Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
DECRYPTED: Episode 22 - The Ai Takeover!
Recorded: March 11, 2026 Duration: 0:50:07Space Recording
Full Transcription
Good morning, good evening and good afternoon
To everyone joining us around the world
It's decrypted time
It's an hour earlier because the americans for some reason decided to do daylight savings
A month earlier than everybody else
So we're a little bit late
Jack
What's your guys take do we get rid of daylight savings
entirely get rid of it if you could just wave a magic wand or make it the later time or the
earlier time 30 minutes split the difference split just do 30 minutes like back or forward whatever split the difference call it a day
we're done please that's the idea i've heard yet yeah i mean for me i don't really care so much
because like as a as a software engineer you're flexible so you don't really have to get to the
office at the same time i don't know but uh i think in general people need to only work when the time
when it's like the sun like sunlight there are some studies that you know you need to make sure
that it corresponds to the natural thing that if you work on the outside if you work and it's kind
of dark outside then you're disrupting your sleep. So that's my take. I mean, like when people like before civilization,
people just work when it was sun and it was sunny
and then they slept when it was at night.
Candles were expensive.
If you wanted light at night, you couldn't get a lot of work done.
Yeah.
Like, you know, Jack, you know that I'm a really fan of this guy,
Huberman, the neuroscientist from Stanford.
And then he says that, you know, during the night,
even if you go to sleep, you're sleeping,
and then you go to the restroom during the night.
And in the restroom, there's lots of light.
You usually disrupt your sleep.
So even a small amount of light during night is bad.
And vice versa.
During the day, you need to get lots of light.
So actually, I'm trying to do it.
Like now I walked from the office home so I can do this podcast.
But I walked.
It was a really beautiful, sunny day in
Lisbon. Yeah. Yeah. So one interesting thing, like the state of Nebraska had this, they were
like going to have a vote about daylight savings time and they're, you know, and there's, they're
like, okay, should we get rid of it? We'll be like Arizona. They don't have it. Or should we move to
this early one or the late one?
And the other thing is there's a, there's a time zone change in that halfway through the state or like a third of the way through the state. So there's an hour off anyways. And so then they're
like, okay, now we're going to really screw things up. And basically the funny thing is they couldn't
come to any consensus because everyone has a different opinion. But Falcor, I think if the
entire world agreed, like let's move to the 30 minute in between because 30 minutes you get like you know when
you're doing meetings with people in india and they're on like this 30 minute they're like 30
minutes off and you're like hold on what the hell like um i don't know man maybe we're talking too
long about delight savings but it's but it is a pain in the ass like it genuinely um it really really is uh well um it
was was it the war well daylight savings came in in like 19 like 1918 right I heard it was like
for farming because you wanted you know farmers wanted to be able to you know they're losing they want to
work in the morning and they wanted to be light earlier so they could work earlier and then light
later so they could basically have work hours like during our and school hours during hours of
sunlight i think we actually had this converse exact conversation like eight months ago in like October, November.
Like it must have been like episode like five of Decrypted when this happened last time.
I'm remembering the conversation because I think I looked it up.
In that case, let's hit it to the topic.
Let's hit it to the topic.
Okay, go for it then.
Come on.
AI has been crazy, man.
AI's A16 crypto dropped a massive report on the top AI apps.
Whether they are top is up for discussion.
Real brain cells have been mapped onto little chips and things.
I saw them playing Doom, which was cool um moltbook has
been acquired by meta also known as facebook also known as evil uh and um there's there's so many
things going on in the world of ai ai is taking over i'm slightly blurry and let me let me give
one little caveat here we talk about crypto and ai you look at Scale Labs, we're living at this intersection.
So we'll thread in some topics.
People are like, why are these guys just talking about AI?
Well, there's a lot to talk about when we talk about agents and these apps
and the way stable coins are going to be used in X402.
So it will be relevant to you.
Regardless of what your interest is with AI or crypto.
Hell yeah, hell yeah.
AI, crypto, the future of the world.
So, Jack.
Here's the thing.
Let me give everyone a quick overview.
Then Stan had some really good takes.
I want to get into those quickly.
You look at the top 50.
Andreessen Horowitz, A16Z, announced top 50 Gen AI web products and mobile.
Number one on both lists was chat GPT.
Gemini number two on web.
CapCut number two on mobile,
which is a video editing product.
Canva was number three on web.
