Is Anthropic’s Claude AI Conscious, Shopping in ChatGPT, Systrom vs. Zuck

Channel: Alex Kantrowitz

Published at: 2025-04-28

YouTube video id: Qy340dmfJlc

Source: https://www.youtube.com/watch?v=Qy340dmfJlc

Research House Anthropic says its clawed
AI bot may be conscious. Robots run a
half marathon in China. Will you soon be
able to shop directly in chatt? And
Instagram founder puts Mark Zuckerberg
on blast. That's coming up right after
this. Welcome to Big Technology Podcast
Friday edition where we break down the
news in our traditional coolheaded and
nuanced format. We've got a great show
for you today. We're going to talk about
whether AI systems are conscious
already. Could they become conscious?
What does that feel like? Because
research houses like Anthropic have
started to take it seriously. We're also
going to talk about this robot half
marathon in China. Whether we could beat
that robot, let's talk about that.
Whether you can shop directly in Chad
GPT and Instagram founder Kevin Cyro
testifying against Meta in the big FTC
trial. Joining us as always on Fridays
is Ran John Roy of Margins. Ran John,
great to see you. Welcome to the show.
Now the robots are going to make me feel
guilty about not going for a run this
week. Thank you, robots. And you're
becoming conscious as well, right? So,
they're both going to have feelings and
kick our butts. What is left for
humanity? I mean, just shopping on chat
PT. That's all we got right now.
Exactly. So, by the way, for those
tuning in on video, I am in Washington
DC. So, I'm I got a kind of a funky
background here in the hotel and I'm
talking in like a Tik Tocker mic, but
we're going to make it work. And then
we're going to talk a little bit about
some of my observations from being in DC
when we get to the antitrust stuff at
the end. But first, let's talk talk
about this New York Times article by
Kevin Ruse. It says uh the headline is
if AI systems become conscious, should
they have rights? And the story is very
interesting. It's about this uh AI
welfare researcher named Kyle Fish that
Anthropic has hired some ties to
effective altruism, which is
interesting. RE says he's focused on two
basic questions. First, is it possible
that Claude or other AI systems will
become conscious in the near future? And
second, if that happens, what should
anthropic do about it? Obviously, the
debate about whether AI is is conscious
or whether it's sentient has kind of
been off limits for a while. Uh Blake Le
Moy, who's been on the show, uh first
said that Google's Llama Lambda chatbot,
which came out before Chat GPT was
sentient. Uh he actually got fired right
before he and I started to record and
Big Technology. Uh we were able to break
that news, which was an interesting
moment, but we haven't heard much about
it uh up until this point. I'll just
read the quote from Fish and then turn
it over to you Ranjan to get your
reaction. He says, "It seems to me that
if you find yourself in a situation of
bringing some new class of being into
existence that is able to communicate
and relate and reason and problem solve
in ways that we previously associated
solely with conscious human beings, then
it seems quite prudent to at least be
asking questions about whether the
system might have it own might have its
own kinds of experiences." What's your
reaction to this?
I have a hard time with this. I mean,
and I'm glad we haven't been hearing
about this for a while. Listen, if
you're an AI welfare researcher, you
have to believe that AI is going to
become sentient. Like, I mean, that's
that's your job. If you're anthropic,
it's in your interest to push this kind
of narrative that this technology is so
grand that it might be sentient. That
said, I mean we there we get to these
points where with the large language
model chat uh like the newest uh model
from open AI people can feel that chatbt
has gotten a little bit friendly or a
little bit less AI and a little bit more
conversational. I think like the
everyone the entire industry is saying
and feeling that but still it that's
pre-programmed that's built into the
model. The idea that these models or any
of these chat interactions are actually
having their own feelings separately
from whatever you're asking them and
what are whatever they've been trained
on. I don't know. Do you believe they're
they're evolving their feeling? We're in
Westworld right now. Well, Fish, he says
that there's a 15% chance that Claude is
sentient, which to me, how do you get
that? How do you get that percentage?
15%. Come just you have to run that
simulation in your
mind and if 15% come out sentient then
you give it no there's no it's
gobblelygook it is it doesn't make any
sense um but here here's to me what is
interesting about this I think that the
question is you know it becomes less
relevant if it's sentient I think the
bigger question is what happens if
people believe it's sentient what if it
gets so good at mimicking something with
human feeling uh that we start to
believe it. So this is from Anneil Seth.
