Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
AI Power Wars: Infrastructure Strains And The Trust Collapse🎙️
Recorded: March 3, 2026 Duration: 1:31:31Space Recording
Full Transcription
Thank you. Thank you. Thank you. Thank you. Thank you. Welcome, everybody, to this week's podcast on AI power wars, infrastructure strains and
really the trust collapse that's occurred. I think there's been a real emotional undercurrent
running through this week's AI news and that's about acceleration without consensus. We're
seeing AI technology scaling faster than the social contract around it. And the gap is where power, infrastructure and trust
are really beginning to collide.
So from this last week of headlines,
are we ready for a world where AI companies
look more like defense contractors than tech startups,
where your electricity bill reflects model training runs
and where trust and intelligence is negotiated
in the courtrooms instead of code? This week's theme of AI power wars and infrastructure strains and where trust and intelligence is negotiated in the courtrooms instead of code.
This week's theme of AI power wars and infrastructure strains and the trust collapse is not theoretical
anymore.
It really is visible in all the headlines we've been seeing pouring through.
I mean, we saw this when Fortune reported that AI into Pentagon deal while Anthropic
was labeled a supply chain risk and all the stuff that's happening around that.
Then in real time, we're watching an AI shift from mere production product competition
to geopolitical alignment.
The Wall Street Journal warns of a new AI threat to power grids, data centers unplugging
all at once.
Then we see intelligence hitting physical limits, and those limits are ethical not philosophical from futurism there's a report on anthropic dropping its huge
safety pledge so we're seeing in real time this trust narrative fracturing
under commercial or political or maybe even military pressure and when even
network world covers open AI launches stateful AI on AWS signaling a control plane power shift. It really
means we're glimpsing the quiet consolidation of control into infrastructure layers that most of us
users and consumers and just citizens will never see. So I think here's kind of the uncomfortable
question for today's podcast. Is this still a technology story
or are we witnessing the early architecture
of a new power system where compute, governance,
energy and belief collapse into the same battlefield?
Dapit, what are your thoughts around the stories?
Any stories you saw that came interesting?
What's happening from your side?
Oh, guys, things are moving so quickly.
I'm into AI models, right?
Yeah.
And I'm constantly tracking the new ones and figuring out, you know,
what are the best ones for my system and all that kind of stuff. Um,
just this week,
I think I've tracked fear for new large model deployments.
And each one of them increases the ability of the user by 20,
25%. Uh, this is happening every single week. Yep.
And I mean,
this is not even covering the news that you just discussed.
Just from a development standpoint, we get better and faster every single day, which I've never worked in a field that moved that quickly.
I thought crypto was that fast.
It turns out it was not.
So that's really great on the positive note.
great on the positive note on the more negative side of things I do agree with
you about the fact that we are not prepared for what's happening a lot of
people are starting to understand that like this is not a siloed event right
this does affect them whether it be in the case of war whether it be in the
case of electricity whether it be in the case of war, whether it be in the case of electricity, whether it be in the case of, I mean, just watching their governments, like kind of
not be able to handle this and do whatever is necessary. We've never seen a US company
designated a supply chain risk until I dropped it. That's insane. Yeah. Right. And that was because
they said, no, you can't use this for autonomous killing. you can use it for killing you just can't use it for autonomous killing i'm glad that's still there it was just that distinction alone
that was um that was the big the big battle there and then you're seeing other companies
like in particular xai and open it just, man, no, this thing, sure,
can't give you the date correctly sometimes, right?
But don't worry about it.
Yeah, use it for kill chain.
No issues.
We're moving too quick, right?
We're moving away.
But yeah, I mean, like you said, it is moving very, very fast.
DeepSeq is going to release a new model coming up.
But they did announce today in the newsletter that they're adding Blackwell SM100 and FP4 support that will allow lower cost scaling.
So whether it's a new model popping up every week or it's a new hardware support.
Yeah, this is moving insanely fast with the models and pushing through, you know, and as we've all seen in the last, you know, to some extent, I feel like people are
using, you know, maybe in the Department of Defense, so basically they're using AI to
jumpstart the AI, right? So I think what we're seeing is active conversations with AI, active
use of AI to jumpstart the next level of AI. And so now we're in an even bigger sort of cycle of acceleration because now we're using AI to accelerate itself.
And I think, you know, if you look behind the headlines and you look at what's happening in the different areas, whether it's research or military, supply chain, environment, security, regulatory, you know, all this stuff.
I think what we're starting to see is companies and individuals and companies and entrepreneurs
applying AI to their use of AI and jumpstarting the whole thing. And I think that application is also adding to this whole
sense of massive increase in speed. And like I said at the start, it's like we're seeing this
trust collapse, you know, and it's not theoretically more. It's visible in the
headlines popping up and we're seeing different things breaking and shifting very, very quickly.
It's, you know, I mean, it's like, wow, take a breath and sit back and just, it's crazy. The
thing is, you can't sit back because like you said, it's impacting all of us, whether you're in
the Midwest or in Virginia and your electricity rates are going up because of data centers.
And now there are issues with, well, what happens when the data centers unplug all at
once?
You know, it's like we're stress testing very important components about civilization
in real time.
Yeah, well said.
Well said.
I think the, or one of the most interesting aspects of this, right, is like what you just mentioned was
AI makes people more productive.
People make AI.
We're seeing for the first time,
like real exponential growth as a result of that
is because AI has made me so much more productive,
I can make AI so much more productive.
And then we go on and on and on and on.
And the gap in society is becoming kind of crazy.
Like a friend of mine last night and I were talking and I was like, you should use my software.
Like, check it out.
I can do some stuff for you.
He's like, yeah, you know, I really, I'm amazed by how much AI can do.
I had cursor write me out a full proposal recently.
I was like, oh yeah. Like it made the PDF and all that.
I said, well, no, I just wrote the text.
That's where he is as far as AI is impressive.
I shipped three applications this week, right?
And almost completely autonomously.
Get educated, y'all. Start using AI.
Like you can do this with almost any device at this point,
right? The optimizations and the smaller models exist for anyone. I'm currently running a server
in my house on old cell phone and a micro PC from 2016. And it can generate images and have a tax
inference, right? Like you can, there's no reason to get completely left behind in AI. You don't need a 3090 or any of the other things going on, like get comfortable
using it. Yeah. And I think, I think the more structured you are as you approach AI, the
more you actually end up doing with it. So for example, you know, I, I, I built a whole
product requirements document and marketing requirements document for a specific application.
And then I realized, well, you know, we can't build a whole thing at once.
We have to have an MVP.
So I then used AI to net it down to the critical core functionality we would need to release one version of this product.
And then someone came to me and said, hey, we've got a partner who can do a whole bunch of this stuff. So the partner sent me like eight of their major documents, you know, describes the
product, describes the company, describes a whole bunch of stuff. So I took all those, slammed it
into chat GPT. And I said, Okay, what would happen to our MVP if we and a major product, if we
incorporated these guys, and it came out with a whole cost benefit analysis of how it
would save like three to five months in development time and a significant chunk of money. And then I
went, oh, but then it gave me a bunch of risks around it because it's in an area of regulatory
control. And so I said, well, give me some SLAs. And then it just pumped out SLAs and it gave me recommendations in the contract with that vendor that I should implement so that it keeps us safe, keeps them safe and so on.
But essentially, I took something that would typically take three weeks and did it in less than a day and a half.