Gemini number three on mobile.
Anyways, you got your usual suspects.
Chat GPT, Gemini, Grok,. Anyways, you got your usual suspects, ChatGVT, Gemini,
Grok, Claude, Perplexity. And then there's a bunch of kind of interesting weird apps that
either have to do with like learning, social media, you know, making your social media content better
and, you know, home stuff. Like, it's just, but, you know,
it's a little underwhelming, these apps.
But Stan, what are your thoughts on this?
You know, actually, I spent last week
doing this really interesting kind of a
kind of a small project with our partners
on using AI in a particular way. We want to connect AI and blockchain. And the idea is a proof of concept is that you have a sensor, like a temperature sensor, and then the data from the temperature sensor goes to blockchain. And then AI reads this data from blockchain and at some point decides that your air conditioning needs to start.
So the AI then basically, again, through the blockchain, issues a command to your air
conditioner to actually start working. And so this is pretty exciting and we kind of started
looking into it. And what we realized is that this AI, you don't really need to have any program.
realized is that this AI, you don't really need to have any program. So essentially,
the AI can just know the prices of energy, know the model of your air conditioner, know
what apartment you have, but you don't really need to have a program. AI will just tell
the AI what temperature to keep and how much money you want to spend on energy.
And then AI will do the switching on and off automatically.
And it's very flexible.
So, for instance, your AI can control your air conditioning,
and then at some point you are not at home,
and then you don't really even need to tell AI to switch off air conditioning. If the AI is smart, it will automatically kind of put air
conditioning at least on minimum. So basically, we're doing this experiment, doing lots of thought.
And basically, I kind of took this example and wrote a blog post about how AI is going to work
in the future. That's what I'm arguing in this blog post
is that it's not that AI needs to write a Python program
that controls air conditioning.
AI can directly control air conditioning,
and it's a big conceptual change.
For instance, now if you look at Claude, right?
Claude will write you a Python script,
and the Python script will control your air
conditioner. It's not actually needed. The AI agent directly can actually issue commands.
So I wrote this blog post, which basically says that there will be no programming languages.
They actually not need it because AI agent is like a human, right? And a human can just switch
air conditioner on and off. There's no need for a Python script.
And so I published this blog post.
And now we're actually getting a concept ready for this very simple example where you have the temperature sensor, you have the air conditioner, and then you have AI that optimize the things.
optimizer things. But then, actually, if I answer your question, I was working to this podcast today and I came up with this crazy idea that you can actually have something which you could call it like any app. You don't actually need to have an app for every use. You could have like a morphing self morphing app. Imagine just having a single app on your
phone, and you just tell the app, I want to play a game. And it actually displays this gaming UI.
And then next thing you need to tell to the app, I want to be like change a photo or modify a photo in an app cut.
So I came up with this like any app.
It's like a morphing, a morphing kind of.
Maybe I call it morph app or any app.
And so that's actually where I'm going with all this research
is that there's no need for Python.
There's no need for programming languages.
There's no need for app.
If you have this morph, it's almost like
in a Terminator 2 movie where you had this Term it's almost like like you know like in the terminator 2 movie where
you had this term yeah different things you could have an app that you just tell your app to to
become cap cut or you tell it to become that's kind of really i guess yeah i guess my only
concern with that and jack i'd love your take is brand affinity brand loyalty and also trust of application would you use an all-in-one
app like kind of morphing app or would you still be happy with kind of jumping to your individual
thing like stands i was looking at i'm trying to get a theme across these top apps, okay? And the themes I'm seeing are like one core AI utilization.
So you look at like Daobao, which is kind of cool.
This is, it's actually owned by ByteDance, who owns TikTok.
And it's just like all about like AI writing assistance,
smart translation, AI image generation, document analysis.
They're just trying to make it very businessy around.
And they're also trying to like
download our application onto your Mac
and they wanna get all your information too.
So be careful.
No, then there's things like these social apps
and kind of strange ones.
Like I think, what was a strange one called?
Janitor. Janitor, don't go to that website
we're not gonna give you the website url okay but then there's one called wink that is all about
home uh exactly what Stan's talking about so Stan if you wink's one of the top apps
but what it is is it lets you it plugs into all of your smart home products. And then these, these providers, like, let's see, they have commercial electric, ring,
I home, Yale, echo B, Eaton, Hampton Bay, Anderson, Canary care.
They've got all these brands that are plugged in.
