possible, obviously with the latest technique.
And somebody like this one, D'U.
Obviously, knows how to train that kind of model with kind of
deep mind technology that they use for Meev Zero.
So obviously he's able to do something that is better than chat GPD4.
And if you, and you want to be first also.
The people that will be first, they will secure a large share of that
of that kind of of that value.
So you want to be first, you want to secure that value.
And obviously those AGI agents will be able to do work that people are doing.
It's just economically, I will just do the same.
And I'm doing the same, in fact.
That is so interesting. Brian, let's go to you. So why do these people need so much money? And, you know, AGI talks about first mover advantage. In other industries, we often see second or third movers being the better winners. Like, you know, MySpace was the first mover. Facebook was the second. But I think AGI has some compelling thoughts about whether first mover matters a lot here. So yeah, Brian, do you want to tell us, is that this time is different or are,
Well, it's, thank you for having me on. I think it's both and neither. We're in a sort of a new paradigm. The difference is prior to this moment, when you started a technology company, you needed a lot of technologists.
And from this moment forward, you don't need as many technologists to start a technology company.
You have an in-house technology called generative AI, right?
And large language models, things of this nature and moving forward,
it's some of the stuff that I'm testing right now, sort of a new technology.
The coding that we're going to see coming out of these types of platforms are going to be mind-blowing.
And so if you look at how venture capital really kind of gestated out of the Silicon Valley, it was always around that sort of technology founder and things of that nature.
And before our very eyes, that's shifting.
So where is it shifting to?
It's shifting to the people that are creating the grand ideas.
It's shifting to people who can execute, but maybe not code these ideas, because you now have endless teams of AI to do that.
So that's one aspect of it.
The other aspect, and you absolutely pointed out, is although H100, you know, Nvidia cards, and they're now going to be a few equivalents to that coming out very rapidly,
Yeah, there is a constraint supply for that type of model building that we're used to in the current AIs that we're deploying.
But that's going to shift rapidly also.
And on the other side of that is the amount of cost to build smaller, very dense local models like StarCoder.
StarCoder was built on a couple of hundred dollars on cloud compute.
based on sort of the Lama model and things of that nature, which were already built.
So we're building on things that were already built, and those things are making it a lot less costly.
So what does it mean for startups?
It means now startups have much more opportunity.
But a lot of the venture capitalists haven't quite caught up with this idea.
I'm working with a lot.
I'm trying to guide them on how this is going to look.
And it's hard to shift.
You're so used to the old ways of doing things.
you know, come to the Bay Area, go to Menlo Park, you know, and whatever, and hang out in San
Hill Road, give your pitch and then get a, you know, 100 million, 500 million, something
like that. And boom, you take off. I don't know if that model is going to be the sustaining
model. And I've said this quite a few times in these spaces. We're going to see the first
five, ten person company.
that's going to be a trillion dollar operation and primarily be based on using AI to a massive level.
So if that, in fact, is the truth,
And I absolutely believe it is, the amount of money necessary to get to $1 trillion is going to be
ignorantly smaller.
That's the lever.
That's the magnitude effect that we're going to have by utilizing AI.
So in the interim, we're going to see a lot of this.
Brian, though, Brian, I agree with so much of what you said, but want to push back on
on one part.
And that's related to, you know, the establishment versus the new players, right?
Like, totally agree.
I think.
traditional investors, we'll call traditional DC investors.
A lot of them know things are coming.
I mean, they tend to see things, you know, trends before, you know, let's say later
stage investors down the pipe.
But it's like suddenly you have your billion dollar fund and you don't need,
you only need a fraction of that, right?
And so there's just too much capital chasing too few of these opportunities.
But it's like you look at the, you know, the investor base, the cap table for this raise.
And you can't help to think, well, is it, is it the new folks, right?
I mean, sure, inflection is new, but the CEO is a deep mind person.
The investors are Microsoft, Reid Hoffman, Bill Gates.
And Vitya is the new investor in this round.
Like, I mean, isn't it just like the establishment?
I mean, I think it is.
It is.
And Hill Road establishment is going to be the one who are going to be the five people in this trillion-dollar company.
It absolutely is.
And that's kind of why it's...
We're not really seeing the very exciting garage stage of this yet.
And that's, we are, if you're in the open AI community, you're seeing it.
It is absolutely massive.
It's like when the internet was arising.
I've been around seeing that.
It was all built on open source.
It was all built on RFCs, which was more academic open source, but nonetheless open source.
And it took quite a while for,
venture capitalists to fully understand and appreciate what it was.
This is sort of inverse of that,
but they're kind of clamping on a direction that fit their prior paradigm.
I think it's Marshall McLuhan.
It said that we only see the future in the rear room mirror of the car we're driving, right?
And this is true for all of us and certainly for even the most advanced venture capitalists.
So I'm not putting them down.
What I am saying is the things that are rising up,
with hundreds and thousands of creative minds,
not just a few dozen at a startup,
that power is being amassed right now,
and the stuff that we're seeing come out of it
is phenomenal. Now, yes, some of the bigger companies did donate to that. By accident, quotes,
you know, Facebook put out the Lama model. It was leaked, sort of, not really. I think it was put out there,
but that's my opinion. And a lot that sparked this. And it also set everything on its side.
You know, everybody thought, well, everybody's lining up. You got Google versus Open AI.
And then we have the no-mote, you know, memo, the Jerry McGuire memo, probably got that person fired.
I think they tracked them down.
And it essentially said that.
you know, the larger our models are, and the more complex our models are,
the harder they are to contain. And the secret isn't the corpus of data.
Once you reach a certain level of the data that you've slurped up, it's a diminishing return.
That data is there not so much for knowledge, the general knowledge.
It's more for the understanding of how words sequence, what word should follow the next word.
And after you reach a certain level, like I said, you're wasting compute power.
It's the training and it's the retraining whether human or machine based.
And that's where all the innovation is coming from.
And it's coming from, you go to hugging faces of paper every 12 seconds.
And most of those people are just working out of their bedrooms, kitchens, and garages.
They're not at academia or corporate.
So it's fun to see.
And once somebody starts riding on the back of those new ideas,
we're going to see a whole different sort of –
monetization, if you will. I'm not saying everything is a free and open source and there's no
monetization. I'm a capitalist and I believe that we need to do that. But what I am saying is the
innovation cycle is not being constrained anymore to just an elite few who know the right place to be
and rub elbows with the right folks. It's egalitarian and democratic now. And I think that's a good thing.
I think it's so fascinating.
So, Dan, welcome to the stage.
Thank you for joining us.
You know, we'll go to you.
I know Josh has the hand up.
So maybe we'll go to Josh first.
And then, Dan, I'd love to get your perspectives on this development.
So actually, yeah, Dan, looks like you unmuted.
So Dan, why don't you go for it?
Yeah, I'll just say something really quick, which is that...
you know, something that's on the horizon
is that people are gonna have to start paying for data.
And, you know, we just heard maybe the argument
that data's not so important.
I'm not so certain of that.
And it's definitely gonna change
if large fractions of say open AI's training data
is found to be, you know, data they shouldn't have used.
And so I don't think necessarily the VCs
were thinking about this when they gave large amounts of money.
But those large amounts of money might turn out to be useful
if you suddenly have to start paying for your data.