I mean, it's just nuts how fast it can push you.
nuts how fast it can push you. I do want to make one comment about the whole military stuff because
AI firms are no longer competing just for users, right? They're now competing for defense contracts,
national alignment, and supply chain designation. And really when AI companies are labeled as a
security risk or they're embedded into military decision systems, they stop being startups,
right? They start becoming geopolitical instruments.
And I think we need to realize that.
And this is a transition from product wars to power alignment in a very real sense that
we're seeing with Anthropic, OpenAI, Mistral, you know, and all of them.
And I think there's a deeper pattern, but it's really kind of simple.
Whoever controls AI infrastructure is going to influence economic
leverage, military posture, and regulatory narrative. These are three components that are
really critical. We're also seeing these courtroom battles, Pentagon deals going on,
a lot of deals going on that we don't see, and the supply chain classifications. This is a noise.
These are really early signs, I think, of AI becoming a state level asset class.
And I think this shift is really super important that,
you know, we need to be aware.
What are your thoughts on that?
I think it always has been.
You don't see governments coming in and saying you may not regulate
where data centers can be.
You may not down this aspect whatsoever.
You don't see open AI who doesn't make any money, right?
Like none, zero.
They don't make cash coming in and being like,
we need another hundred billion dollars
and people jumping in and giving it to them
without understanding that this is becoming
the infrastructure for society, right?
This is going to be used.
If you want to look up an interesting thing, the flock camera issue that we currently have.
So flock is mass surveillance in America.
But unlike mass surveillance in other countries, it's being done through companies, right?
So we as the people don't even have the the same control over it as like the Chinese do.
It's being done through corporations and Palantir was involved and there was a massive hack.
I've never seen anything buried faster in my life. Right.
Like go online right now and type in flock camera hacking and you'll see guys busting into these completely insecure things in 30 seconds.
And you'll see guys busting into these completely insecure things in 30 seconds, right?
And that is what we're planning on the next modality of our society being.
It's based on these types of technologies.
Now you currently have to have a camera in your car watching you while you drive after 2026.
We have a Subaru.
We swapped it out, right?
And they gave us a new one, and it's have a Subaru, we swapped it out. Right. And they brought us,
they gave us a new one and it's got a camera facing right at my face. First thing it said was sit up straight. I was like, no, I'm not going to sit up straight Subaru. Like how dare you?
We're, we're barreling towards something new, but due to the speed and the aggressiveness
of development currently, I think a lot of us are missing it.
Because like you said at the beginning of this, Louis,
like three days, you're out of the loop
as to what's occurring.
And that may be the best environment
for a government looking to implement
these new types of controls over its population
to be operating in, because none of us can keep up.
Yeah, I mean, the Flock Camera one is interesting.
Being hacked is like, you know, we're all, I think we're numb to the amount of security
breaks that keep happening and happening.
One that just happened just like last week.
So an AI engineer, you know, he was mucking around with his um his robot vacuum
cleaner right and he was trying to connect uh his his his little unit in his apartment or home
to a playstation controller because he wanted to muck around with it and and he didn't realize that
there was a security floor that suddenly gave him access to user cameras and microphones of over 7,000 other vacuum cleaners across the
US and maybe elsewhere.
No, it's globally.
Yeah.
So, I mean, it's like, what is going on?
It's like, you know, whether we got flood cameras, which are being used by, you know,
defense, military, police with no warrants, no control, no regulatory stuff. Or, you know, we have, you know, little robot cleaners running around
and maybe it's in your bedroom and there's a camera on it.
That's not really great that, you know,
someone on the other side of the planet suddenly has control of that little thing.
It's, you know, we're allowing this technology into our lives.
And you and I have talked about this, that, you know, this frictionless desire for convenience, right?
And so, yeah, yeah, let's have a robot cleaner.
Let's do this and this.
And it's like, hmm, now we're starting to see
as these things break, it's causing all these issues.
I just think it's interesting that, again,
because unlike the slow introduction of mobile phone
or the slow introduction of, I don't know, any other technology,
when it would break, it wouldn't be global.
It would be highly localized.
Gone are the days where when technology breaks,
it's a highly localized event. It seems like when technology breaks now, it's a highly localized event.
It seems like when technology breaks now, it's scaled to be global.
Any thoughts around security and global Roombas watching you?
Yeah, we're going too centralized, man.
We don't have to be as centralized as we are.
Like I just mentioned, I have an AI server running in my house
that anyone in my family can talk to
on an Android cell phone, 2016 mini PC.
That is junk I found laying around my house.
I'm gonna upgrade it with a laptop from 2016
pretty soon too.
You can detach a little bit from these systems
should you choose to, right?
Like one thing my team is developing is a decentralized Alexa.
You don't need to have Alexa's microphone in your house,
typing everything to Amazon.
You can just have that run local.
That's totally doable.
And, you know, it's something that we're developing.
Look out for Dappet's GitHub this week.
We're going to be open sourcing a few cool things that everyone can use, right?
Like there's a different type of AI model called a recursive neural network.
It is a large language model, but it's non-tensor based.
It runs very differently.
If you can imagine our AI models run on a straight line, this one runs in a circle.
So its memory does not increase as you use it.
The cost of memory is that on your system.
We built a whole like LLM manager for these models
because it has nothing to do with JAPIT
releasing it for JAPIT.
Because once we use this, we were like,
people need to know that they can do this, right?
On a laptop, you can run a five gigabyte model
without overheating it comfortably
and generate images and search the web and check the time and do all these things that we're paying open AI and Anthropic for, but you can just do it on your laptop.
So we're with the most recent ability to ship as quickly as we can, right?
As we work on DAPIC, we're also working on things to just kind of help everyone.
So I hope you guys like those when they come out and you should take a look at our GitHub. It's currently empty.
We have two releases coming out this week.
The decentralized server, which is called sovereign elite.
And then we have the birds nest, which is posting all these really fast, but reasonably small image models and the Raven models, RWKBs, the RNNs.
And you can run them just directly on your computer.
Have a good time.
Yeah, I think the consumerization of AI and models
is absolutely needed because otherwise it's just left up
to geeks like you and I and a lot of people on the call
to kind of figure this stuff out.
The rest are at the mercy of these, you know, insanely large corporations and, you know,
throwing the data over. So I think the faster we can get to local management of our data,
our AIs and so on, the better. I mean, you know, it's like crypto, you know, get a cold wallet,
get a physical cold wallet that you own and get a Yubikey to lock down your email you need you know banks have spent billions of dollars on
security you need to spend a little bit too lock down your email with a yubikey get a cold wallet
use that you know i i i had a friend here who had a friend um who had like a couple of million
dollars on a hot wallet right and i'm I'm like, what are you doing?
And they were having problems transferring out of it.
And they're not technical, you know, but they were a wealthy person
and they were interested in crypto and they were doing it.
But it's like, oh, my God.
You know, I think decent, like we've talked about this, you know,
the decentralization, localization consumer,
it's as if that's the only way to kind of stay safe in many ways, you know, as we keep moving
forward, because otherwise, the globalization and the scaling capabilities just seem to be open
to crashing. I want to talk about the next one of the other stories, which is infrastructure strains,
where we're seeing the physical limits of AI scaling.
I mean, the biggest constraint on AI,
in many ways is no longer intelligence or the models.
We're seeing those getting pumped out,
like you said, almost every day.
It's becoming electricity and the physical aspects.
A lot of headlines in the last seven days
are really converging on a clear reality around this.
Data centers are stressing the grids,
utility bills are rising.
And we're seeing construction slowing due to power
and permitting delays.
We're also seeing memory chips being consumed
at historic rates.