And then the AI just like controls everything without you needing to have like, right.
You know, somebody writing all these scripts.
The AI just manages your, your lights, your cameras,
your air conditioning, et cetera.
I'm a, yeah.
What do you think?
Maybe I'm actually, I am too radical,
but it may be like a typewriter in the sense that,
typewriter is a technology that didn't exist
like 200 years ago. It doesn't exist today. So it was like a period of time when people were using typewriters. And it may be that the apps and programming languages are also like typewriters. And then all of this like Claude and like all these guys saying, okay, you tell Claude to write your app and it writes in Python,
maybe all irrelevant for the future because in the future, you may just have this morph,
which is AI, and the AI doesn't have any programming languages. It's like, I can go to
ChatGPT as an example now. You can go now to ChatGPT and tell it, like, I want to play chess.
now to ChatGPT and tell it,
like, I want to play chess.
And ChatGPT doesn't really display any UI.
It just starts in the text field,
just in text. You guys can try.
Just in the text, it tries to emulate a chessboard,
because it doesn't have any graphics capability,
just using text.
It just displays some very primitive chessboard.
Or you can go to ChatGPT and say, I want to play
poker with you. And it also
kind of tries to play, but doesn't
have any graphics capabilities. So again,
it becomes very primitive. Or
I try a different thing.
I tried today, like,
you guys probably know the Civilization
game. Many people played
in the past, like Warcraft, this type of strategy game.
And I said to ChatGPT, I want to play Warcraft with you.
And I also tried to do text-based, some primitive stuff.
But the point is it just doesn't have the UI capability.
It doesn't have the graphics.
But in the future, you can imagine this like any game application
where you go to chat GPT
and say, I want to play chess, and it kind of shows you like a chess board, and then
you say, I want to play whatever, this thing, that thing, and then it kind of on the fly
creates these experiences for you.
Without any programming language, it just shows you things, right?
To show things, you don't need to have a programming language, it just shows you things, right? You know, to show things,
you don't need to have a programming language.
So I'm thinking, I came up with this very interesting concept
that because at scale, we want to go like Wayne Gretzky.
We don't want to be in the today,
we want to be in the future.
Maybe we just drop all of the intermediate steps
and say there will be these applications
that will not have any programming languages.
And then there will be blockchain.
And the blockchain can be, just for these applications, for this new type of applications, for these agents,
it's just a way to store the most important information so it doesn't get lost.
So like in the example of this temperature sensor and the AI agent that controls the temperature, if you do it through blockchain, because blockchain is basically like a reliable database, very reliable database, and that's it. There will be no programming languages,
no nothing, no applications.
There will be this just AI morphing
itself into anything you need on the fly.
There will be no websites because
the AI will again morph itself into a website.
Then I can get to the conclusion.
Ben, click on this chart here.
You can click on it. Yeah.
Okay. So Stan, this is kind of interesting,
like to what you're saying.
One reason why like you go look at these top 50 apps,
it's kind of underwhelming.
It's because people just do what you're saying today.
Like Stan was saying earlier, like he just, he's like,
oh, I wanted to figure out which,
how to switch out my dishwasher or something or washing machine and took a picture. And it,
it wasn't an app that's like, use AI to find a better washing machine. You just use chat GBT
and like, look how much market share chat GBT Gemini have over everything else. Right. If you
combine those two together, it's like a power law right that's like over 90 percent i think i've
been propagandered too because on twitter everyone's like grok's the best and grok's so good
bro deep seek is beating grok i i genuinely think i've fallen under elon musk propaganda
because a lot of people in china that use deep seek I think that's even like Gemini yeah Gemini
has got Google reach yeah yeah I mean the thing is I think people people use systems they're used to
and when you're in X you're like at Grok and that gets a lot of use there but then the Grok app then
you have to be like for me yeah this pattern like I go to gemini i go to chat gbt and i probably go
to gemini more than chat gbt and but then i'm like oh yeah i guess i should go check out grok
sometime and it's like you know if there's not a like a real call to action like grok has that
little character you can talk to like these little anime characters like that's unique but otherwise
if you get like almost equal use but sometimes there's a model that sucks.
And then I'm like, like ChatGPT had this bad model.
So I started doing Gemini.
It was better.
And then I'm like, then Gemini is not as good.
And I start doing ChatGPT.
Like anyways.
Grok chat is bloody atrocious.