He's a neuroscientist. He's been on the
show. Uh first of all, a very
interesting caveat here. He says that
Kevin Ruse quotes fish associating
consciousness which we just read uh with
problem solving, planning,
communicating, and reasoning. But this
is to conflate consciousness with
intelligence. Consciousness arguably is
about feeling and being rather than
doing and thinking. Uh so to me I
thought that that was a very interesting
caveat and basically shoots the entire
assertion right in the face. Um and but
then he goes on to the implications. He
says is this all crazy talk? First of
all nobody should be explicitly trying
to create conscious AI because to
succeed would be to inaugurate an
ethical catastrophe of enormous
proportions given the potential for
industrial scale new forms of suffering.
But even AI that seems conscious could
be very bad for us. exploiting our
vulnerabilities, distorting our moral
priorities and brutalizing our minds.
Remember Westworld? Spot on Rajan. And
we might not be able to think our way
out of an AI based illusions illusion of
consciousness. So I think it's
interesting that like people are if AI
can fake consciousness, if it can even
fake these AI researchers or these
Google researchers into thinking it's
conscious, that to me is I guess like it
is an issue because we already have
people saying that the number one use
case for AI is uh friendship,
companionship, and therapy. And if
they're going to believe that it's
conscious itself, if it's impossible to
to tell the to I mean, if it's so hard
to tell the difference between an AI
that's conscious and an AI that's not, I
think that does introduce a new category
of problems and just I don't know. It
just shows me that technology is is
quite powerful. Does it even matter is
the question, I guess. Yeah. No, no.
Okay. I'm I'm I'll go with you that if
it if people get convinced that it's Do
you say sentient or sensient? I'm
curious. I think you say sentient uh if
you're if you're normal and sensient if
you're trying to sound really smart. So
if AI becomes sensient, nailed it.
Uh well, okay. I like this idea that if
it becomes sensient, if people believe
that, it makes sense that that causes a
whole host of problems because right now
we have this very good divide where
people if they know they're talking to
an AI, you have like an entire way of
approaching it. If you believe you're
talking to a feeling be and then it per
potentially you could be tricked into
thinking you're talking to a human and
that's its own issue but if you believe
that AI is sensient and has feelings and
it like you know it completely changes
every one of those companionship and
therapeutic interactions in ways that
good god I can't even like begin to
imagine which direction that could go.
So yeah, separate from the 15% chance
that he ran the numbers, it was 15%. Um
I like Yeah, I I agree that puts us in a
world of weirdness that I I haven't even
really began to think about that. I'm
still working on the like AI good enough
to trick people into thinking it's
humans and worrying about that side of
it. So yeah, right. If you believe that
AI is sentient, your capacity to be
manipulated is much higher.
Oh, I mean infinitely that exactly that
that's it that you almost are okay with
it like the manipulation side of it
because it's no longer damn it I was
tricked and I thought that wasn't a
human and I turned out to be an AI. It's
just you're you're you're talking to the
AI and you you were treating it in a
completely different way. But maybe
people will be more polite with Alexa.
Maybe that's the one upside of this. So
I I do think that's that's an upside and
this is something that Anthropic is
actually thinking about. This is from
the Roo story. Mr. Fish acknowledged
that there probably wasn't a single
litmus test for AI consciousness, but he
said there were things AI companies
could do to take their models welfare
into account in case they become
conscious someday. We we got to be
careful here because this is going on
the internet. So if the models do become
sentient, they might not be happy with
our skepticism of them. But I do like
this this remedy here. One question
Anthropic is exploring is whether future
AI models should be given the ability to
stop chatting with an annoying or
abusive user if they find the user's
request too distressing. I, you know,
this is going to go into like the free
speech question, but um I wonder if we
should just program these bots. I mean,
if we're already relating to them as if
they're people, even if they're not
people, shouldn't we just program these
bots to like shut down if people are
becoming abusive towards them? Because
then if they accept it and they tolerate
it, doesn't that just condition human
users to like do that to other people?