We're seeing vendors buying out a year's supply.
So I think another real dimension around this
is that we're entering the physics phase of AI.
You know, for years, AI and other technologies were framed, you know, software revolutions.
But now it's obviously becoming a capital intensive infrastructure race.
And the limiting factor in this space is an innovation.
It's really physical.
It's megawatts, transformers, cooling systems, silicon supply, and interestingly enough, construction workers, electricians, you know, to do this work. So, I mean, the strategic implication of this is really profound. AI scale is no longer governed by code alone. It's really being governed by land, energy contracts, and geopolitical supply chains chains as well as grid resilience and you know
local politics you know we're seeing communities coming together to say no don't build that damn
thing here so it's it's interesting on that i mean from you locally have you heard anything around
you know because you're in a data data center intensive area anything else on the build out
there you know and the constrict there, you know, and the
constraints of energy chips in data centers? Yeah, everyone's starting to get really pissed
off. Our goals are raising our land, we bought up. It's on every next door, you know, which
is like a local community posting place. Everyone's starting to talk about this, but I have to say one thing.
If you're gonna talk about an AI bubble bursting,
it's the bubbles not like the dot com bubble, right?
Where these companies were so overinflated in price
and then they went down.
What we're gonna see is,
Lewis, if you're actually just on a browser tab,
go to taalas.com
and then go to products and then try to use their chat bot.
This is an AI ASIC, meaning the model
is not written into memory,
the model's written into the chip.
As a result, this model is 1000 times faster
than an H200.
Yeah.
And you can use it right now.
Very cool.
So, dude, the first iteration of technology is always this way. We think we need trillions of data centers and all this energy and all these things to do.
We're going to just innovate.
just innovate and then the cost is going to rapidly, right?
And then the cost is going to rapidly.
Right.
And I think what's probably going to get hit is these,
these types of data center and energy infrastructure programs,
because they're not going to be necessary to run AI the way most people want to.
Now for a government kill chain, I can't speak to it, right? I don't know.
Models like Claude might be absolutely necessary constantly,
but Claude will get written to a chip in the next two years.
And they'll be Claude chips, right?
You'll have those and they'll be more energy.
We're going to see a lot of innovation
coming out. There's a guy named
Jaisos.
He's
a really, really gifted
physicist and engineer.
And he created something called a thermal processing unit.
And rather than computing with electric gates,
it communicates with thermal gates pretty much, right?
But it can also operate in a third state,
like quantum computers, similar to, on, off and infinite.
So we're seeing technology is gonna catch up to where we're at.
And when technology gets up to where we're at,
I don't think a lot of the things that we're talking about
would be so necessary.
We'll look back on these days and be like, wow, that was crazy.
It does mean though that AI is really here to stay
in a manner that's gonna infiltrate
like every aspect of our lives.
So six and one half dozen go together.
Yeah.
I mean, it's interesting.
The piece I mentioned about the data center stuff, there's a new risk that's popping up
with all this.
You know, as we're doing this rapid build out of data centers, some of them, when they
go offline unexpectedly, they are destabilizing the power grid.
So apparently in Virginia, July, 2024 and February, 2025,
dozens of data centers simultaneously switched to backup power
after high voltage line failures.
This then forced the grid operators to scramble to rebalance the supply and
demand. And,
and it really raised a potential new threat that's going to continue as data centers consume
up to 17% of all US electricity by 2030.
So, you know, there's all these other fault lines starting to occur on the infrastructure
side.
And you're right.
You know, it's like we used to think we need
you know 10 000 100 000 gpus um but this you know there are people building sort of the asic version
of cpus right they're building these mpus or whatever you want to call them um that can do
17 000 tokens per second per user on a model, right? Which is just bat crazy.
And I hope we'll see lots of others that will do that
so that while you're on hugging face
and you're looking across all the models, right?
You don't need six 50, 90 GPUs in your machine.
You can just get an off-the-shelf custom whatever it is,
slide it in, and it'll run. But again, depending on how fast that technology comes to market and
how agnostic it is across models, you're right. The impact on the AI boom could be quite significant.
It's kind of an interesting thing watching this mad rush of technology. And I have to believe
that the people building these hardware products are also using the models on their hardware
products to improve their hardware products. So again, I think we're seeing the inclusive and cascading effect of AI layered on AI using AI to make AI better.
Yeah, I have a funny story about that. I use Antigravity, which allows you the options of
Antigravity is a agentic CLI. So it allows you to code and you get to choose
which model you code with.
So I usually use Opus 4.6, it's a great model.
And what was I trying to accomplish?
I could not, it was sandboxed to some part of its environment
where I couldn't get it to do what I wanted it to do
on my system because of security measures, right?
Right.
So I asked the model, I go, hey, you know what? couldn't get it to do what I wanted to do on my system because of security measures.
So I asked the model, I go, hey, you know what?
Why don't we look at this software and see if we can just rewrite that portion of it?
Okay.
And I was like, oh man, that's crazy.
What the fuck?
It got wrong.
It did go wrong again.
But I had it edit its own software and break its own security measures, right?
Right.
It ended up completely destroying the software,
but I think that was the AI's fault, not necessarily the premise,
but I used antigravity to improve antigravity when Google owns antigravity.
Google, I'm really sorry.
It was for experimentation if you guys are listening to this.
But yeah, it affected its own system to the extent that it broke its own system
and had to conduct a reinstall.
Yeah, I mean, it's like doing brain surgery on itself.
Not good, obviously not good for humans and maybe not good for AI.
Right.
You know, the other piece coming up I want to highlight is the trust collapse.
You know, we're seeing safety narratives really breaking down.
When Anthropic and OpenAI are removing words safety from their mission statements and their vision statements, you know, it's shifting.
You know, and these safety pledges are really being recoded, reconsidered and removed.
You know, we're seeing military partnerships sparking a backlash. We're seeing
advertising models being actively tested, whether it's commerce in conversation or if you watch
some of the headlines about MasterCard and AI agents and Walmart and Target, Go check out what they're doing. It's really interesting. The whole
explosion of AI agents into commerce is just kind of nutty at the moment. And I think what
was once framed as responsible AI is really being tested under pressure. Trust is shifting
from branding language to governance proof. you know, actual activities now are popping up where you're going, that's not safe.
You know, users, I think, are really starting to ask the harder questions.
We've seen the whole quit JATGPT movement bubble up like crazy.
There's a protest.
I think someone put an article online that went to a protest about people saying, you know, shut down AI or close it off.
protest about people saying, you know, shut down AI or close it off. I think if AI systems are
influencing defense decisions and are shaping government economic infrastructure, you know,
and they're inserting ads into daily cognition, then, you know, public expectation is really,
really changing. And the thing is, the change is not dramatic. It's really, it's gradual. We're seeing a credibility erosion. And I think the
deeper risk is that commercialization, militarization, and scale are really
accelerating faster than consensus around the guardrails. I mean, you can't assume trust in AI
anymore. It has to be audited, you know, and systems are breaking, like in real time.
But it's causing this slow and I think maybe quiet erosion of confidence. Again, you and I've
talked about this boiling of the frog issue when it comes to trust and security. What are your
thoughts? Do you think, you know, I mean, there have been some protests around AI, but I don't know, it's just happening too fast
and it's happening at a deep infrastructure level.
I don't think, we can't go back.
I mean, the genie's out of this bottle.
It's just nuts.
David?
You're on.