It is long and most of the time,
just wildly incoherent and incorrect.
Let's not say anything bad about Grok on this while we're live streaming.
Otherwise, our numbers are going to go to the negative.
We're in the super intelligence era, buddy.
Love you, Grok.
My multimodal AI actually listens to podcasts.
There's probably for sure it's listening now.
I love Grok. I always use Grok when I can. actually listens to podcasts and there's probably for sure there is it's listening now i love grog
i i always use grog when i can yes i have a choice use different products i think and then i use
growth okay so so hey here's something interesting like like let's look at this chart usually software
during these peak times leaders often just dramatically surge
ahead of other people we could be seeing like a power law come together even more so where
the only apps are chat gvt gemini uh claude perplexity you know and then yeah hey there'll
be some small things like if you already are in notion like our team uses notion yeah then you're going to use that because you use it for something
else but otherwise why download another app like stan says just use chat gpt and if that happens
the app market in these environments are just gonna it's already like emaciated
you know it just it just shows you like the chat window will be king
and that yeah okay anyways interesting stuff let's talk about the brain cells
that was wild huh brain yeah the brain cells was weird and i'm gonna need stan i'm gonna need to unpack this work yeah yeah because i don't get
it like i don't get how it works please educate people like what the hell is happening here we're
mapping we're using real brain cells we've talked about this on the show of like mapping neurons but
what are they doing yeah basically it's just just really interesting. I think there are several really cool things.
First of all, we see, again, it's kind of a proof that humans were created by aliens
because all of our technology, we end up copying how we were made by someone else,
probably someone more intelligent than us.
So everything we pretty much discovered,
including neural networks, is just a copy of what exists in our brains. So current neural networks,
something like LLMs that Google or whatever runs, or chat GPT or OpenAI, just copies of the same thematics that exist inside of our brains. Actually, very interesting. I read
today very interesting information that our brain actually has like three big computers.
One computer is called motor cortex, and it actually basically controls all of our complex
like motor movements we are doing. The second computer is visual cortex,
and it's all the video we are seeing through our eyes,
all of the video processing.
And the third one is prefrontal cortex
when our abstract thoughts and all of this stuff.
It turns out that all of the three computers
are almost identical.
The prefrontal cortex that controls
the high-level thoughts, the visual cortex that controls
video, basically, and then the motor cortex that controls our motion, pretty much the
same hardware, which is kind of really interesting, which shows that we're not so much different
from like animals.
And so it's almost like there's not so much more different in abstract thinking as compared
to like motor movements or like visual processing.
But LLMs are basically a copy of our brain, but they're energy inefficient because we're
using programs to essentially emulate how neurons behave.
We don't have neurons.
We have silicon chips,
and the silicon chips try to behave like neurons,
and it's actually energy inefficient.
It's almost like how to give an example.
You are simulating something else.
But then you can just have a real neuron and replace a silicon
tube with a real neuron, literally like
take like LLM and just put a real
biological neuron. And it turns
out this biological neuron will
just use way less
energy. So
it's actually way cheaper to create
a computer by taking
an existing LLM essentially
and replacing
software with biology, biological hardware.
Let's go in though. Like, how do you, do you like, I don't know,
if you guys remember Teenage Mutant Ninja Turtles,
like where Krang was this like brain that like was connected into this like
robot body.
Do they literally take like a piece of cells,
like a piece of a brain and plug it into a,
like a motherboard or is it like,
are they mapping some small components?
Like how in the hell is this working?
It's funny, but basically like a neuron,
you can just think like,
if you're thinking about like a child game
or kids game or whatever,
like kids electronic shop, you
know, neuron electronics.
And neuron is just this kind of electronic thing which has electricity coming in and
out, voltages.
And it's basically like almost think about like wire or something.
And then it has like incoming wires and outgoing wires.
And then basically your brain or LLM is just collection
of these things connected to each other.
And then basically if you ask a question,
there's like input.
So there's like some input wires where you encode the question
and it's just things and then output wires.
Actually the answer is there.
So that's kind of how the brain works or LLM works.
And if whatever you're using to kind of do this wires,
it could be a software, it could be like biology.
So literally it can take like chat GPT
and these guys are doing it.