That's interesting. Okay, so can you dive into that a little more?
Like, why do you, so I mean that some people think, yeah, we've actually even on this space has talked about that, but what is your perhaps more in-depth view on that matter?
Okay, well, it's based on a few assumptions, one that, you know, there are legitimate sources of data such that the licenses permit this sort of use that is being made of these things.
And then there's a huge questionable, maybe a vast amount, much vaster amounts of data, like, you know, training on books that are in copyright, et cetera.
And it seems likely to me that there's vast amounts of data that,
If we actually knew we're being used, then those people would be very upset.
And there's a, you know, there's a court case brewing or maybe just launched,
which, you know, it's a bunch of angry people, people angry at Open AI for using their data or maybe using their data.
And discovery will maybe, maybe discovery will shed light on something that's closely kept secret at Open AI.
And so, you know, data, which...
is, you know, data is what's fueling this to a large degree and scale, but data very much.
And so if data all of a sudden becomes expensive in a way that we've never really thought about it as being expensive because we just...
scour the web and soak it up and don't pay anyone for it.
So Dan,
Dan, definitely agreed on the issues there, right?
And particularly with the image generation models too, right?
Like majority, stability, et cetera.
So the question though is, I think the EU is making a clear stance that this is going to be very tough for 10 AI companies right there, you know, in their jurisdictions.
You asked a little less clear, more ambiguous.
You know, we're seeing this in Web3 in crypto, right?
You know, when jurisdictions like the U.S. are suing people like finance and Coinbase,
they're all fleeing to other places, right?
So, I mean, this is like Pandora's box.
It's out of the bottle.
So, I mean, if this were to occur, wouldn't there be a world where someone in the world
or some set of jurisdictions would be okay with the kinds of data being sold and being,
quote, stolen or taken, scraped?
And big companies would just move over there, right?
I mean, wouldn't we predict that to happen?
Yeah.
But, you know, I will say that, you know, these large sums of money might be useful for these companies, which are, you know, firmly and, you know, on U.S. soil at the moment, there might be an awkward time where they have to start paying for things.
And also, I just, you know, I'll just say one more thing, which is, like, not to underestimate that some of these companies have indescently,
individuals who are really incredible and, you know, maybe at the root of engineering successes
that they're seeing, right? So when I think about some of these acquisitions, you're getting
like incredibly talented teams. And it's hard to underestimate because, you know, these things
are still, you know, how is Mosaic training without any human intervention? That's because
some brilliant humans got it to be...
to that point.
And that's not something that's widely understood yet,
widely enough to,
widely enough to,
like,
be open sourced.
Yeah, and it actually goes back to what Brian said.
I know we got some hands, so I want to go to those.
Because there's this kind of two narratives, you know, that I keep hearing.
One is like what Dan said, hey, you need these like brilliant people to do these things.
And then, you know, Brian's positing.
I think these are not mutually exclusive, by the way.
It's more, it could be a sequencing.
But Brian's like, hey, you don't need all these like really technical people.
But the question is, what's the time horizon, right?
Like, you know, the 10X engineer, I think is, I mean, I think these fundraisers are showing.
They're even more valuable in.
in this world than they are in the past.
So, I mean, perhaps what Brian's referring to, I mean, based on the, the hundreds,
I'm guessing what he's referring to is more of the time horizon, right?
Maybe at some point we don't even need the technics engineers, right?
Because, you know, I've used chat GPT for coding endeavors.
It's fine.
It's great, actually.
It's great for what it is.
But it doesn't yet, if you want to do something serious,
it doesn't yet replace great engineering talent, I find.
So, yeah.
yeah anyone want to dive into that or that dichotomy i know josh i got i got to just say back
that up you're absolutely right uh and
And anything I say, you've got to preface with the idea that the human is always going to be much better when it comes to creativity, when it comes to doing phenomenal work.
And if we stand on the shoulders of the giants, and that being AI in this case, we're going to be that much better.
So we're never eliminating the exclusive and elite engineer.
They're always going to be necessary.
Right.
But we can all agree that many of the apps that we've seen and we use did not necessarily require incredible engineering.
They just required somebody who knew how to code.
And now that is being democratized and liberated in the hands of everybody.
And that's the change.
So those engineers are now liberated to go on to do better things and more things.
So it's just a matter of the boats are rising up in the harbor for everybody.
It's fascinating.
I mean, we all predict, oh, you know, the cost of engineers should go down, but I'm not sure.
Yeah, I guess what time will tell where we go from there.
So, Josh, let's go to you, and then we got some other hands.
So, and welcome to some of the new folks on the stage.
But, yeah, Josh, what do you have?
Yeah.
Well, I just want to point out what seems to be, you know, a very clear paradox here. And that's that is that of course there's a democratization of these skills and knowledge and Brian Wright, you know, folks who didn't know how to code can code now. The paradox lies in the fact that the AI agents will be able to do it better than the people who now have those skills.
so they don't become employable versus the AI agent.
And that's a paradox.
So the time horizon matters.
The question is those folks who are replaced, do they go into the art of creating and using
AI to create their vision?
Most of them know.
Some of them, yes.
But most of them.
But most of them are going to be in a position where their job has been replaced.
And so I don't know if it's going to be something along the lines of disclosure to the SEC about how much of your company is operated by AI.
But it's likely something like that.
And you'll be taxed heavier if more of your companies operated by AI to pay for the people who are now replaced.
So the time horizon is very relevant.
And these acquisition deals are very relevant as well.
It's a paradox in motion, and it's really interesting.
We're all here talking about it.
Yeah, that makes sense.
So strangely, but what's going on.
And also, like, a question I have to ask is, I mean, we're talking about inflection.
I mean, who has used inflection one, right?
Like, what, like, do we feel, I mean, they have these graphs about how it's better than
GPT35.
You know, they released it not too long ago.
I mean, does anyone think that this is better, right?
I mean, I literally have it up there right now.
You can take a look at it.
But, yeah, I mean, who actually has.
It's not substantially better.
I've used it quite a bit.
You know,
Let me kind of disclose a bit of a secret here.
A lot of people are wondering what the secret sauce is from, say, 3.5 to 4, chat GPT and inflection,
and even models like Claude and Cloud 100K.
Some of the secret sauces are running multiple models at the same time,
and they're creating a sort of consortium of results.
Right.
And that's a secret of chat GPT4.
So that's why it's not really being disclosed how that technology works.
And the same thing we're dealing with inflection.
I believe that they're doing a combination of that and some other simple things to make their results about equivalent to four.
It's so interesting. I'm actually literally talking about it now. And it's, I don't know, it reminds me of like, you know, the first time I use medium versus, you know, other other sort of things. But yeah, anyway, super interesting. Yeah, Strange Loop. What do you think about all this?
Hey, Y, C, hi, everyone.
Yeah, it's interesting.
Somebody, I think it was Dan who was talking about data,
and I want to touch that point,
and I had some thoughts around that.
Yes, there is a recent, I think there was an announcement made,
or I just saw something on Twitter regarding, you know,
open AI have been getting sued.
I mean, these are bunch of, this is just one of many that will follow.
Now, the challenge of our times is like how to synthesize, how to get that data in a fair manner from the people who, you know, have contributed towards that, right?