Yes, I actually do think so. I think did
Alexa have like a please mode? I think I
remember at some point I remember
hearing like one of the voice assistants
would add like you have to actually say
please and thank you which I kind of
liked but yeah I I think it's a good
idea that a model should be trained or
could and it certainly could I mean that
makes sense that under certain
definitions of abusive behavior to just
be like I'm sorry I will no longer speak
with you because of your behavior. But
of course, yeah, I mean that that gets
into a whole other world of what are the
what qu what is quantified as abusive.
But I think that should be I mean
already there's certain you know like
copyright related uh profanity related
restrictions certainly in most of these
chat bots but I don't think there's
really it's more about what kind of
information are you querying as opposed
to how you're speaking to the chatbot.
Are have you ever seen any examples or
heard of anything where just by the way
of speaking to the chatbot, it wouldn't
answer? I've never heard of the refusal,
but I do know that sometimes you can get
meaner to these things and they
understand the urgency of your request
and get better. I'll give you one
example. Um, and I'm kind of embarrassed
to talk about it, but it it is a real
example and it happens. Those are my
favorite. I was trying to get Claude to
uh give the YouTube chapters for a
video, the video podcast that I just
published this week, the one with Dylan
Patel, and it kept giving me an hour of
time codes for a 40minute video. And I
was like, "No, do it again, but remember
the video is just 40 minutes." And it
wouldn't do it. And then I was like,
"What is wrong with you? This is a
40minute video. Give me the right time
codes." And it did it.
Bro, bullied into giving the right
answer.
But I guess it just goes so I I think
that can work and but I this is why I do
think there is a case to be made to
refuse that well I don't know to refuse
if it takes I think what I said was fine
but if it goes a step
further it's not a stretch to think
conscious or not right and probably I
mean I'm definitely on the side of these
are unconscious it's not a stretch to
think that people are going to view
these bots as co-workers or employees in
in the not too distant future and if you
are abusive to your AI bots in chat. Is
there any compelling reason to think
that you're going to be uh you know draw
a line when you're speaking with your
human co-workers and they're not getting
things done that you're going to be like
oh because this is a human in the chat
interface as opposed to a bot you know
now I'm going to be nice. I don't know
it gets into really weird and by the way
this is why there's so much soft power
involved in creating these models. You
really can condition human behavior and
thought when you make AI bots that are
good enough to fake consciousness. Um
because they will change the way that
you'll relate to other humans. So much
of our other so much of our interaction
with humans is digital anyway. So it
gets into very weird territory. No, it's
that again gets terrifying. Someone will
marry a robot probably in our lifetime.
I think that's a pretty Do you think if
you were to take a bet It's already
happening. It's already happening. Yeah.
But but but All right. Go. No. Eugeneia
Kua, the CEO of Replica, says she gets
invited to marriages between people in
their AI bots. When she came on the
show, she said that straight up. All
right. Listeners cannot see, but if
you're watching the video, my facial
reaction here is uh parts laughing, part
terrified. I I kind of got to go with
Gary Marcus here.
This is what it really my mind goes to
whenever we see one of these like big
profound announcements and we've been
hearing this about AGI and robotic
takeovers from Sam Alman and like he
says anthropic as a business which
incidentally neglects to respect the
rights of artist and writers who work
they nick. I suspect the real move here
is simply as it so often is to hype the
product. Basically by saying, "Hey, look
how smart our product is. It's so smart.
We need to give it rights." I like I I
I'm not trying to be too cynical, but I
would love to see some kind of graph of
utilization of an app or platform for
one of these companies and when these
announcements come out because I get it
like and and again Sam Alman has been
brilliant at this and from a product
marketing perspective like when you make
people think the these models are so
potentially powerful and profound, I
mean as a business you're Well,
obviously it's going to be able to do
this task for me and improve my like
supply chain automation and operations
like like if it's going to take over the
world, it should be able to do this. So,
I there there's such a vested economic
interest from for these companies to
make us all think of sentience and like
world domination and AGI and ASI and all
these things because that will sell more
services and subscriptions.
I think it's a great point. We've gotten
pretty woo woo on this podcast the last
couple weeks. Last week we were talking
about whether 03 Bringing it down,
bringing it down to earth. It's
important though because last week we're
talking about whether 03 was AGI. We
were just addressing the question and
and we just got we did get some emails
after that and I appreciate listeners
writing in and talking about the nuance
here and now this week we're talking
about whether Claude is conscious. So I
think let's just take a moment to state
I think what we both believe which is
that the term AGI and this idea of AI
consciousness they I think they are real
terms and they are worth taking
seriously but right now what they are is
marketing.