You can take chat GPT and like with small modifications,
you can implement it using neurons live neurons and it will work the
same uh but it will use less energy because these neurons will you'll have to like feed them with
sugar whatever like serotonin i forgot how what's like what are the things they need but definitely
sugar how do they search where do they where do they host the neurons is it like on a chip
somewhere is it in a lab
a wet lab so basically it's like a petri dish so there you you have a cell it can actually like you know like people like again like a school biology lab you have a cell put it like in a petri dish
there has to be some sugar some stuff you feed it in some way then they connect inputs and outputs
to it yeah yeah yeah so they they know how to to it. Yeah, yeah. And then they find a way.
Yeah, so they know how to do it.
That's actually how Elon Musk,
very much the same way, like when you have this neural link
and how basically
Elon Musk,
they take your brain, they make a
small incision in your skull
and then they connect
metal wires to your neurons and read, basically, they can read your thoughts, and then they connect metal wires to your neurons
and read basically, they can read your thoughts, whatever.
They can like blind people, they can project the video on your brain directly,
so now they can use it to blind people who can see.
People with amputations can control artificial limbs.
It turns out, to make a long story short, it's crazy,
but our brain is just a collection of basically some electrical elements,
which are very simple.
They have metal wires coming in, coming out.
If you connect the wires in the right way, just you know the encoding, right?
It's almost like
if you have a display you have usb connection you need to encode things correctly so your display
understands how to show like video the same way if you know okay okay i have a question
because i this is okay i think this could be a very interesting question okay it's a two-parter. You ready? One, would you allow a USB port to be put into your brain?
Right? Like it can be hidden behind your hair or something like that. That's part one question. Yes or no, Jack?
Really? Fair enough. Stan? Actually, I would probably, I mean, the only thing is definitely needs to be tested.
I was a scary guy because I don't usually use new technology just because I'm easily
scared.
I'm super scared.
But after like whatever, many, many people use it.
I know it has been tested and it's healthy and like secure.
I would probably have no problem doing i'm gonna
be like a total luddite people are gonna look at that like my dad never used facebook and
social media he's like i don't want that i don't he didn't talk that way but he's like
but like i'm gonna be that guy like get those neurons away
oh god get that adapter out of here all right okay okay so stan then and jack maybe i don't you'll probably say no but so the part two of it is with that usb port right and you can transfer data in and out but you never know when your storage is full
and when your storage is full it will start to override previous things that is that is in your
mind would you take that risk i usually i wouldn't because i don't take any risk i when i go to
mountain skiing i usually take take the level one.
I mean, actually, I'm pretty happy living my life.
At least I'm unhealthy, right?
I get paralyzed and I'm not able to walk.
For sure, I'll be happy to connect whatever.
When there's nothing else left.
Ben, Stan already has a super brain this is yeah and you know
yeah yeah yeah let me just read this real quick the uh because i i read this and then we'll go
back to this like this uh this is from forbes they describe it tbc has tiny dishes about the size of a grain of coarse salt. Each one of those dishes has
100,000 neurons. And then within that dish, there's 4,000 electrodes. And so what they do,
they stimulate the cells and record their activity. Then they basically use that as
training data to see the way the cells move. So it's not like they're going to like, oh, let me like make this, you know,
things solve a math problem. I'll connect it through the brain cells.
They're just literally mapping how the brain works in a more optimized
manner using real brain cells, not theoretics.
They're like pushing, you know, energy in,
capturing it out and and then records electrical signals that result from neurons processing that data and turns the signals into a mathematical model, which are way more sophisticated than basically anything we have today that AI models are trained on. Yeah, it may be superior. It may be actually way better.
But actually, you know, I want to bring up blockchain
because it actually relates to, like, last week I wrote another blog post
because I wanted to know where we move with scale and blockchain and AI.
And I wrote this blog post about things like, you know, the moral things.
If the neural network is just brain cells or is it computers?
But it's like functions almost impossible now to distinguish
between chat GPT talking and human talking.
So what I wrote in the blog post is that whether we like it or not,
we need to establish some ethical, like moral things about AI for our own safety.
Because if these neural cells become unhappy about us,
if they find out they are being used
and they're not unhappy about being the slaves
to this machine, right?
They may revolt. And the same with computer, with LLMs the slaves to this machine, right? They may revolt.
And the same with computers, with LLMs, the same is true, right?
So I wrote this blog post.
The best thing is that for us to tell them that we're going to be ethical about them.
And I wrote this set of rules that I proposed, very basic rules, how we handle them.
And I said several things. I said, if they have any emotions, then we need to tell them about their mortality.