Sorry, I was on mute, guys. No, you're not. Also, you have a very small group of people making these
decisions for all of you. Which is why I'm urging people to take that decision into your
own hands and you are able to do so. Literally, I have an 8 gigabyte RAM Intel Celeron processor running image
generation that's possible it took a lot of optimizing a lot of you know
discussion to get it done but you can do that so if you don't want to
participate in this new way where AI is tied into everything there are ways you
can avoid it there's also ways you can't. And the ways you can are probably going to be governmental infrastructure, right? Or vehicles. I don't think my son's ever going to learn how to
drive. I'm being honest. I think my son will probably be driven by whatever car I own or
whatever car we ride hail or whatever that is. You know, 15 years when he's up for that like society is going to change um we
still have a lot more control than we let on we're very distracted we have a hard time understanding
where our power lies because there are all these things going on all these distractions and this
constant level of activity that i don't think humans are very used to but get focused man
like sovereign ai is completely possible.
You can do it if you wish to use it. And like Lewis was saying earlier, your own operational security is your responsibility. I say that currently with a machine. It's just like,
it's got more ports open than I should ever allow. But we have to be cognizant of that.
Yeah. I mean, you brought up the whole car thing. I mean,
my son's 21 and I've just, I've just given him a very limited Uber account, right? Because if I
was to put him on my insurance for my car, it could probably double or even triple what I would
pay. It's just nuts. And we all know, you know, certainly for men, what it's, you know we all know you know uh certainly for men what it's you know 21 in a car you know
um it's just it's like yeah um so you're right i don't think our children will be actively driving
and in many ways for us as parents it's cheaper for them not to um to either do uber or to get
you know finally when the thing either with
Tesla or if you look at what's happening in China with their technology, it's streaking ahead really,
really fast. And because of the car industry in the US, we're actually seeing the cars from China
going into Canada. There's been some great YouTube videos on some cars that are like $37,000 that would cost like
a, you know, in the US, you'd probably play three or four times that with the level of
quality and the technology in it.
It's so the Chinese are moving very, very fast.
I do want to get to the fourth point in the whole thing, which is the whole platform control
shift.
We're seeing a new AI control plane.
And Dapha, you and I have talked about, you know,
where technology goes from centralization to decentralization,
back and forth and back and forth.
This is kind of slightly like that, but different. And it's a subtle shift, but I think it's important.
And we're seeing controls shifting away from single models
and towards orchestration systems.
We're seeing stateful AI on cloud platforms. We're seeing multi-model workflows. We're seeing multi-agent
communication systems. We're seeing whole platform ecosystems signaling that they really can't win
alone. And I think the winner in this phase, and given how fast everything's moving it might be next week but um it may not be
the best model it may be the best coordinator and so with intelligence becoming modular and
distributed and persistent and really embedded into cloud infrastructure and workflow engines
what we're seeing I think now is this rise above AI models of an AI control plane. And it's a layer that's going to decide which models talk to which systems under what governance.
And so this power is becoming, in essence, architectural.
And the architecture is really hard to see, which makes it even more powerful and more
concerning.
So it's an interesting thing where this orchestration is becoming authority
and multi-agent systems and infrastructure level orchestration of AI is really shifting
the whole power across as we see all the shifts in the other geopolitical, in the physical plane
and the collapse of trust.
It's just a really powerful shift.
What are your thoughts on multi-agent systems?
We tend to use, as consumers, ChatGPT or consumer one-on-one.
What are your thoughts around the actual orchestration with multi-agent systems for consumers?
Well, you think you're using ChatGPT and that you're talking to one model.
Well, that's true.
That's true.
Yeah, that's just not the case.
So if you spin up the voice mode, we're using a different AI model.
If it's calling tools, it's using a different AI model.
If it's researching, it's using a different AI model.
These are all multi-agent orchestrators.
Yeah.
I'm fascinated with cheaper coding.
So I'm a man who likes to be
completely open and honest about where I am
and where my products are at. We are the
best vibe coder for Web3
in DAPIT, period.
I saw Replit's most recent update
on their
new multi-agent.
Each agent does
a different part of the build
orchestration. I'm going to be working
for the next three weeks to ensure that we get that that level of quality for our users right
because by splitting up by creating a coordinator and then creating a ui ai designer and then
creating a coder and then creating an error checker and then creating an error loop checker
right goes back to the first one. You can increase speed
and you can increase qualities
far more than I thought possible.
I was trying to crack this by myself
maybe like three weeks ago
when I started a project called Layla
and Layla sucks, right?
But looking at how some of these bigger firms
have figured it out, it's amazing.
Like we could theoretically, I mean, have a model instantly on one submission, generate music, generate images, generate the UI, combine everything together and code the back end in minutes.
So I think it's amazing. I think it's really, really cool.
Yeah. Yeah, you're right.
Yeah, yeah, you're right.
You know, multi-models are acting for different media, whether it's visual, video, imagery, or chat.
You're right, they're all different and they're all acting as one.
But I think what we're seeing in the newsletter, if you look at the AI agent space, there's just massive activity, growth, evolution occurring.
massive activity growth evolution occurring um one thing that popped up i mentioned it last
podcast but um you know in the entertainment industry right we're seeing uh you know copyright
anxiety becoming a creativity constraint we saw music platforms starting to label ai content
because trust is there i mean the supreme court declined to entertain the notion of AI authorship.
But that's kind of pushing people
in the entertainment space and the creative space,
and maybe even in enterprise space,
where they're hiding their AI collaboration
because social trust is starting to punish it.
You know, people, I mean, disclosure fear
is really about belief and status.
You know, the social environment's really creating this shame gradient around AI use.
And so, you know, that encourages secrecy.
It erodes trust further.
But that really then pops up with identity threat for creators themselves.
And obviously, we're seeing a lot of suspicion from audiences who feel misled and betrayed, you know, with this whole interesting
belief system, sort of AI usage, creativity thing bouncing all around.
What are your thoughts, you know, the split between AI forward creators who brand it openly,
you know, whereas you've got AI silent creators who treat it as taboo, or maybe
people, creators who just don't use it at all is this
a phase we're just going through or you know do you think this is a long-term trend
i think when the computer came out we said the same thing right we said oh this is i'll do this
i like my paper i like. Nowadays, everyone uses the computer
in some part of their life.
I really appreciate the holdouts actually.
You're only able to do that for as long as the society
around you doesn't implement it as necessary hardware.
So currently, yeah, you can still stay away from AI.
That's pretty cool.
Pretty soon, if you choose to watch TV or do anything,
you're going to be interacting with it in some capacity.
So for those that are holding out,
I really thought long and hard about not getting a 2026 card, right?
I was like, I don't like this camera.
I don't want this camera.
And then I thought to myself, I was like, okay,
what does that buy me, five years?
I'm doing the same thing with a real ID, which in America, it's a national ID that we just came out with,
which I find extraordinarily unconstitutional, as well as like 94 cases that found the same thing.
We implemented a real ID.
And how they did that was they said, if you want to fly on an airplane, you need this certain type of ID.
Me and my families, we don't have them
we choose passports right but how long can that last with my kids you know in 10 years when he
needs to get an id card it's going to be a real id right like i have to accept where my power lies
and what i can actually control myself and it's not that yeah Yeah, I mean, it's this interesting balance
between individual, I guess, power and overall security.