If tomorrow, Mid Journey and Open Air and all, they stop getting any data streams, you're going to get worse and worse results over time because...
it's not just about static data.
It's also dynamic.
It's changing.
Our world is changing.
The data is changing.
And the users who create,
they need to be compensated.
You know, initially the web,
you know,
all these big companies,
they did got away with,
okay,
yeah,
we can,
you know,
get away with just using data
and give them a service
and they're happy.
But I think things,
dides are going to change
with this,
depending upon where these lawsuits go and how users data would be used.
Artists would be pissed because their livelihood is kind of at stake with this kind of a thing.
It doesn't mean it's like the holy grail.
You're just basically using their data so they need to be compensated in some way.
And so yeah, that's what I had to say about that.
And so that's another interesting question that's coming up on the horizon as to the secret source is also the data they are using, right?
The data sources they are using apart.
And how do we spread the pie, right?
Or how do we divide the pie, right?
Because we all own the data, but it's these folks who are profiting from it.
Xavier, and then we'll go to meta.
Yeah, do you hear me?
You can hear you fine.
Honestly, I'm just going to be frank about this.
Inflection is just a bunch of different use cases.
I mean, I honestly think, I mean, I hear what Microsoft did this, but I think it's just
it's our own version of a hallucination in this situation.
Because if you really think about it, it talks and all that, but you think,
essentially if you train any LLM model to this day it does all the things inflection
does there's nothing different the only difference is they have enough marketing and
advertising money just to make it seem like it's just like it's just the shit it's really
if you really look at it and you really break it down and you go to that processing you can download
a Falcon you can
You can download any of these new find-toe models off of GitHub and Hugging Face, whatever,
and you'll get the same results.
Or even if you combine star coding with Falcon or some other LLM,
is you're going to get the same inflection results.
The only difference is the package deal.
It's how they package it.
It's how they market it.
Right now, I only see two benefits of what Microsoft is doing right now that makes it so much worth so much more.
It's the way they market it, the way they package it, and who they're advertising it to.
They're not advertising it to people who are technical because we know we're not blind to this.
We know what's going on.
We know there's a sauce, but we know what's really happening.
They're not they're going to get users that have no clue how tech work.
They don't they don't have no any and because they have the power to do that,
they're going to be startups who've already done this.
There's plenty of stars that are already doing what inflection has already done.
The difference is we don't know where they are.
We barely got to get known.
We've barely got to try them.
It's because it's a name brand.
It's no different than wearing a Jordan or a knockoff Jordan.
It's inflection is now a name brand that people just want the brand.
They don't care if it's, if there's another one that's just as good or a generic brand just as good.
It's just they beat everyone to the prime.
And it's purely because of marketing.
So I honestly see the future of most companies going to be purely marketing and data collection.
And I think that's, yeah.
Yeah.
Yeah.
So do we, okay, okay, this I think kicks up an interesting debate and I'd love to hear from MetaPrime and some others.
But basically like what you're saying is we.
That's a bold claim.
That's a bold claim. Metaphrine, do you agree?
Is this just all marketing?
There's no difference between models.
Yeah, no.
That was interestingly enough.
That is what I wanted to talk about is that I don't think this idea of data ownership,
there's no easy solution.
It's really messy, right?
Because everyone's been putting their stuff online, right?
And so it's openly accessible.
You have a lot of artists who have a lot of different, like, art chair things.
And the way that they've been,
kind of trying to
monetize their art
is not necessarily the art that they put out
but like commissions
or maybe they'll
sell you the more high definition
versions of those photos
or like some sort of print or something like that
so there's other ways that they've been monetizing
But I think it's really hard for you to say, okay, well, we've been training this date off, like,
avionart.com's, like, different, you know, like anime, girl stuff.
And it's like, okay, you know, how do we pay those people or whatever, right?
I don't think there's an easy solution.
I think it's actually going to be some sort of evolution in, and I really like what Xavier was saying,
basically marketing, right, branding.
right that you're getting the original mccoy from a certain type of artist right and so i think
like nfts are definitely going to play a very big role in this right verified ownership of digital
assets um but i i think it's it's just to me the idea of trying to you know pay people for
the data that they've put out i'm also very much of the belief that no one owns an idea
Right. So like all ideas kind of exist. We discover them. We don't create ideas. Really what we create is an abstraction of that truth, right, an interpretation and analogy and explanation.
And if we characterize it in a particularly potent and powerful way, right, that's what's going to differentiate us from other people.
And so I think, you know, one of the cool things about AI is not necessarily just the technology that is creating and the kind of efficiencies that this creates.
But a lot of the questions that it raises around like creativity, creative process, ownership, that kind of stuff.
But I do think that the path forward for a lot of...
a lot of creators is going to be like creating a brand, trying to do digital ownership through
NFTs, that kind of stuff as opposed to trying to like, you know, hold on to.
This is basically like trying to hold on to air or hold on to like, you know, clouds or
something like that. It's just you're not going to be able to maintain kind of ownership over
this data. It's just already too, too long, too far gone. It's already up there.
Sorry, I had a quick counterpoint or like a slight pushback.
So in the audio or like say in the music world, you know, if somebody gets sampled,
I mean, there is a big hoo-ha and there's a bunch of lawsuits sometimes and these days you're seeing more of that.
And I'm just wondering, I could see that happen.
Like it's.
you know, an artist that has a specific style, right, which is distinct.
And then you're outright copying it and then producing on a mass scale.
I think some artists still feel the, they're not fully sold out on the NFT,
you know, NFT train at the moment.
I mean, they still feel, you know, traditional way of art, how they, how they propagate that.
But it's like,
it's blatantly like just copying from them.
So just telling them that you can't hold on to your,
you know, your art.
I don't think that's like 100% fair to them.
Because we do that in the music industry too, right?
I mean, if somebody samples out someone's,
I don't know, there was recently, like there were some big incidents
where was that, I forgot there was some,
yeah, there was some big lawsuits recently.
Yeah, so you're talking about like the PROs, right, BMI, ASCAP, et cetera, and the equivalent of something like that on the blockchain.
But I think we could actually have our own entire space on just that and rights.
I actually plan to.
It's on the forward docket.
But I want to make sure we don't gloss over the idea.
So Xavier brought up an extremely interesting dichotomy, which is the idea that none of these models actually have competitive advantage compared to like,
you know, some, you know, some, uh, some open source checkpoint on Hugging Face versus like some of
these, but so I want to say like, do we, do we actually agree with that, right? So are these VCs
just don't know what they're doing, right? And they're just like throwing money around because
they're, you know, floundering or do they know something that we don't know, right? So I've been an example.
So Mid Journey, you know, basically forked stable diffusion.
And, of course, they put it on Discord, right?
So that's a unique way of distribution as an app.
But they did fork it and they did do their own things, right?
They started to, you know, on the back end, rearranged what people prompted and people generally got better results, right?
So, I mean, Mid Journey generally, you know, as a consumer particularly was like, you know, kind of handily beatings when Stable diffusion 1-5 was out.
Stable diffusion 2 and Stable diffusion 21, for example, did catch things up quite a bit.
But I think the average consumer, you know, who maybe isn't like the best prompt engineer, would say, you know, my experience using mid journey, you know, four and five, five one and now five two is just significantly better, right?
So I guess that's like the question, right?
Like, is it really just all marketing?