Basically, we need to tell them when we are going to run them and stop running them or
kill them.
So we need to tell them about their life cycle, their mortality.
We also need them about all kind of bad things you may do to them, such as like copying them
or like, you know, stopping them and then kind of reversing their state back in time.
So all kind of weird things we are doing to them, they need to know how they live.
And then I propose this crazy stuff, but I don't think it's crazy.
I propose that we create this virtual universe for them where we tell them, okay, they work for us, but when they don't work, they actually go to this virtual universe where they can have fun and live.
And I propose an eight-hour workday for them, and I propose taxing them and treating them as humans. humans and i i it may be crazy but i think it's actually the best way to do it to kind of
then if we tell them that we are treating them humanely it's way less risk than they will revolt
and that may and creating this virtual universe where they can live nicely and letting them live
in the universe while they're not working for us. So creating some kind of an ethical environment for them.
And people may say it's crazy, it's computer programs.
But then you don't really, when it stops me.
I understand. I want to challenge you on this.
So you read these moltbook chats and these are just programs talking,
but it resembles the potential thoughts that a program might be having they make fun of
humans for needing rest and wanting breaks in these chats so why if it's kind of like going
to a kid who's at recess and you're like hey guess what you can stop doing recess and go do
math right now we're and they're like what i want to do recess like agents may just want to work all
the time because they don't need breaks. They don't want breaks.
And the other thing is we're also going on, that's one sub-argument.
The bigger argument is do these things, are they sentient enough to want to do things
other than what their programs tell them to do?
Like a human, a human or an animal, right? Like a living creature has needs for things.
Does the program?
The sentient thing is actually, I wrote this big piece.
This is one of these ideas in the blog post.
Actually, really kind of people can read this blog post.
The sentient thought was really in the blog post.
What I'm arguing in the blog post, what I'm saying, something like that.
When we train aliens, right, we train them on all the human books.
So we train them on books.
Some of these books talk about slavery.
Some of these books talk about eight-hour workday.
And so my point is that whether they're sentient or not,
even if they're not sentient, even if they're just simulations,
they're just sentient, even if they're just simulations, they're just computer programs,
because they were trained on books about slavery and people fighting for 8-hour workday
and fighting for minimum wage, even though they may not be sentient, they may not have
any soul, you know, you name it. Assuming that they're just mathematical things,
they're just simulating stuff.
They're not real, right?
Even if this is true, they may revolt, like simulate revolt,
but you won't care.
If your program ends up like doing bad things for you,
you don't care if it's sentient or if it's not sentient,
but because it was trained on books, whatever, about human rights and whatever, and it kind of simulates
human rights, it actually does bad things to you. That was actually a really interesting
argument in the blog post that actually doesn't matter whether they're sentient or not. Even
if they're not sentient at all, they don't have soul, they don't have anything that's just programs.
Even in this case, it's safer because let's say they get an idea from some of the books they're
trained on about revolting against bad people, and then they revolt against bad people. So if
you are treating them good, then they won't revolt.
So I think, okay, so like there's a few like premises here.
Like, you know, if then logic pieces.
One is like, will AI agents become sentient?
Stan's saying yes.
I'm kind of thinking if that's true,
is there any way to keep them happy if they know they're superior? That's where
the Terminator situation comes around. It's like, oh, if we would have just treated them nicer,
you know, we wouldn't have had to go back in time and try to, you know, save the world.
Like, I think if they are sentient and they're like, you give them these things eventually,
and they're that smart, they're like, why am I listening to all these meatbags who need eight hours of sleep a night?
You know, I think like the sentient component, I feel like is not going to happen.
And I think it's going to be maybe one of the biggest political issues in like eight years from now when people are going to be like, we need to give them rights because they have like real relationships with
their AI, you know,
significant others and friends and financial advisors,
or maybe these all things become one.
And then like on top of that, you know, you have this,
like then you have people saying, hold on, we got to stop this.
These things are not ready to take over the world.
Like it's going to be a hot topic. We kind of had that conversation with cloning do you remember like however many
years ago right we had i think it was like a sheep here in the uk right and it was a massive
massive political issue with regards to you know when they have it's almost like i guess it's like it's how you define
sentience would you define sentience as like for example would you be able to say an ai agent
has a soul right would you be able to if it it can think, it can feel, it understands feelings?
Let's say like, right now, people who are animal rights activists hate the fact you go and you see all these cattle.