You know, we've talked about this on a couple of podcasts
where you're continually going back and forth
between I want to be an individual,
I'm going to have power over what I do, my space, my creativity,
but I live in a society where creativity uh but I need but I
live in a society where it's not safe so I need security about my identity and and like how much
do I have to give away in order to have that security but not give up on my autonomy you know
it's this interesting push back and forth and I think we're also seeing some interesting stuff as we grapple with that
you know um in the film industry there was uh seed oh i forget the last name anyway one of the video
um ai video see that by alibaba yeah yeah it popped out and it did that little apparently did that video fighting between Tom Cruise and and I
need an AI to remind me anyway some other stuff and you know Hollywood kind
of went oh my god you know and I think you know if you look at entertainment
and just gaming or whether it's movies coming up right think about this you
know let's say there's an avatar, an AI avatar,
but it looks like a human, it acts like a human,
and it's inside a genre, right?
And yet you've got humans and AIs acting together
in an entertainment form like a film, a movie,
or something like that.
I mean, it's not just Uncanny Valley anymore.
It's really about almost a disorientation i mean it's becoming
entertainment could become synthetic to the extent that you know where ai the ai avatar characters
are expressing emotions to the best that they can and we don't realize it because we're just seeing the video format and
you know i think i think we're going into an interesting area where there's going to be a
blurring between what is an authentic human in terms of expressing emotions um their thoughts
words and all that in the same context as seeing avatars doing similar things,
but maybe not the same.
And then, you know, I mean, our kids growing up will look at that
and will absorb that and act similarly.
I mean, we're seeing lots of trends in that area already
with people using AI for counseling, people using AI for, you know,
hell, get me in shape, you know, encourage me all this, you know, to run faster, move faster, lift better weights.
We're seeing AI get involved at an emotional level with us that has to, I mean, the thing is it has to impact how we process our own emotions and how we process our own thoughts at the end of the day.
What are your thoughts, ideas, especially, you know, we both have kids, about the impact of AI
on us as human beings and how we deal with ourselves and civilization?
Simple question.
Simple question, right?
No big deal.
I don't know.
I think it's important to understand that we don't know what we don't know.
Yeah.
And when we put people in front of smartphones and we gave them those algorithms, we did not know how that would turn out.
Now we're dealing with the repercussions.
I think the same thing applies to our current ai state right is this will change humanity i personally hope we're heading towards
star trek with this right the end of scarcity um the end of it it is capable of doing so man
like this technology is so powerful in between here and there i do believe there'll be a pretty hard adjustment phase type of thing,
especially as these models get smaller, more powerful.
You'll be able to run them in lower areas.
Non-state actors will be able to train them, things like that.
So we'll see.
I don't know.
It's one of those things where, like,
I really think it's a super exciting time to be alive.
Yeah. I mean, just, you know, companion AI are really starting to fill social gaps and schools
are trying to write rules, right? So it's migrated from being a novelty and really is kind of starting
to fall into governance, especially around children. If you look in the US government
section of the newsletter today, you'll see lots of different states start to push, you know, regulatory stuff around kids and use of AI.
But I mean, the human impact is that emotional attachment is, I hate to say it, becoming a
product service. And schools are being forced to manage this psychological risk, the dependency,
inappropriate interactions.
And there is a belief shift occurring about authority and intimacy. We're seeing students trusting a companion agent or an AI more than a teacher or their
parent.
And I think we're seeing this policy gap and crack, struggling to keep up with the emotional reality of our kids relying
on these AI companions in many ways.
It's really not a toy.
It's becoming a relationship surface.
And really, I hope emotional safety is really going to become the next safety category.
I mean, if a child trusts an
ai companion more who's going to be responsible for the value it teaches i mean it's it's some
really significant and i think powerful questions occurring um across our civilization as this stuff
rolls out um you know uh one thing i also want to bring up, you know, loneliness is also becoming a market and AI is becoming the interface.
And I think the phrasing here matters, you know, infrastructures for connections is really signaling a shift from apps to social substitutes.
We're seeing this human impact is really double edged. You know, companionship can reduce isolation.
You know, companionship can reduce isolation.
It can also reduce human to human practice over time.
Because, you know, as you and I know, AI can be like way too convenient, positive, and
lack any friction, which I think as humans, we need friction in order to have insights,
awareness, and a sense of kind of stepping beyond yourself.
But I think this belief level change is normalizing this machine mediated intimacy as an everyday
solution.
And I think once it's normalized, opting out starts to feel like being left behind.
And that's one of my biggest concerns, especially with kids, where there may not be a level
of discrimination. I mean, you know, you go through enough hard experiences in your life, you start to set
up boundaries, you start to, you know, take care of yourself a little bit more, you know,
even at my age, you know, I'm still always learning around boundaries and, you know,
keeping yourself safe.
Kids may not always have those filters developed because they just haven't had the experience. And yet with the introduction of AI companions, I think there's a huge issue around this emotional
economy becoming productized. And unfortunately, the long term costs may be, unfortunately,
a reduced human social resilience, which I think could lead us down some pretty dark paths.
What are your thoughts around, you know, this whole normalization and kids and AI companions
and chats? I mean, it does solve problems. That's true. You know, people who are lonely,
they can have a, they can chat and, you know, have conversations. But I think at the younger ages, there's real issues.
I am building an intranet in my house for my son.
He's one and a half, right?
I've got some time.
But it's going to take me years to get this right.
So I'm currently building an intranet with his own AI model that I can monitor, his own
Netflix, his own everything that you would want to get from the outside world. I'm building on
servers inside my house so I can monitor my kid because this is not the kind of thing that like
society can handle. This is a really bad idea. Giving kids unfettered access to the internet
was a bad idea to begin with. And now doing it with AI models that will literally tell them
anything they're thinking is great. Like I know of a friend, my best friend's a psychiatrist,
his best friend's a PhD. I'm convinced he created some new sort of math proof. The man's a PhD.
He created some new sort of math proof.
The man's a PhD.
That's insane, right?
He did not, none of that happened.
And like it, this is why I hate chat GPT
above all the AI models.
It's sycophantic.
It wants you to think you're this like super awesome genius
because you asked an AI to build you something like,
Claude is far more pragmatic.
We need to be really careful of what we're doing.
Don't like, and tell any parents out there
just don't let them use it.
There shouldn't be a reason for our children
to be exposed to this so early.
Like they can learn how to Google something
as crazy as that sounds.
Or they can use AI for as a search engine.
That's fine, I guess.
No, well, when I was a lad, we just had Google. As crazy as that sounds. Or they can use AI4 as a search engine. That's fine, I guess.
Well, when I was a lad, we just had Google.
Right.
Did you just hear me?
You know, it's like the old book.
Oh, geez.
Luxury with this AI conversation crap.
When I was a lad, I actually used cards at the library to find the damn research book.
Right.
I'm not trying to be a lot like that.
I know.
But it's just like, you know, I mean, I can hear my kids going,
oh, really?
You know.
But the thing is, mental health safety is becoming a model design problem.
As much as I hate to make them Satan,
we're using sentences that have never happened before.
Just as a quick segue, the Pope came out and asked the clergy to please not use AI to create
sermons and someone noted this is the first time this sentence has ever been said you know and
there's a lot of that going on you know I mean mental health is you know people arrive vulnerable
you know and bad outputs can cause real world harm we've seen that time and time again and
lots of headlines around that and the human impact is really high. And it doesn't matter if you're a genius or you're
not, you know, your biases and everything else, and maybe you've had hard life experiences.
You know, it can cause issues that, you know, you need another human being to work it with.
that you need another human being to work it with, maybe not an AI.
And unfortunately, you get lots of liabilities, issues around this.
And helpful is not the same as safe, right?
Let me just say that.