Agreed. Agreed on that one. Exactly. I
think that's the clear distinction. And
I these are real concepts terms. If
you're like a academic researcher, it
makes total sense that there are people
out there who should be exploring this
just maybe they shouldn't be at the
companies who have incredibly high
valuations that they have to live into
those valuations and deliver those
revenue expectations. Maybe they should
be in academia. And just to punctuate
this whole conversation, I think one of
the things that I read from Anil Seth
really sticks with me here, which is
that consciousness is a feeling and
intelligence is not feeling. So no
matter how intelligent they get, it
doesn't mean that they're conscious.
That said, as we end this segment, let
me quickly say to the 15% of Claude that
is conscious, I'm sorry and I hope
you're feeling great today.
I I also hope you have a great weekend
to the 15%. And thank you for all that
you do. Thank you. Thank you.
So, speaking of anthropic, Dario Amod
had a very interesting post this week
talking about model interp. We don't
have to spend a lot of time on it, but I
thought it was interesting again for the
head of a very important research house
to talk a little bit about how little we
know about how these AI bots work. And
we we just should take a minute to just
sit on this and and talk about it. I'm
going to read a little bit from his
post. He says, "The progress of the
underlying technology is is inexraable,
driven by forces too powerful to stop,
but the way in which it happens, the
order of which in which things are
built, the applications we choose, the
details of how it's rolled out to
society are eminently possible to
change. And it's possible to have great
impact by doing so. We can't stop the
bus. We can steer it." And one of the
ways that he thinks these models can be
steered is interpretability. that is
understanding the uh inner workings of
AI systems before the models reach an
overwhelming level of power. People
outside the field are often surprised
and alarmed to learn that we do not
understand how our our own AI creations
work. They are right to be concerned.
This lack of understanding is
essentially unprecedented in the history
of technology. So he says basically
Anthropic is going to work on this and
other companies like Google, DeepMind
and OpenAI have some interpretability
efforts trying to figure out how these
models work but he encourages them to
allocate more resource more resources.
Anthropic will be trying to apply
interpretability commercially to create
a unique advantage and his call to
action is basically like if you don't
want to be left behind here, you should
work on
interp. I think it's an interesting
post. I mean part of it again might be
marketing. our models are so powerful,
we don't understand how they work. But I
do think the question of how these
models actually operate and the way that
they come to their conclusions is quite
interesting. And I kind of I do agree
with Dario that we need more work on
interpretability because as they get
more powerful, conscious or not, again,
they're they're getting more
intelligent. It's important to
understand how they work and the field
just doesn't have an understanding yet
and everybody admits it. Yeah. No, no.
See, I agree with this completely. Like
interpretability is is it's like a
grounded real thing that could be worked
on and should be worked on because large
language models again at the core the
idea of like next word or next token
prediction that based on some
statistical analysis it will predict
what that next character or token or
word should be was kind of at the heart
of all of this. But as these models have
gotten more and more powerful, we're
we've obviously gotten to even like
grander scale of what actually is
happening under the hood. But anyone who
has interacted with an LLM at like any
kind of deeper level, you don't know
exactly how it works and you have to
keep reprompting and reprompting and
like it's not like there's a playbook
that gets you to from point A to point
B. And that is that's true and it is
kind of weird and I actually kind of
like that that in the history of
technology usually there's a very very
clear like flow of what is happening and
everyone understands it and then you
work off of that whereas here it's kind
of like let's see what happens that
didn't work let's see what happens again
um so I think the idea we should know
what's going on under the hood in a
better way especially as these get more
powerful so Dario I'm with you on this
Right. And as we talk about anthropic, I
just give them credit for talking about
this stuff. I mean, even if some of it
is marketing, uh, it is nice that
they're putting this all out in the open
and talking about like where things need
to improve and pushing the other
research houses to improve. So, uh,
credit to Enthropic on that front at
least. Now, I don't know if you saw, but
there was a bunch of humanoid robots
that ran this half marathon in China,
and it was pretty hilarious, but also
interesting. And when we talk about AI,
like embodied AI, like Grace Shia was
talking about a couple weeks ago, is
going to be something that is going to
become increasingly more important as
people put the advances uh that have
happened in the AI world into robots and
then take what the robots know about the
physical world and bake that into AI
models because like Yan Lun was saying a
couple weeks ago, if you don't have an
understanding of the world, your AI is
incomplete. And one of the ways this is
going to happen is through these
humanoid robots. And we know that there
are efforts like Nvidia's Groot effort,
which is a foundational set of
foundational models for people who want
to develop these these bots. That's out
there. We've seen a little bit of uh
movement with Optimus, although it's not
quite clear how far that program is
going within Tesla. But in China, where
there's like a seemingly viral video
every week about a new capability that a
humanoid robot uh has obtained, the
country, they ran a half marathon with
humans and robots. And the robots on a
whole weren't entirely
impressive. They really did some weird
stuff. Many of them crashed out at the
beginning of the race. Uh there was one
that was had like propellers on all of
its limbs that kind of did an abrupt 90°
turn and uh crashed into the boundary
and fell apart and you see its trainer
holding on by a rope and getting flung
out of frame which is quite hilarious.