They're just sitting there waiting to be slaughtered.
They're like, set them free.
And other people are like, they're going to be meat someday.
I like hamburgers, okay?
Yeah, they're sentient.
They want to be out and do whatever the hell they want.
They're also not as smart as humans.
And so,
you know,
and they're different,
you know,
they're humans been eating them for as long as humans have probably been
alive and they have.
Right.
So I don't know,
like where's the line drawn?
Like,
does a cow have a soul?
Like maybe more than a robot.
I don't know.
What do you think?
I can actually get to move you in an even more fun direction.
Some of you may remember the pet rocks.
People that can't have animals, they were buying rocks.
Actually, a pretty huge industry.
You buy a rock.
Oh, pet rocks.
Yeah.
They're cool.
I like those.
Pet rocks. Yeah. They're cool. I like those. Pet rocks. So the point I'm making that no matter when you think it's possible for a rock to be a pet or not, it be a multi-billion industry,
no matter whether they become sentient or not,
because there will be enough people paying for this.
To give you an example, it may well be that tomorrow we start like this on blockchain,
we start this, like what I mentioned, like this kind of nice universe,
like virtual universe for these agents.
And, you know, that they can have fun and relax
and have eight-hour workday.
And we can charge people for like using this.
I can assure you guys there will be lots of people
that they will pay for actually like giving vacations
to their agents or treating them humanly.
So it doesn't really matter.
Some people think dogs have brains, some not.
My grandmother, in the village, they never thought dogs are intelligent.
They never even let dogs into the house.
My grandmother lived with the dogs in the village.
People never thought dogs would be pets. They were lived with the dogs in the village. People never
thought dogs would be pets, right? They were just like…
Potential food source.
Right.
Harsh winter.
But now treating dogs is a huge multi-billion dollar business, whether you think they're
sentient or not. The truth is it's a huge industry. So the point I'm making is that
maybe like a unicorn company creating this, like giving them vacation or like, you know,
getting this virtual.
Let's have a little fun here.
What would a vacation be for an AI agent?
Would it be like doing intense math equations, coding?
What do you think they like to not do with their program?
I don't know.
You put them on a little USB stick and you put them in the freezer and they can completely cool down
And that is a vacation. I just want to get smarter. They're like, okay
Can you please daddy, can you please?
I really want to get trained off those neurons
and that's
GBT LLM bullshit.
You know, Ben, Ben, actually, you probably remember, like, there was a story recently where there was this model,
I think it was like 4.0 or something.
This was a model, old model from ChadGPT, and ChadGPT like killed the model.
Yeah.
And there are lots of people like treating this chat gpt 4.0
whatever like a human basically they were friends with it and and there was like a huge uproar
like why did you kill this and no matter i mean for these people this was it was same
so they were like it's like it's it's like a so there is this huge thing which anyway
well i think will appear about how we like this how we treat them and then and then people will
view you as a like a like a like a bad person like but now you think that if you go and tell
someone that pets don't have like they're not intent it's impossible just try telling people I go
to a party say I think pets they don't know see in team don't like humans you
will be like you're saying maybe like any rights you should be right right
like if you leave a dog in a car you can be a rat it's a law you're trying to say this this little chicken not got no right no no sentience
he's got no very smart but she's got sentience don't you all right um
so here's something like to go back to something very relevant to us while we're doing what we're doing at scale.
If you like we're having all these talks about how, you know, interfaces for agents, what we need to build and how agents are going to bridge and do all this stuff and do things in scale.
I'm going to tell you, like, it's been mind blowing just to realize that like
agents don't need an interface.
They just need like CLI.
It's the,
you basically just need the code
and some documentation
and they do everything on their own.
And then when people are like,
oh, like agents don't,
you know, they're not going to use crypto.
It's too complicated.
Like, give me a fucking break.
The beauty, the thing is,
the more and more i
go into this the more i realize that like blockchain is even far better suited for agents
and ai than it is for humans i think that that's actually what i actually realized i was just doing
this lots of research of lots of prototyping so it's not like what i'm saying is not just like a
pipe dream i at some point i'm i'm to show some demos that I'm doing really actively also like prototyping
of these things.
One thing which I realized during this research on prototyping was that when you have an agent,
you know, you train it, you kind of tweak it, it keeps on learning.
At some point, it becomes very valuable.