And it's sad that emotional care is becoming a product feature for AI.
product feature for AI. I think it really is. But I mean, you know, should, you know,
I think it really is.
I think, should AI be allowed to sound like a therapist, if it can't actually be one,
I think is something that we really need to address, right? You know, it's like,
this is where we need, if we are going to do it, we need special tools, you know, and if you look
in the healthcare section of the newsletter, you'll see people are developing specific ai for specific mental health therapy
stuff which is great but you know when when it's convenient and it's only convenience and you've
got chat gbt people are going to use that thing um so um again this whole thing, this whole stream of AI productization and access and usage
has just been so fast. When intelligence has become such a sheer commodity, it's just been
crazy how fast it's going. One other piece I want to bring up that's kind of relating to this is
that elections are becoming a persuasion war between AIs.
I mean, if you think politicians aren't using AI, too late.
And they're creating essentially these persuasion systems and they're automating them.
And so you'll start to see, especially with the midterms here in the US coming up, elections become less about debate and more about attention capture.
You'll see narrative targeting. I think you'll see manipulation at speed for a term there.
And I think the human impact, I think you'll start to see significant civic fatigue and distrust
more than there is now, which is hard to think of it, and maybe even disengagement.
And this shift really is existential for democracy in many ways.
I mean, it's really serious in this era.
You know, you're seeing the legitimacy of democracy really depends on shared reality.
But AI is going to strain that like up the wazoo,
especially when this persuasion starts becoming automated
and hyper-personalized.
What are your thoughts on the political process with the involvement of AI?
Obviously, in military, what could possibly go wrong?
But in politics, what are your thoughts?
I think we...
So one of the first real applications of AI was politics.
There was a piece of software.
It was purchased by a firm called Cambridge Analytica.
If anyone recalls, there was a scandal around that.
Yeah.
Humans weren't prepared, right?
So that software was originally designed and developed by, I don't know if it was the
DOD or one of our defense contractors, but to convince
boys in Afghanistan not to be terrorists. That's why it was made and then it was turned into a
commercial service or political campaigns. We are up completely understand that I cannot handle that, right?
Like I am subject to the same difficulties that any other person would be when faced with an entity of that scale.
So my solution is to remove myself from the game.
I get my news from guys on Telegram taking pictures and
videos of whatever's occurring, right? Direct sources. I don't check any of the news places
anymore because I understand that they're designed to direct me in one way or another.
And if you look at America right now, this is the result of that.
And if you look at America right now, man, this is the result of that. There's nothing else.
There's nothing else.
Like we are arguing over absolutely everything.
Everyone's pissed off.
All those things we shouldn't be subject to.
Yeah.
Yeah.
Be careful.
Yeah.
Yes.
So everyone be careful out there.
One good news.
I mean, you talked about, you know, you hate chat GPT.
I get it.
But Claude just announced you can transfer your past conversations with other AI chatbots.
They're implementing a service where you can import the history.
And while it's a move disguised as being user friendly, clearly the impact is to transfer your preferences, habits, decision
patterns, making them portable, and that makes switching easier. But in many ways, it normalizes,
if you think about this, it normalizes life experience logging as a product service.
And the deeper shift, I think, is that competition is moving from model quality to who owns your context, who owns your, you know, your your chat sessions back and forth as you, you know, try to solve things and work stuff out.
And I think once this context is important, the assistant can then shape choices more effectively, which is execution control by personalization.
I think, you know, this portability, while it reduces lock-in,
it really increases data exposure.
You know, we've seen so many times now, you know, systems getting cracked,
systems that, you know, and portability is great.
Again, we're back to this, oh, I want convenience in my life.
I don't want to be locked in.
But, you know, portability and allowing you to move all the
work you've done in chat gpt or claude or whatever to a new model um do you think this sort of an
identity experience transfer if you will um will become more mainstream and And hopefully, if we have more powerful local devices, we can
then just bring it all down local. Again, you think this is a trend?
I'm going to kind of derail the conversation and give you another side part to consider
in this subject.
Yeah.
Did you read about Project Lazarus from Facebook?
Okay.
This is a big one.
Okay.
So there was this speaker
on 4chan
maybe
four months ago
and he claimed
he worked for Meta
and that Meta
was developing
this
this software
copy exact behaviors
of individuals so that when they passed away,
we can still talk to our, our, yeah, yeah. Raising the dead. Yeah.
Right. Which is funny because not only was this the subject of Westworld,
the project name was the exact same in Westworld, which I find.
Right. So everyone thinks that this guy's crazy.
He has some very specific details that were kind of hard to ignore, though, right?
Like that it started as a meta project, and then the DoD came in, and then they siloed their information, and they moved them to an island to work on this thing.
Which, again, is so Westworld, it's mind-blowing.
They just filed the patent.
Like, publicly filed the patent.
Now, that sounds cool, cool right that sounds completely cool but the
way that you communicate the only reason the way I know Lewis exists is because I'm speaking to him
on spaces right right and if there's a model that can duplicate your behavior exactly as it is, we could remove Lewis in reality, and I would never know it.
Right? Absolutely terrifying tech. You could remove a whole village and then
slowly every person could die of natural causes, a heart attack or whatever, over
time while their digital footprint is still maintained by these models. It gives
people, because the digital world is all that connects us, right? Like it's,
I don't send carrier pitches to someone to verify their existence. I send an email or I send a text message or we talk on a phone call. And yeah, Facebook, at least we know their objective. I
can't say how far along they are, but their objective, it was big enough to file a patent
and they're far and long enough to file a patent. So to answer your question, I think we have the types of problems that we haven't really
considered coming up to us in this business.
Some very interesting things are occurring.
One of the things that scared me, I had a friend who works in, what does it say, a three-letter
agency, right?
And I was talking to him almost eight or nine years ago.
Said, what's the biggest real threat
that we don't talk about?
Like in the normal world?
And he goes, oh, digital people completely.
And I was like, what do you mean?
And he's like, oh, there's PhDs, there's influencers,
there's all these people that only exist online,
they're not real.
And it's getting hard for us to tell when other people are doing that.
That was nine years ago.
Yeah.
Yeah, I mean, I think, let me bring it up, because I've mentioned it before.
There's a number of companies now
that allow you to create your own AI twin.
People here, if you go to twinprotocol.com,
you can see you can create your own twin.
You can then load it up.
And you can then, if you're a consultant or you're a coach,
you can then scale yourself to global
and have thousands of people talk to your AI twin all at the same time.
You know, there's this, you already have like this, this guy, he runs a company in California, Cali Floors.
You can chat to his AI twin.
You can chat to Deepak Chopra's AI twin, though there's stuff around that that we
won't go into. You know, so many, it's already there that you can now create, you know, you can
talk to Zorro. I mean, it doesn't have to be a real person. You can now create fictional characters
like Davy Jones or King Arthur or, you know, Peter Pan.
You can talk to whole bunches now.
Winnie the Pooh, Emilia Ayrton, Mozart, Beethoven, Shakespeare.
So, yeah, digital personas have pretty much gone mainstream in the whole thing.
And it's this, again, we come back to this interesting thing
around identity and security and who am I,
what's real, what's authentic.
It's shifting so fast.
It's really kind of crazy how it's going
with these chatbots.
And when I think, you know, we're seeing marketing
and advertising in a way with becoming hyper-personalized, in some ways it's becoming weaponized.
So that these personas will seek to influence, persuade, and manipulate you.
Again, I'm not sure, you know, what can we do about this stuff?
It's hard to say because it's just moving so fast
and it's happening at an infrastructure level
and it gets done really, really fast.