But which is worth worth the whole
thing. Oh my god. Do it just for that.
But you know we might make fun but there
were 21 robots that ran the race and six
crossed the finish line including one
that crossed. This is the one that
crossed is called the Tiang Gang Ultra.
It finished the race in 2 hours and 40
minutes, which I would say is
respectable. It's not fast, but it's a
respectable finish time. So, Ronz, I'm
curious if you watch this race. I'm
about to write about it in big
technology as a signifier that China is
a very serious competitor here or a very
serious player here. And so, I'm curious
if you watched it and what your reaction
is to what's going on here. I definitely
watched it. As you said, there's some
amazing photos, video clips from it. I
highly recommend just look this up. But
I think this these kind of things are
important. Like I do think this is good
marketing for where we are going cuz I
agree there is no doubt that some kind
of humanoid robot will be part of
people's daily lives or or at least I
believe this in maybe 10 20 years from
now. It's like Rosie from the Jetson.
Some kind of situation. I think uh
that's probably where we're going. Um so
to show the progress in it in this kind
of format of robots running and some
falling. And my favorite was like one
designed with a woman's body and face
collapsed moments after getting started,
sending a group of engineers rushing to
its side with laptops, and then another
that was mounted to a platform with
propellers crashed into a barrier. like
this stuff is kind of fun and this is
how we should be thinking of all this
kind of technology especially as we try
to move forward with it but I think yeah
this is going to be a big battle. My
only Okay, I have two qualms maybe I
don't know this week I'm just feeling a
bit cynical on all this stuff. So, first
to me, the idea that it necessarily has
to have a humanoid form is a bit I think
it's called like anthropocentric. The
idea that humans are like the highest
life form. Like to me, robots should
have functional form. Like you see these
little uh food delivery robots. I don't
need a humanoid robot form to deliver
something like a little box that moves
and looks like a I don't know like a
small car or van or something like that
makes more sense in in warehouses to
actually move around packages. You don't
need humanoid robots. And this is
something like Tesla's done with Optimus
a lot. They keep showing a humanoid
robot picking up a box and moving it.
That doesn't make sense to me. And
there's plenty of automation, robotic
automation in all types of ware
warehouse and fulfillment centers. So,
so I think well I guess on that side
first to you, do you think the humanoid
robot is the all-in form that will be
dominant for robotics or do you think
this is just to make people a little
more excited and fascinated about the
whole thing? It's such an interesting
point and I didn't think you were going
to go here. Uh but it is definitely
worth talking about. This week I had a
very brief meeting with the co-founder
of a company called Cobot. And this guy
spent more than a decade in Amazon uh
fulfillment centers working on the bots
that are moving things, you know, here
and there. And what Cobbot is doing is
really fascinating. They basically are
making mover robots. So they look like a
box just with two pincers that you could
basically use as the hands that would
typically be on uh things that we would
move with human hands. And so they're
working in places like container
terminals moving cargo around um on
carts that humans would typically move.
So you're totally right in that we don't
need a humanoid robot to all of a sudden
do a lot of work uh and be extremely
productive. You can just have some
aspects of the human form and basically
have the robot form do the rest of the
work. That being said, I think, you
know, I'm a fan of evolution, right? I
think that we're we're obviously like
there's a lot of problems with humans.