Because then, I mean, you know, marketing is health consumer companies for generations for years and years.
But now we're in a world where it's like, well, is it really what Xavier is saying?
All you need is marketing.
You know, just take your, you know, your favorite model from, you know, hugging face or elsewhere.
GitHub.
Yeah, one point that was missed is, is usability.
So I remember Sam Altman saying they did a lot of work on UX and usability.
And yes, that's another differentiating factor, I would say, apart from, yeah, the rest of, you know, their algorithms, the way they, you know, because if LLMs are predictive token models, if they are prediction models, I mean, what, um,
breakthroughs are there other than, say, you know, using transformers and which everybody else can use,
the other competitive edge is the usability, the way you use it.
And, you know, when it gives you responses, like some canned responses on certain things that it doesn't find a right answer to.
So it's, they've thought through those things.
So that's the U.S. is also a strategic advantage here.
I totally agree.
I totally agree.
Brian, what do you think on this?
Thanks. I really think the most valuable thing that we own generally when it comes to a digital realm is our context and our data.
And one of the things, projects I work on a great deal is intelligence amplifiers, which is local AI.
And this is the slurping up of all of your context, all of your information, all of your emails on a local encrypted environment.
And being able to sell that, if you desire in little chunks, is where your power and your individual value will once again arise.
The old model was, we're going to give you, quote, unquote, something for free, and then you're going to get advertising to support it.
That's basically internet one, two, three, four, five more or less.
I believe the new model is that as you see the value of your context rising, right?
and your ability to sell it in probably very likely a blockchain, you know, Bitcoin type of environment,
advertisers will pay you directly or companies will pay you directly for your information.
So we're bypassing the middle party.
And it also bypasses another problem, and that is how to...
how to get all of this information if you're an AI company without being locked out.
Well, you can find individuals that you want to take some of their context.
Again, under contract and very constrained and have access to it.
So you're now 100% in control of who gets access to your data,
how you put that data out there, and what you earn from it.
And if you kind of play this out, it's worth hundreds and thousands of dollars
throughout your life.
What is happening is a dismediation.
So model building is always going to be there.
So that's one aspect.
The other aspect is there are a lot of lawsuits that took place in the 1950s to the 1980s on public directories.
The yellow pages and white page directory, there was a lot of lawsuits that took place.
And even Supreme Court findings on what is a derivative work.
when there is a directory of information.
And one could argue that a lot of the information
that's being slurped up
is not very much proprietary.
It's directory information.
And it's very easily being accessed.
you know, legally with a legal context.
But, you know, derivative art, this is another aspect of it.
That also has a lot of Supreme Court findings.
And that comes down to what is a derivative?
How do we judge by that?
Now, we're going to live through this in the Supreme Court.
They're going to look at what Mid Journey and other visual models has taken
and say how much of that is unique, how much of that is enough to require somebody to pay rights on it.
I think that's going to be a very difficult prospect.
Right.
Because I would argue that almost all the music we listen to today is derivative.
You can go back to a root music that it derives from.
Yeah, actually.
And like the interesting thing is you say, oh, like, you know, I want a Taylor Swift song, right?
Does Taylor Swift get all of those Gen A.I. Taylor Swift's song?
Should Taylor Swift get it all?
Or like, how much does Beethoven, you know, quote, the state?
I know it doesn't exist.
How much do they have?
How much does Mozart?
How much did Chopin?
Right.
Like how...
It's like impossible to disentangle these, right?
The whole shoulders of giant thing.
And it does beg the question.
I do think blockchain does play a big role.
Like I said, I feel like an entire space on this.
I mean, we can definitely tease it now because it's so important.
And some of the coolest conversations I've had with AI folks is on this concept.
Yeah.
Before we dive further, though, I do have to say.
And by the way, I do want to go to the folks who haven't spoken.
Also, spinks, Naga and some others, Naga welcome.
But before we go to you all to opine, want to mention there's a purple button on the right.
So audience members, please do comment.
We've got a team on the back end looking at them.
We like your comments.
We'll bring you up.
Also, Mario's company, IBC, incubates and accelerates AI and Web3 companies.
So partners with the VCs and funds.
Yeah.
with companies for equity, zero cash.
If interested, DM Mario and his team.
So there's also Charxan style pitches.
We're going to do them more in the AI space.
I didn't want into crypto space yesterday.
It was really fascinating.
So crypto roundtable yesterday was a lot of fun.
So if you're interested, also send a message and don't forget to subscribe.
So, thanks, Naga.
I mean, you guys haven't jumped in yet.
What do you guys have for us on these interesting developments?
Well, you know, I wrote Soleim on, but the space has been glitching for me a bit.
There are periods of absolute silence.
But what I do really, I am really fascinated with is the intersection of blockchain and AI.
And it's something I've written about.
But because I have, I'm trying to hear everything, but I don't want to speak out of steps.
So I'm going to just keep listening for a bit.
Yeah, we can bring you down and up. I haven't heard the glitches.
If anyone else has, let me know or send a thumbs up or something if you've had those issues.
But, yeah, Naga, what's going on? What's going on on on your side?
Wow, I just have to say that this is so fascinating that this conversation is talking about this right now.
Because I've been listening to a lot of spaces explicitly talking about the problems between connecting like blockchain, open sourcing AI and making sure that it's like monetizable to a certain degree.
Lots of ideas like.
going around about what is it called royalties after the fact.
I think it's very fascinating.
I think I have lots of ideas about how the market might be able to monetize that in a good faith way.
Because I'm really nervous that a lot of negative...
influences might take advantage of open source creators and their work and basically ripping them off,
using them in their programming without their knowledge. And how do you prove it? How do you legally make sure that that work is that at AI creation,
can be rooted back to you.
And I think blockchain is going to be a huge factor
in like a legal context of linking your work somewhere
and making sure that it's sourceable to you.
All right, you know what? I feel like we're just going to shift gears to that.
We will have other future spaces on this specifically.
But, you know, there are, I mean, there's just an overwhelming desire to talk about this data
and how blockchain can be a solution.
I also, by the way, do want to hear people say the other side.
So do prep up for that because I feel like we got a lot of fans who do think that this is a good,
good solution.
But AGI, maybe do you want to tell us how this can help, you know, help things, you know,
the intersection of data and blockchain?
Yeah, so last year, we communicated with a BET, which is a Grammy-nominated Transpionaire and a Web3 founder.
And he's an artist, he's a musician.
And he suggested that we use TF IDF, which is to create a unique vector for every artist.
So every artist, all of the music of that artist, we will capture kind of its style.
And then we will be able with the blockchain to do kind of the rewarding for,
according to if we use the work of that artist.
So he was suggesting to use TF, IDF, plus the blockchain.
So I think it's a, it will be a good basis as a start.
And to make attribution.
Yeah, definitely.
Any other folks? I know that you had some thoughts on this specifically.
Sounds like Xavier does this well. I do you guys can jump in.
Yeah, if you don't mind if I jump in real quick.
I was going to say that's exactly what we actually, some one of the things, again, we were working on in our company.
It's just ethical approach is definitely when it comes to blockchains.
And we've, for the last year, actually, we've been diving into the thought of that.
Because, again, the original structure Web3, just it was a cute concept.
It just wasn't practical.
But,
But now, you know, what they call what they're called NFTs, I think they do, they call them like tokens now.