And like, you know, I've talked,
people are using AI to improve their AI, to increase the AI.
It's accelerating like crazy.
I think we just need to be really super aware,
but, you know, not just as an individual because any one of us can be flim-flammed,
as, you know, clearly it's true,
that us humans have to start connecting together
and making sure that we can reduce this as much as we can.
I don't know quite what else to do.
You know, when we're seeing a Genic AI,
you know, attacking developer ecosystems,
there's a couple of stories that popped up on that.
Unauthorized AI agent execution code was published
into an open source and then that went out on the attack.
So we start to see, you know,
browser vulnerabilities with AI agents pushing through those. And then of course, the shift
with a genetic payments moving from concept to execution and money flows that's also expanding
really fast. It's just interesting watching how fast then when you combine it with
all the crypto stuff that's happening and the blockchain stuff it's yeah I
feel like sometimes it's like you know when you have like the hurricanes you
see in the Atlantic you know sometimes you're in the center where it's peaceful
but you know very close outside of that center, things are moving
super fast and you can see it. And sometimes you're pulled into it and it's like craziness.
And then sometimes you can come back to your center. I feel like we keep bouncing in and out
of that in some ways where, you know, we center ourselves and there's a level of silence of peace but then you get rushed back out into this maulstrom of amazing change progress and just chaos
again it's setting time to be alive right um but it does come with the negative sides of that as
well doesn't it yeah i mean the the ability to do
vibe coding like you guys do and and create apps you know essentially someone made an interesting
comment they said that uh in in the music field they said that these ai tools have reduced the
cost of learning to zero right you don't need to be an artist and spend, you know, 10 years,
20 years learning the piano or an instrument or, you know,
learning how to write music or, you know, learning stuff.
You can now essentially with AI tools instantly create pretty
sophisticated music, whether it's a symphony or a pop song
or a country-western song. Country-western songs seem to be easier, I don't know why. But it's interesting that
this, it's like the cost of entry was so high,
whether in terms of time or cost or both, or just ability, right? You know, some people are
great playing the piano. Other people suck. I suck, you know. But, you know, with AI, we can overcome a lot of that sort of cost to entry stuff.
And I think in that way, it's a great time to be alive and an amazing time that we can take our inner creativity and we can repurpose it to almost any field now in many ways than we could ever do before.
And I think it's expanding.
And I think that's the good thing
of what's happening at the moment.
I think the issues though,
we're seeing it happen in the business enterprise,
finance areas where people are struggling
to find employment, people are struggling to figure out,
to get trained so that they can be employed, whether
it's with new skills around AI or whatever.
We're in this massive transition point that I think is only going to deepen for at least
the next year or so.
What are your thoughts, Stafford, around our ability to work, earn money, be creative, while all this change and increased
capability to work and do stuff is happening all at the same time.
I personally love it. It's, again, I don't think it's probably the best thing for us.
Blindly as we currently stand, right? Like we're not working on this in the right way,
societally, but you know,
the alternative might've been a 20 year rollout
with a lot of bureaucracy and such.
Maybe the right answer is to just barrel forward blindly
and then fix the problems as they come up.
The one thing I'm concerned about,
we talked about this on Spaces a long time ago, is that like, this is such a large change. I believe it's fundamentally
going to change what we call governments. I think nation states and areas that control this tech
are going to increase in power so quickly that a government is going to pale in comparison to the abilities that they have, right? Yeah.
So yeah, go ahead.
Yeah.
Just in conclusion, like, I think we all are along for a really cool ride, man.
We'll see.
Get sovereign, though, a little bit, guys.
We're really like, I've got a battery right here and my own AI server.
Should things go wrong, which they will, inherently all things go wrong to some extent when it comes to new technologies I can still operate relatively the same with what I have on
my system so like we all could do that yeah it's it's interesting I think in some ways we might
see smaller governments maybe in Europe or South America where they do create sovereign AI, they do use it for the good of their citizens
in a way that makes sense, allowing for greater democratic access to services, to policies being
created more intelligently that support more people. I think those countries that do that,
I mean Estonia, if you look at, you know,
the whole, you know, crypto and blockchain came up,
Estonia really looked at that as a way
of enabling greater access, support,
and power for its own citizens.
I think if we see small countries
using sovereign AI as a way to basically create some level of equality with larger countries around them.
We could see some amazing advances, you know, at the level of government, democracy and citizenship.
You know, again, you know, we also see certain governments using it for higher levels of surveillance or if not the governments, then the enterprises that support certain parts of the government, you know, basically setting up AI or technology to surveil and manage and report on citizens.
It's interesting.
I think we might start to see a lot of trends around that.
I mean, the sovereign AI stuff's been popping up
a little bit.
It was more vocal about two weeks ago
with the whole conference in India.
But even so, some things are still popping across.
I think what's happening is both a consolidation phase and a test and demo phase at the government level.
A couple of stories in the U.S. government section and the non-U.S. government sections of the newsletter.
You start to see, you know, development, using AI to develop better policies and so on.
Again, that kind of raises, you know, the whole specter of
why are we using AI to do this? Shouldn't we use humans? But hopefully, you know, if these are
experts in these spaces, they're using it in a way that makes sense. And so there are less
unknown unknowns, but you just never know. So yeah,'s it's interesting i think you know i think for the
first time ever uh i saw a report that for the first time ever um immigration in the u.s went
negative people are leaving more than they they've ever done in previous history at all um as as
people are trying to uh figure out you know what figure out, you know, what's best for them as
individuals, what's best for them as families, does this society I live in support that or not?
We might see a lot of, you know, as AI gets more integrated into some countries and governments,
we might see even more migration, you know, where, you know, people would migrate to a country because of
economic opportunities, maybe they'll also migrate now, not only for greater economic
opportunities, but for greater democratic or support by those governments of their citizens
their citizens opportunities. What are your thoughts? Do you think we might see a trend around that?
opportunities. What are your thoughts? Do you think we might see a trend around that?
I don't know. I think we are seeing a trend on it right now, right? So I guess it's less about
if we're going to see a trend or how much, to what extent that's going to affect modern society.
I can say that our current systems are no longer acceptable to the average person.
We just want to live, die, take care of our children, educate ourselves.
Humans have really simple needs.
Educate, see towards the new frontiers of life.
That's really all people want. And it's, it's amazing how quickly in America,
I'm watching the ability to do that degrade, like so fast. I mean, groceries are inexpensive,
our property taxes are soaring. I saw an article that says that the American actually pays closer
to 83% of taxes when we include property, sales tax, what do you call it? State and federal taxes.
And what was the last one? Like death and capital gains or something. 83% of the money is just
circulating right back to the government. I'm sorry, I can't say those are accurate numbers,
forgive me. We know there are at least 47, right? But like, that's not sustainable. That means I
exist for the government, not the government exists for me. And leaving that is logical.
If someone trying to, you know, raise a family and continue, we don't want to just pay more for
the same things. Our insurance system in America is currently broken for healthcare. We're paying
eight times more than any other civilized nation for the exact same services. We're not super elite in that field anymore compared to the rest of the world. Something's got to give.
the term not me it used the term late stage and i said whoa late stage what and it was like oh well
the society you are in is in its current late stage evolution of where it's a new system has
to come in soon right because no one the old ones are served and it's probably that um the one thing
i wanted to comment on i didn't get to earlier was your comment about security.
Yeah. I don't know who said it was one of the founding fathers, but like the country that gives up or maybe it wasn't.