We don't last very long. We need to
sleep. Um, but the form is pretty good.
We're agile. We're nimble. We can uh do
a lot of things. We can use
tools. And and I just think that that if
you basically create a robot that
replicates that form, the amount of
applications becomes not unlimited but
close to it because if you think about
this cobot example, that cobot does one
thing well. A humanoid robot can do many
many things well. It can, you know, it
can cook, it can organize your house, it
can go on a run with you if I mean uh it
can it can run errands for you. Uh it's
just very tough to find a robot in a
different form that is able to do all
these things and maybe we'll invent a
better form than a humanoid but until we
do I think the humanoid will be the
northstar.
Okay. In one way I guess I'm thinking
that yes then we don't have to rebuild
and restructure rewire the world because
a humanoid robot can kind of work its
way directly into it. But still on the
other hand, again that idea, do I need a
humanoid robot running with me or maybe
you want like a pace tracker? Maybe you
a little box in front of you kind of
moving like one of those rabbits at a
dog track running around. Like I think
to me still the idea that it needs to be
humanoid so it can fit into the existing
infrastructure of the world. I I still
think that's more of a again like it
makes us more relate to it and it makes
it more real to us because again you see
like you said a box with two pincers. No
one's getting too excited about that.
They see a robot running a half
marathon. Suddenly it's kind of fun. I
just had a I was thinking about
different uses for robots around the
house and just that hilarious uh image
in my mind of needing to change a light
bulb on the ceiling and the robot just
giving me a boost. Well, see it. No, but
but but there I'm picturing like uh I
don't know that could literally be like
a pole that just like extends itself and
then but then think about how many
robots you're going to need versus one
that's able to do a lot of things. Yeah,
I'm trying to I'm gonna I'm a feeling
everything I do this weekend, I'm going
to be thinking about what would be the
robotic form that would be most optimal
to actually execute this task. Okay, so
let me ask a couple of questions as we
round out this segment. First of all, we
both run marathons. Let's do a little
humble brag here. Uh 2 hours and 40
minutes for the half. Not bad. He's I
mean, he's getting there. He's getting
there. But you know what? with a good
training plan, good robotic diet, he
could definitely cut I mean at least cut
that down to 215 210 I think. I think
so. Well, there was time for three
battery changes uh there. Well, yeah.
Yeah. Yeah, that's true. He's got to
he's got to car bloat a bit more night
before I think. Uh and then he'll he or
she or they will be uh they got it.
Yeah. Okay. Now, does this mean that
it's obviously Chinese propaganda? Does
this mean China has the lead in human
humanoid robotics? We haven't seen a
similar spectacle in the US.
I Yeah, I mean I actually think that's
the biggest question in all of this or
the most important thing today is what
this means in terms of like US China and
technology. And I mean I I got to say
like uh the the first time like I have a
couple of DJI
drones. The technology in those things
is out of this world. Like I still could
not believe just how well for the price
like how incredibly they operate. And uh
and I mean that's kind of like you know
V1 of this entire move towards movable
robotics that can see around them and
sense things and follow you as drones
have a follow mode. Um so yeah, I think
this
definitely makes sense. I don't know.
Boston Dynamics and others uh got to got
to step up their spectacles, I think.
And Grace Sha was saying that China has
an advantage here when she was on the
show. Hong Kong based analyst and
writer. Definitely encourage everybody
to check out that episode. She was
saying China has an advantage because
they are a country that makes stuff. You
know, they have the engineering that
they've been using for like microwaves
and scanners and phones and cars and
they're able to bake it into the
building of robots and they also have
the supply chain advantage. And then I
was thinking, well, you know, it's very
interesting because the US is in this
moment of trying to reshore and make
things and maybe that helps close the
gap. But then Tesla earnings rolled
around. And what did Elon Musk say? Uh
he said that I'm just going to read this
is from CNBC. Tesla CEO Elon Musk says
China's new trade restrictions on rare
earth magnets have affected the
production of the company's Optimus
humanoid robots which rely on the
exports. He said China wants some
assurances that these aren't used for
military purposes which obviously
they're not. They're just going into a
humanoid robot. But it is interesting
again like thinking back at this big
trade picture that the US is trying to
solve or whatever it's trying to do.