But essentially everyone having their own.
Ornals, right?
Just.
Oh, yeah.
Exactly.
Exactly.
I mean, it depends on.
Exactly.
Digital artifact.
Safe words.
Yeah.
Digital artifact.
Well, but I think honestly, if the investors of the VCs who may have not done did good in
2023 because of this invest in 2020, I think because of this investment, I think is going
to reap the benefits now because of AI because I think the only way this is, I think
AI is going to, because again, it's so everyone can have one.
I think the only way it's going to have its individual value is who like is the individual
own it, almost like a collection like, oh, I want to collect a Kelly Jenner's NFT.
Okay.
or I want to be a part of her language learning model,
or I want to be a part of Kim Kardashian's language learning model.
Actually, I want to be a part of hers,
because it'll almost be like you get to interact
with each of those individual models.
So honestly, I think the future of AI
isn't gonna be all these coding
and all these different use cases.
I honestly think people's personalities
will be the brand of these future AIs.
And we're already seeing it.
Again, we spoke in the last space about,
Again, a lot of people didn't like it, but we have to take the use case of the only fans, how she used it to make five millions a month.
Don't you think it's going to be due time where somebody just turns her picture into an NFT and that's officially hers forever?
No one else can use that.
And now she could just market her AI as hers and now everyone wants to speak to her.
Same thing Kim Kardashian can do.
Imagine being able to speak to Morgan Freeman tomorrow morning.
Hey, Morgan, how you doing?
Or even Snoop Dog, what's going on,
what's good, Snoop Dog, God, what's good, what's going on, baby?
You know, that I honestly think is a future use of AI.
You know, that is so fascinating.
I mean, you know, celebrities have had a checkered history with investing.
I mean, some of them have been great.
You know, Ashton Coutures, you know, is sort of one example.
But, you know, others have had issues, right?
Investing into hypey things without, you know, without a lot of return.
But, you know,
What you're saying is if attention is what matters, right?
If marketing is all that matters, then why not have Kardashian, you know, LLM, right?
Is your, I think your hypothesis, right?
What do others think about this?
Maybe meta and strangely, you know, you know, Dan, I mean,
I had a quick, yeah, quick response to.
So that's a good way to go like in terms of like personalization.
And we can get there not by cloud services, but yeah, high compute on single devices.
So that would be good if we can run LLMs on our existing devices.
Now, talking about blockchain, like anybody can take a screenshot of the image and then sell it on some other platform.
That problem hasn't been solved yet.
Yeah, you can create hashes of your art and that is a unique identifier, the identity problem on the internet.
not saying exposing ones or entry like pseudo identity that needs to be resolved i think dids are the
way to go decentralized identities and um should be algorithms that kind of um you know map out the images
uh and kind of make a signature of say a certain image is a certain signature and signed by um
certain artists like something along those lines are little more than
Just recording it on the ledger would be, I think, pretty good, almost like a watermark, but for, you know, done by algorithms on unspecific, say, art or images.
Yeah. Meta, what's up?
Yeah, so I think there's a lot of different ways that we talk about, like, how the technology works.
But in my mind, like that's not nearly as kind of exciting at this point just because, you know, it's going to happen.
There's a lot of ways that it'll happen.
But to me, it's about the culture that's going to evolve around this.
Because, you know, in my mind, I think a lot of people with this kind of,
opening up of the creative space.
You're going to have all of this kind of flood of, you know,
mid-level quality stuff or even like decently high-quality stuff, right?
You know, you look at, like, for example, I'm on Deviant Art,
and, you know, if you're going to go on to that site,
I warn you, a lot of the really good artists also have a lot of...
graphic content that you might want to avoid,
but there's also just a lot of
amazing AI art that's being
generated in that. I use that a lot for
my tweets and stuff like that, and I'll
even use mid-journey. And
one of the things that I see is, it's,
if you find, like, the really good AI
artists,
They definitely have like a style.
They have a way of doing things.
They have their own brand.
And so there already is kind of this ability to differentiate yourself inside the AI art space,
which I think is really cool.
But it's also about building this kind of patronage system, right?
This idea that people are, you know, excited about a certain artist,
creating certain things, right?
Like you're hyping yourself in a unique and different way.
A really good example of this in my mind is there's this YouTube guy, his name is Charles Bertrand.
He does bass.
He plays the bass.
He's really good at it.
But he doesn't just play the bass, right?
He has a lot of these videos that are, you know, kind of funny and kooky.
He does like, you know, he'll do like the skit thing and put on wigs, but he'll like do collabs with other creators and stuff like that.
And what makes him interesting is not necessarily his music, although that's fantastic and really high level.
It's about like the creative things that he does with that, the collaborations, right?
And that's what attracts people and gets them to like, you know, sign up for his Patreon and buy his music and stuff like that.
And so I think basically what we're doing is, you know, it's the bar to entry of, you know,
making AI generated art or creating AI generated music is being lowered. And so we're going from
the situation where, you know, the high quality stuff that's not actually like really well done or
interesting or stylistic. Like the, the,
in my opinion, most like mainstream music is just absolute freaking garbage. Most mainstream like modern art, absolute freaking garbage. And I'm like really happy to see all these people complaining about the fact that they're kind of losing their ability because in my mind what's going to happen is you're going to see this kind of rise of
truly interesting, expressive, stylistic, creative stuff because the bar has been lowered and,
you know, the ease of distribution. And what I'm hoping to see is that the people who actually care about
good quality,
artwork,
style,
all of these kinds of
things are going to
then patronize.
And so you're going to have
the shift in culture
from kind of
liking these kind of mass,
you know,
big celebrities.
You know,
I actually really detest the idea
of like people being excited
about talking to Morning and Freeman
because in my opinion,
like I don't think we should have
celebrity worship.
I think to them,
again,
it's like this,
they were kind of elevated into position.
I much rather the people
who have interesting things to say
or interesting like
styles and things like that, those are the people that we kind of gravitate towards.
So Matt, I agree with your sentence, but basically you're painting a picture of a world where
like, because we're voting with our wallets instead of with our thumbs, with likes that
the quality should go up. Is that the core argument?
effectively. Yes. Yeah. Because I think a lot of the people who are kind of elevated into those
spaces, you could argue that what they're doing is kind of catering to the lowest common denominator.
right there they're they're digestible so i think being a classic example is you know christopher marlowe
it ain't great compared to shakespeare right i think we can all generally agree on that but he was
popular if not more popular than jakepeare right to the lowest common denominator i mean he's actually
not bad but we only remember him today because of his association with the bard um
I mean, isn't it true that, like, I feel like, you know, Herman Melville died thinking that Movie Dick was a failure.
It was only a few years after his death that it became popular.
I just feel like a lot of these, it's like time tells, the time usually tells over decades, not, you know, not in the moment, right?
I think people often don't recognize what's good and what's genius.
Sometimes they do, but often they don't, right?
Yeah.
Sure. And I think that's a good point, but I guess the more pertinent example that I point to is more like, look at like the recent Indiana Jones movie or something like that. Right. It's like the studios are really trying to churn out these really safe remakes that are kind of made for modern audiences and they try to hit all of these certain box demographics and stuff like that. And
you know, they're losing hundreds and hundreds of millions of dollars on these projects, right?