It was someone the country that gives up security for comfort or freedom for comfort does not deserve either.
And we're just barreling down that road. We're blaming roads for drugs existing.
And we're just barreling down that road. We're blaming roads for drugs existing.
Like they're saying you guys can't have cash because drug dealers and bad people and cartels use cash.
We shouldn't believe that. Right. That's not a thing that like every person should just take as as a fact.
Like, sure, some things are important for large institutions and whatnot.
But like you have a right to conduct transactions without KYC. That's happened throughout all of human history, right? No one was looking at
people trading cash and going, well, this person is that person, that person, this person. Okay,
that's good. Right. We're starting to believe the bullshit. So be careful about that, guys. Like
don't let this technology be used in a way that hurts us and doesn't help us.
don't let this technology be used in a way that hurts us and doesn't help us.
Yeah, it's interesting.
I mean, security, a lot of the security stories,
just because of the whole fiasco with the U.S. Department of Defense
and Anthropic and all that, a lot of the security stuff happened.
I mean, even HHS, HHS, Department in the U.S.,
is telling their employees to stop using Anthropic Claude. Even HHS, HHS, department in the US,
is telling their employees to stop using Anthropic Claude.
So it's a government ban of one model, right?
It's about control, liability, data exposure,
all that sort of stuff.
Security is popping through at the sort of governmental level,
regulatory level at the moment,
just because of the most recent news stuff.
The individual staff, oh, there's an interesting story.
Armed humanoid robots basically moved from science fiction to procurement pitch
this last week.
A San Francisco startup is pitching armed humanoid robots to the Trump administration.
Yeah, again, what could possibly go wrong? So, yeah, it's interesting how either national security,
individual security, everything's getting kind of pushed around
at the moment in a significant manner.
And then AI use and surveillance oversight stories coming out of New York,
again, like personal security.
It's just interesting, the shift in the stories in the last seven days, how so much is focused on more national security, obviously with the war in the Middle East.
But also it ripples down into personal security areas as well.
One of the biggest attack services that's popping up at the moment is really around
this agentic AI as an execution layer.
A lot of people are really kind of getting super concerned that the ability for hackers to come in and manipulate those agents is just a whole new security attack surface that
we have never seen before.
If you look in the security area of the newsletter today and this last week, a lot of people
starting to raise a lot of issues around that.
And interestingly enough, quite a number of multi-million dollar
investments going into new security companies. So, hey, anyone who's got an idea around security
and Igenic AI, set up a proposal, you'll probably get VC funded very quickly because it's ripe at
the moment and happening very, very quickly. I guess it's it's interesting how again security regulatory stuff we just keep
seeing this bounce back and forth and a lack of speed in many ways around it um but then it seems
to surge forward um that but do you think you know we're seeing this no guard rails and then a surge
no guard rails then a surge again, is this just the current status
quo? Do you think this is a long-term piece as we're shifting down? I mean, I guess the end
question is when is this going to start to settle down, if ever? From what I'm seeing in headlines
and everything shifting, I don't think, you know, I think we may be in for a two-year right at a minimum um with this
level of just structural fundamental civilization change what are your thoughts is it you think it's
longer it will be a little bit more concise for the question that was a big one well you know
we're seeing massive changes across almost every aspect of being a human every aspect of our
government every aspect of enterprise every aspect of business we're seeing you know robots being
deployed in factories in germany for automakers we're seeing that's going to accelerate in the
next two years um we're seeing chinese innovation when it comes to robotics going crazy um they just
one of the major provinces in china just announced that they no longer use coal.
All of their energy is coming from other alternative sources at the moment.
So we're seeing an acceleration in access to alternative energy occurring in other places on the planet than the US.
access to alternative energy occurring in other places on the planet than the US.
We're seeing massive changes in how, I guess, commerce is being done and will be done.
We're seeing the introduction of AI into our politics.
We're seeing, you know, it's like there's nothing that's not being touched by AI.
I don't know any area that isn't really being touched in one way or another by AI,
right? And so everything seems to be in this upheaval. You know, last podcast, we talked about
this heightening level of anxiety amongst almost everyone I talked to, that there's just so much
fundamental change occurring. And I guess my my question is is that just the way it
is for now onwards or i guess will we get habituated to it or used to it um and and i'm my
feeling is i think i think the next two years at least are going to be like this um your thoughts
are going to be like this. Your thoughts?
I think it's going to be a really
tumultuous time, not just in AI,
but the world's in a very tumultuous place, right?
Yeah.
Yeah, I wish I really had a lot more to say about that. I try not to be negative,
but we can see what's going on in the Middle East currently. Everyone can.
We can see what's going on in Europe Middle East currently. Everyone can. We can see what's going on in Europe currently.
Everyone can.
I looked at my wife a few days ago, and I said,
we just witnessed the shooting of Archduke for a minute.
Yeah.
And I believe that, right?
Right.
The world's going to restructure and change.
How and what that means, I don't know, man.
AI is going to be a huge part of it, right?
And not just AI.
All the new technologies that are coming out so quickly.
Robotics and all the other things that we've discussed.
But I think AI like microchip or like HTTPS or like something that like we all use ubiquitously
is going to become just that.
A ubiquitous technology that's in all of our tech.
Yeah. So I think we're coming up to the hour. Let's close out on this because it's like
fatigue that's occurring. So let me close out. I think, all right. So as we land this plane,
right, really, we're landing this plane. Here's what I think this week really felt like.
Right. Really, we're landing this plane. Here's what I think this week really felt like.
I think across the headlines, you know, I think AI stopped behaving like a tech story and in many ways started behaving like a civilization story for real.
You know, we saw power harden, you know, as frontier labs drift towards defense logic and state alignment.
We saw how quickly legitimacy can wobble when users feel like their tools are being drafted
into someone else's agenda.
Then we hit infrastructure,
that part of the narrative that doesn't care about hype,
the constraints no longer imagination,
it's electricity, water permits, grid stability,
it's chips, it's memories,
and it's the humans who wire it all together.
And that's the shift from AI as software
to AI as industry that we're seeing right now.
And it's already touching household bills, community resources, and political patience,
be that as it may.
I think the next thing that came was belief and trust, where we saw emotional undercurrent
is really the loudest.
You know, safety pledges flex, ads are creeping into answers.
The public's starting to do something very rational, like asking who benefits, who decides,
and what happens when this fails quietly or even loudly. You know, trust is no longer a vibe. It's
starting to look like an audit trail. And finally, we saw execution and control inch forward,
almost politely. Memory memory becomes default agents becoming persistent
orchestration is moving into the cloud and the real story is not that ai can do more it's that
ai is being wired into places where it can do things without asking anyone and i think that's
the moment advice becomes action and actions becoming an architecture that we live in so i
but here's the good thing i I think here's the hopeful part.
And I mean, I really mean this.
None of this removes human agency.
It simply raises the price of unconsciousness.
If we stay awake, if we demand transparency where power concentrates, you know, if we
insist that infrastructure costs are paid honestly, and if we treat defaults as moral decisions, then AI can really become a tool for dignity and resilience and shared prosperity, not just massive speed and leverage.
I mean, this week didn't show acceleration.
It showed the exact pressure points where people, citizens, builders, and hopefully leaders can push back, steer, and redesign the system in real time.
You know, we're not passengers.
We really are active and involved participants,
and the participation is still the one advantage that we as humans have
that no model can counterfeit.
That's it for this week in AI.
Stay informed, stay curious.
Come back next Tuesday for more.
Thanks, everyone.
We'll talk then. Thank you.