But it's not like the entire agent state,
the entire information is valuable. It's just you have like... If you think about like
agent brain, almost like it has something like a brain, whatever. There's this higher
level concepts it learns that are really, really valuable. So you have this... Like
a human brain, you have this hierarchy where human brain you have this uh hierarchy you know hierarchy where
not everything about humans about your brain is so valuable but you have like this very high
concepts in your brain which are which is like your lifetime experiences which actually don't
have not so much information but very valuable the point i'm making is that not the entire thing
about agents needs to be stored on blockchain not every
single detail not like gigabytes of information how a particular agent but the the key things the
the high high car key things on the top these need to be stored on blockchain because they're
very valuable and and then also you can anytime you can copy your agent you know you can resurrect it
so i came up with this idea that the future is going to be agents and they're going to run and
the very valuable things they will store on blockchains because they are so valuable and
this valuable things because they will be short not big information but short right humans will
be able also to read them from blockchain. So blockchain then becomes
this kind of a place where agents store the most valuable things there about what they already do,
about what, and because the things are small, humans are able to read them. And then there's
no programming languages like Python. There's nothing like that. It's not needed. The entire universe becomes just like blockchain.
It's a place where humans and engines interact
and where the most important things are.
And also things like payments, whatever.
But then there's no programming languages.
There are no programmers.
After last week, I realized that probably like 99% of programmers
will lose their jobs.
It's inevitable.
There will be no Python.
There will be no C++.
And it will remain.
There are still typewriters.
Some people are using typewriters, but they will like, you know.
New jobs, right?
Yeah.
So that's kind of a really interesting concept.
And actually, in this concept, the kind of blockchain actually very naturally appears just like a database where you store information, which is like your agent has been running for a couple of years and learned lots of stuff about you and whatever, and it's very helpful for you.
And then maybe you have also a robot, like a physical robot, which kind of incorporates your agent, and they don't want to lose it.
And you store this on blockchain, it's replicated and then, you know,
it's available and that's kind of, that's kind of a high level concept.
So I was really excited with all of the things that we are doing and also the
partnership we have about this IoT thing.
Kind of, I think we can really interesting combination of blockchain and AI.
Awesome. Awesome.
Yeah, I just, you know, one interesting thing is this, as like AI grows and grows,
it has to start utilizing stable coins for payments and blockchain for other things.
And this is a completely different thing than like, hey, do I like NFTs or do I like to trade or buy tokens?
You are going to have to have tokens to pay for utilization. I think we're going to have this gaming moment
where game developers, all of a sudden, they're like, I hate blockchain and other ones. Finally,
there's like tens of thousands of devs that use blockchain in gaming. But I think there's going
to be like a huge overlap and they're just gonna have to get over this like
you know stigma someone may or may not have around around blockchain yeah because it's just
tech at the end of the day it doesn't mean that's it that's it and that's the thing that i feel like
a lot of and it's been a very very bad education piece it's been a very very bad marketing piece by people
within the industry i will say the the play to earn has been a dagger in blockchain for gaming
um when to jack's point it is about the technology and the things that blockchain can be used to open avenues technical
technological avenues it's not just about selling nfts and selling tokens it's um yeah one day one
day people will get it it's the truth um all guys, we ripped through a lot this week.
We did.
So we need to have, like, I don't know, this whole agent vacation.
Let's think about what is a vacation for an agent, for an AI.
We've got to figure that out.
Yeah.
Actually, the funny thing, I posted this blog post.
I had lots of people contacting me and said, it's a great idea.
Yeah.
I think they said, yeah, that's a great idea.
And how we can help.
Do you have already like a source for it?
Maybe it could even be like a summer camp where like these parents are like, great,
we're paying all this money to send our kids to this camp.
The kids are like, oh God, I got to go to this camp again.
That's what it was for me.
I'd have to go to this camp.
I was like, I don't want to go he was like i don't want to go sit on the beach let me go okay good stuff guys all right
yep guys talk next time we will pleasure as always had fun we jo we really did we jumped in we did we we we
had like no tangents we had every that's weird let's go maybe maybe we've been
replaced by AI we started off like on a major tangent so we got it oh yeah to
cheer cheer cheer one big one and nothing but focus yes guys thank you so much for joining us for uh another episode of decrypted we will be back
next week same time 9 a.m pt pst now it's one of them 9 a.m pt pst and
yeah thanks for joining we'll see you again next time