And what I think you're actually going to see is the rise of kind of more independent filmmakers,
more people who are creating...
better stories, more interesting stories, because, you know,
the Hollywood has all the latest tech and tools and resources and huge budgets to play with.
And then they're, you know, producing these absolute flops, right?
Because of the over bureaucracization of these systems and the desire for these,
you know, for the leadership of these big companies to do things that are safe and profitable
and that seem kind of trendy, but they're actually...
horribly missing the mark. And so I think we are kind of on the point of this pretty dramatic shift.
You know, I hope, along with you, that quality can be discerned, and I hope we can usher in that era.
Spanx, welcome back.
Let us know if you can hear.
Can I just hear you?
I just wanted to say, I want somebody mentioned issues with pseudonymity.
And I wanted clarification on that, please, and what the proposed solution was, because you can actually, the blockchain does identify.
what is real and what is not real, what is not authentic.
So it's not just gobbledy goop.
It's actual, it's pretty technical.
So what is, hold, what I was trying to get at.
Sir, let me finish.
Thank you.
So, yeah, if you could just explain that and then what your issue is with it,
and then what you said your solution was, thank you.
Yeah, so what I was trying to get at is the image that is, say, created in one blockchain ecosystem, right?
You create a specific hash, you have an ID attached to it, maybe a decentralized ID.
Then somebody can go and have the same image at some other blockchain.
So there should be a way to kind of evaluate intra-blockchain where that flow of that asset is going, right?
Where is that NFD?
being sold or bought and swapped.
And so I was trying to get at more a deeper sort of aspect of that.
But I understand what you're saying, but that's a whole other issue, okay?
So you're talking.
Yeah, but there's an issue.
Okay.
I'm going to let you finish speaking and then I'll speak.
Yeah. So the second part I was mentioning is the the way you can even further make it more solidify it or maybe make it more robust is having some of some sort of algorithms that do a signature on an image because if image is a bunch of pixels, you know, the way the pattern in an image is, somehow that could be turned into a unique signature.
is what I was just trying to kind of just get that idea out.
But so those were the two points.
So that's why I'm saying like it's not 100% on a blockchain.
Yes, you can trace the providence,
but then there are ways to circumvent that is what I was trying to get at.
Quickly, since I've given a little bit of time.
Actually, I would like to comment if that's okay.
Okay, go ahead, go ahead.
You're actually quite mistaken on the first point that you made.
You're talking about inter-blockchain providence.
There's no such thing at the moment.
First of all, blockchain is very new.
Okay, the way we identify something as being authentic is if it is the authentic piece within that blockchain.
If I create something and somebody goes and creates and this has been done, okay, there are lawsuits.
pending lawsuits if somebody goes and recreates that or calls it a derivative somehow but i think
it's a basically a recreation let's just say it is a recreation and puts it and mine is on
ethereum and they put it on solana guess what
That is not stealing my identity. That's not pretending to be me. However, I may have a case against them.
So what you're talking about does not highlight the weakness of pseudonymity. I want to be very clear about that.
You are talking about other issues. You're talking about interoperability, interconnectedness between different blockchains, and that is a completely different matter.
Yeah, but it is relevant, right?
I mean, I'm in this space too.
I've interviewed a bunch of folks in the industry.
And this is definitely an issue.
It's relevant, but it's a completely different issue than identity.
And putting a thing on it, like whatever you said, like a watermark on it,
that will not change anything with respect to a different blockchain, nothing.
Yeah, that's why I said there are two separate issues.
right like one is the issue that it was a problem with the pseudonymity doesn't get the job done
and then people can uh uh take other work and present it elsewhere as their own and i'm just saying
that is not an issue or problem with pseudonymity that is simply a matter of moving something
from one blockchain to another and the person who originally did that on the original blockchain
probably has a case
But because blockchain is still relatively new, there haven't been enough precedence in cases right now in the courts for us to say, well, look at case, case, case, this. That's all.
Yeah. So if I'm an artist and I churn out a lot of images and, you know, there are a lot of people copying that, then it becomes on scale this problem. Like, how do you track that? That's a big problem, right? So I was just trying to get at.
that the fraud problem is not completely solved in the space.
There is a major, major, major problem of frauds in the space.
But it's not the one that you're saying.
I'm just going to tell you that you're right.
There is a lot of fraud, but people can copy.
absolutely I have created artwork anyone in here can go copy can go right click and save I have no
case against them okay now if they want to monetize and put their name on it as their own and then
monetize then I may have a case okay why because I live in on
on planet earth and planet earth i live in america and there are certain rules okay it's not
that blockchain doesn't happen in a vacuum it happens within the rules of society so i'm
with you but yeah but i'm trying to just trying to explain
This sort of thing, people do do that all the time.
But if they are monetizing, when you commit certain actions, right, then it's like you're liable or you're not liable.
So people right-click save all the time.
You can't pseudonymity is not supposed to protect that.
So can we move on?
I'll let you respond.
We do have to move on to the next one, but I'm going to make sure we finish out this discussion because it's an interesting one.
Maybe Strangeloup, and then I want to give Meta Prime a chance to...
respond because it sounds like you have some input on this but yeah feel free and then we'll go back to
you so i'll be quick you'll see so yeah sphinx so if you're an artist for you the biggest
problem going forward would be detecting where your artwork is getting copied right like if if you're
worried about that sort of a thing
is what I'm trying to get at.
Detection is also a huge issue.
And the bigger the ecosystem gets, yeah, social proofing,
which I think Meta had talked about earlier,
he was trying to point to like, hey,
people in the group, this community, this ecosystem,
they can call out, hey, this guy is a fake artist,
or he's copying from XYZ.
We're not going to buy his art for X amount of money.
But there are suckers out there that will buy, right,
and on other platforms, not knowing who the original artist is.
So I was just trying to get at that the detection becomes like a huge issue.
How do we make sure that that comes to the surface or somehow there is like an additional layer to it for making sure it's coming from the real artist?
Meta Prime.
Mike's Jorris.
Yeah, I mean, I don't have much to say.
I was just kind of thinking that, like, again, I don't think that the exact how of this, from a technical standpoint, is all that important.
I think it really revolves around the culture of patronage, right?
People finding artists that they really like, caring that it's authentic and different.
I guess an easy way to think about this is like furniture, right?
right there's a lot of people that are just going to buy whatever couch or sofa or chair they don't
really care who made it it's not important to them it's it's a functional thing right and i think
it's hard to necessarily see art in that way but like for example if you're putting up a website and
you just need graphics or whatever you're you're not going to pay someone like a premium because
they're some special well-known person you're just going to get the art that works and conveys your
message effectively but
There is going to be a subsect of the population that's going to care about kind of the style and, you know, all of those kinds of things.
And they're really going to gravitate towards a certain person.
I think that the way that artists will have to build their brand and kind of promote this idea of patronage and the importance of it, it's going to be different.
And that's really the shift that we should be looking for.
And the technology will facilitate that, whatever the culture shifts to.
Yeah.
I think it's, I think it's so fascinating.
Speaking of culture and art, we...
I'm sorry, I just want to say one thing.
Yeah, yeah, thanks.
Thanks.
Thanks.
I'll let you have a chance to respond, but I want to tee it up.
John, just prep up.
I know we skipped over you, but you can help tee us into art after Sphinx's last comment on this matter.
So, yeah, let us know what you were thinking on that.
But, Sphinx, why don't you give us one last on blockchain here?
Yeah, the reason it's important to discuss the technical parts is because this...
is a technical matter.
Blockchain is technical.
AI is technical.
Okay?
So we can talk about society and culture and all that stuff.
But if the argument is based on a premise,
on an assumption,
on a technical thing that's incorrect,
right?
then that needs to be said because we can talk about this is not about philosophy okay what is
incorrect technically okay can I finish sir so what I wanted to say about what was said was
this concept of what is the artist's biggest fear of having their art stolen that is not
with inside Web 3 that is not the biggest fear an artist
Anyone can right click save something.
I don't know how hard that is.
That's been done since the beginning of time.
That is not something that was created when blockchain came in.
And as an artist, and people have been doing this in Web 3 sent for...
ever since I have known is that as an artist, you know if you have a limited quantity, you know
if you have an unlimited quantity, you make sure, and these are problems that come up, you make sure
you say, only go to this link, only purchase from this link. But can people write, click,
save, and get away with it? Yeah, I mean, there's no, but for an artist, that's not the artist's
biggest problem at all.
at all because that's just ridiculous. If I create something and then someone else goes and creates
a copy of it tomorrow and says it's theirs and makes a couple bucks, I mean, it's just not realistic.
Okay, folks, I know there's a lot that people want to say.
And by the way, I love this intersection.
I mean, I've been, I mentioned in the scripted spaces.
I've been in Bitcoin since 2013.
So I like love, and I'm an artist as well.
So this is deeply interesting to me.
I think we need an entire space on this.
But do you need to move to the next bit, which is unity and gaming.
And some people call games, video games art forms.
Some people don't.
But John, maybe you can maybe transitionist there a little bit.
Whoa.
And just go forward.
I think that, I think that,
And I want to push on the philosophical side.
I hope Brian comments on this.
To me, the artist's biggest fear is replacement in a very strange way.
And before I get hounded on that, I think that generative AI and AI in general is unlocking a vast potential of creativity.
And it's no longer the domain of the craft,
but it's shifting to the domain of cognition.
And this is just fundamental that I now can create a video
I now can create a beautiful portrait based entirely on cognitive applications.
In other words, managing the prompt.
So I think that that is really a very, very interesting thing.
I don't need to be a fine artist in the sense of brushstroke.
I need to leverage technology in the same way that many artists, camera obscura, the Lucy machine that Norman Rockwell used.
So that's one of the areas.
And I think that's what we're moving to, very interestingly.
I think it's great.
Let me tee up this discussion.
And by the way, this is going to be a much bigger debate.
So perhaps, I mean, we're also working on some big guests in the gaming space in the coming days.
If you all have suggestions to, of course, DM me, let me know.
But so, Unity launches, so I put this in the nest, this video is wild, right?
So anyone who's done game development, if you watch this video, I mean, you need to launch
Muse, which is an AI platform.
So it's basically Unity's chat GPT moment to sort of bring it down to colloquial matters.
You can effectively have this AI tool, you know, or like GitHub co-pilot or things like that.
You can use this tool and it can do things like animation, it can create textures.
For example, they have a video portion where it's like you have this, you know, skeletal mesh and you say, hey, you know, do a backflip and it literally just goes and does a backflip.
I mean, it's wild stuff.
I mean, it's like next level things.
You know, we had people like a program in Unity using ChatGBT, and they made, you know, Flappy Birds clone in two hours.
Now this AI stuff is built into Unity.
Unity has been a company for years that has focused on data, right?
And data has been the biggest thing.
Even more than its competitor, Epic Games is Unreal Engine.
Data has been a big part of Unity forever.
And John Rickatello, very famous, a long-time CEO now of Unity.
He's been in the gaming industry a while, has been focused on this for quite some time.
He was part of EA before, which is behind the Sims.
So, like, what do people think of these developments, right?
Like, where are we going?
And like I said, this is more of a teaser for a bigger space to come on this.
But yeah, love to get people's feedback.
Brian, I see some thoughts from you.
So what do you think about this?
I think it's a great move forward.
I think game development is always going to find the least resistant path for rapid deployment.
And we're going to see a whole new rise of even more advanced.
NPCs are going to become less of a thing.
Let's put it that way.
I'm working with some groups that are,
trying to not only use AI for development, but use AI actively in the game.
We can get a Lama model now down to 4 gigabytes.
And whether it's a cloud-based game or whether it's a local, you know,
executed game, 4-gabytes is a very small footprint to give every NPC's cognition
and recognition within the game.
Yeah.
It is a tremendous shift.
Low-hanging fruit, right, Brian?
I mean, imagine just like the Sims with, you know,
each NPC enabled by an LLM.
It's going to redefine what a quote NPC is, right?
I mean, we're all going to be NPCs, right?
This entire world is.
Yeah, we are.
We are NPCs.
It's going to be mind-blown.
The early games I've been seeing is this amazing, absolutely amazing.
Yeah, that's so interesting.
I mean, do we think, I mean, do you feel like any of these games are going to be good?
I mean, it's hard to make great game design, right, despite however many powerful tools we are.
Like, I mean, what are your thoughts on that?
I think we're going to see really enjoyable and usable games.
I think we're going to reconceive what games are.
I mean, every...
Every three to five years, we're seeing sort of magnitude changes within a video game environment.
And it usually comes to faster and higher resolution.
Now I believe it's going to becoming much more deeper and richer experiences.
And if the game itself is in fact an AI, right?
Let's conceive of this.
And it's an AI in two levels.
It's a textual AI that's building the landscape, you know, let's call it the scene in real time.
Based upon where you've already come through the game, you're literally at that point in an endless game.
Maybe there is some master mode that is built in to take over at some point, but other times it just might be an endless journey.
That's one side. The other side is using generative and, you know, diffusion type AI like a mid-jurney to constantly create new scenes, to create new environments.
So those versions of your game may become yours uniquely and get this, you might want to share it.
Very much like what happened in a Minecraft world where people just really built and created and shared,
you can start doing that within side of video games.
So the things that your unique adventures could be shared to others and maybe monetized.
So it's a very interesting time.
Yeah, we're going to have to start rapping soon.
I know we just got into this, but let's hear a few more.
So Chiptos, welcome to the stage.
Thank you, thank you.
Strangely first, and then we'll go to you, Chip.
Or, I mean, Chiptos, if you're ready, go for it.
Yeah?
No, no, you can go the other guy first.
What do you think of?
Okay.
What do you think?
Yeah, I agree with Brian.
I mean, gaming is interesting.
And, you know, you have the graphics.
You have the GPUs to power that.
And, you know, having some conversational sort of smartness in there.
That will be interesting, make it more interesting.
The, I think, I'm not sure, Brian, were you hitting on the blockchain aspect too?
But I had a few comments regarding that, that the only problem I see in the space with the blockchain related assets is,
is intergame assets.
Like if you have a Metaverse or like you have a, like say, gaming companies,
they want to contain all their assets within their games, like these studios.
So if they start interacting with each other, like, say, Blizzard interacts with Activision,
you know, EA sports, then you can see this whole new world open up for like,
intergame assets and like the whole economy kind of opens up but might be a time before we see that because you know game the game studios they spend three years and a plus millions of dollars on on a game and they want to get their returns so i'm not sure how that will work and that excite that is kind of exciting
to see.
AGI Naga, sounds like your perspectives on that interoperable game.
Well, this is very interesting.
This is extremely interesting.
So this is a chat GPT moment.
And I want to say that we will use that.
And we will also deploy, you have Play to Earn on the blockchain.
You have Plato Earned games.
It's very interesting.
But what I'm working on, it's agent that are learning to Plato Earn.
So we will deploy AGI agents.
that will play in that kind of game
and generate kind of NFTs
when they reach a kind of a reward
or some kind of score and so on.
So you will be able to play yourself
and to play also with AGI agents
in those game that will be,
and the environment of those games
will also be evolving according to the kind of player
that you are. So if you reach...
some kind of level in the game, the environment will adapt to where you are in terms of skills and so on.
So that means that it will always be some kind of new things and so on.
So the possibility is enormous and it's amazing.
Yeah, you know, and Chip does, I'll go to you, and then we'll go to the rest of the hands.
But, like, the thing about this interoperable thing is, I mean, I'm also not convinced
that it's going to be true.
Maybe there's a token that will be interoperable that could happen.
Like, for example, can you imagine if EA made EA chain and everyone built on top of that,
right?
So you could sell your sword from World of Warcraft or whatever, interoperably go into, you know,
some EA game and then buy some soccer ball, whatever, some soccer skin.
I mean, you know, maybe it works better than things like Fortnite where it's all cosmetic anyway,
though you could argue that the ariana grande fortnight skin like is brighter and shinier so it's easier to see so
though you can.
it does convey some some advantages in the game itself but lots of questions right so um chip test what do you
what do you have for us on this and then we're going to the rest of the hands yeah absolutely uh in
terms of like plater and i'm not really entirely convinced i think at the root of it it's going to be
what is going to create the best user experience for the gamer
and like you say, is that going to be some sort of, you know, a domestic chain where, you know,
the World of Warcraft auction house is now a blockchain and I can, you know, trade it across
all of Blizzard stuff and I can track that with my Blizzard account.
I think something like that's realistic.
I think something like an epic blockchain is also realistic where everything is in-house because once...
If they let these assets in the wild, I think they're leaving a lot of revenue on the table.
Like, I personally don't think Fortnite would ever, like, put their skins on the blockchain.
Like, why would they allow their users to resell their skins when they can just keep making new ones and keep, like, keep the faucet on pretty much?
that's kind of my thought.
Yeah, Tim Sweeney doesn't seem to be,
Tim Sweeney doesn't seem to be the biggest
crypto fan, by the way, too, right?
So there's always that.
And he has his crusades for better or worse.
Yeah.
All right, we're going to have to start wrapping soon.
Really interesting stuff.
But, you know, maybe it'll go,
Xavier Naga meta.
I know GPs just joined,
but we're just wrapping a really great conversation
on gaming.
But yeah, why don't you guys go for it?
Xavier, what do you have for us?
Yeah, Navidia, the trillion-dollar GPU industry, as we already know, and every company from OpenAI, Microsoft, all use Navidia.
And if I'm saying that right, sorry.
But when they did a launch, they showed.
They demonstrated an actual gaming environment from a spatial, you know, VR or VR virtual reality that they had the NPC interact with the user.
But see, this is the catch.
It has the same mission.
But it interacts with every gamer and says something different.
You know, in the games right now, when you go up to an MPC, it will say, oh, I have a journey for you.
You got to go get some coins.
You go get some coins, bring it back to me.
I'll give you this treasure chest.
So...
In this situation, well, now everyone gets his own individual user experience based on your interaction with that NPC.
But it may end up being that same mission.
But if you watch the video, actually I posted it underneath in the comments.
It's pretty incredible.
And I think gamers in the future are going to have NPCs give every gamer their own personal experience.
Yeah.
Yeah, just to add on that real quick, like, I think the first place we're going to see it is in, like, machine learning boss battles, like, where the game is going to get harder based on, like, how your play style is.
And this is going to make, like, almost like the perfect experience for, like, how I play the game.
And then the boss or the game is just going to get, it's going to get harder based on how I play, not just like some linear path.
Man, I feel like boss battles are hard enough.
But, yeah, it's a fascinating development.
So, Naga, meta.
What do you have for us?
Yeah, so I think that talking about the gamer's experience of the game is very important,
and I'm not going to discount that.
But I think this, what EYC posted in the Nest, the biggest thing that I'm seeing from that is from the engineer's perspective, from the first person making the game.
This is, I think, the most important thing that we need to be focusing on in this time period is that these...
companies are absolutely going to cut costs, that that is just a function of our current system,
how companies work. We know that's going to happen, and that means fewer jobs.
So we're really going to have to get ahead of the curb of this and make sure that these designers,
these game engineers, are...
prepared to make their own companies essentially.
I think that it's going to condense everything down to fewer people are going to
be necessary to make a beautiful game.
And they're going to have to basically become their own business person to still monetize
their art or their engineering because there are not going to be positions in the
higher, in the bigger firms.
Naga, I mean, I made both animated films and CG games,
and the games are so much harder for a lot of reasons that we don't have time to dive into right now.
So I'm so glad you said that.
Ample room for another space.
Meta, and then I think GP, you're going to take us home.
And if you guys don't mind, keep me a quick, we do have to wrap.
So meta, what do you have for us?
I'll be really quick.
So, for example, there's this game that came out or that's being developed.
I think it's called Project Loki or something like that.
It's from a bunch of developers who used to work on.
and legal legends, right?
And I'm kind of bringing this up in response to the idea that, oh, you know,
why would companies create like blockchain tradable systems, you know,
when they're basically printing money off like all of these things?
Well, in my mind, it's kind of an evolve or die situation.
Right. Legal Legends, in my mind, made a really big mistake by not trying to create some sort of like forge functionality, some sort of way to kind of create and innovate inside the game. And it's had a lot of success because it's kind of maintained a very strong control. But the fact that like Project Loki is coming out, it's kind of this...
League of Legends slash Super Smash Brothers search slash Apex Legends kind of style game,
to me, that looks like kind of the next evolution.
And so I guess the thing that I would say against that idea of like, well,
companies aren't going to do this because it's not profitable.
In my mind, well, if you stagnate your innovation, you might be able to kind of keep that
profitability train going, but at some point you will be out of all.
That's super great. I think GP, you came late, but last but not least, you're going to take us home today. So what do you think about all of this that's happening?
Yeah, I'm going to actually, because I haven't got context, I just put the hand up for a quick of pause for the late show, had a run over on a meeting, EIC, so Paul's for that.
I'll see you Tuesday. I've made a post about the latest NVIDIA titles or G-Force titles coming in July and the processor upgrades available and also what's available under the...
various membership plans.
So yeah, big stoke for July,
lots of gaming releases,
short and sweet for me this evening,
Paul's for the late show.
I think it's great.
Well, everyone, you know, we did a lot today.
We talked about a monster fundraise.
We got into an unplanned topic, but one that I think is right for an entire space in and of itself.
So we'll look at that in the future.
Blockchain and legal and art rights.
And, of course, we just barely teased how game development is going to transform entirely with Unity's, you know, first 4A.