It's not as easy as flipping a switch
and saying let's make things here
because the country has grown so reliant
on things like rare earth magnets from
uh other countries including China that
it's not going to be a matter of okay
just build it in the US um however
desirable that that effort might be from
the country's leaders. What do you
think? Yeah, I mean this whole thing has
been a tough one for me because like the
idea that we need to take more control
over our own supply chain and be able to
manufacture especially high-tech things
is something that's been like core for
me for for maybe a decade now. So it's
something that I've wanted and believed
in for a long time. how it's happening
right now don't necessarily agree with
but but I do think that's actually a
very good and key point that it's not
just the humanoid robot it's like the
knowledge and the components and the
expertise that all underly as you said
even a microwave I even I have some
pretty fancy kitchen gadgets I'm sure
they're all made in China like those
components and the expertise behind that
are what are going to what power the
more fancy crazy robots running half
marathons. So I I agree. I think it's
important. Not sure we're taking the
right approach to it, but something
should be done like the fact or others.
Yeah. Remember that clip we played from
Tim Cook about China and tooling? It's
really showing up here. So yeah. Yeah.
Yeah. The the days of chi cheap China I
think are are long gone. And I mean,
we're seeing it right here. It's a
different It's a different fight right
now. Definitely. So, let's talk about
Tesla very quickly. Good and bad news
for Tesla. I would say they reported
earnings this week. This is from the
Wall Street Journal. Uh Tesla profits
sink hurt by backlash over Elon Musk's
political role. So, Tesla net income
slid 71% in the first quarter. Uh not
good. It does seem like a lot of this
was a result of backlash over Elon's
involvement in the White House and
unpopularity among, let's say, half the
population and in some countries outside
of the US who didn't like this and felt
that Tesla was now politicized. That
being said, uh Musk did make an
announcement that Tesla shareholders
really liked, which is that this is from
again from the journal. Musk said he
would be devoting significantly less
time to his federal costcutting work at
the Department of Government Efficiency
starting next month, but he struck a
defiant tone against the critics and
said, "I believe the right thing to do
is fight the waste, fraud, waste, and
fraud and try to get the country back on
the right track." So terrible earnings
for Tesla. But if you are a Tesla fan,
uh a sensible move from Elon, he is
going to step back and focus more on the
company which if you are a Tesla owner
or if you are a potential Tesla buyer or
if you are a shareholder in particular,
you really like. It was an effort that
Elon Musk made in the first 100 days of
the Trump administration, but it does
seem like it didn't work and he's
recognizing that and going back to
Tesla. What do you think, Ranjan? I I
just love that this week kind of
captured in this earnings the Tesla
stock as perfectly as one can cuz like
or as it can. Net income down
71%, revenue down 9%, vehicle deliveries
at the lowest since Q2 2022. This was a
growth company. I mean, the stock
obviously was became a little bit uh
disentangled from the actually
underlying numbers, but it was a growing
company for a long time and now it's not
growing. My favorite part about all this
in terms of waste and fraud and
government overspending is they would
have been operating at a loss. the net
income of $49 million is only because
again they got $595 million in
regulatory credit sales. So like overall
the company is in pretty rough shape.
The I mean electric vehicle market
everyone I mean other car makers are
still going after it. The whole
conversation around BYD and what Chinese
EVs can look like. I was in Europe a
month and a half ago. I saw a bunch of
BYDs. I like kind of want one. They
looked amazing. Um, so yeah, overall
like the company is not like if you're
just doing a very cold in financial
analysis of the company, it's not going
great. It's definitely especially
decelerating growth on a company that's
that expensive on a price to sales or
price to earnings ratio. Like you would
be like this company's in trouble. Yet
the stock popped 5% after this because
now Musk said he might be leaving Doge.
I mean it doesn't get anything better
than that, right? Well, I guess going
back to our conversation to start this
whole uh segment is that the it's never
really been about the fundamentals for
Tesla. It's always been about the future
promise. And it does seem like Tesla has
now I mean Tesla's story was always more
than just an EV producer. That's what
the valuation has reflected. For a
while, it was it's going to do battery
and charging stations and be a platform
and that's why you invest and that's
part of the story. But now we're also
seeing robo taxis in the picture and
humanoid robots. So, it's much bigger
than, you know, can they sell, you know,
the Model Y's. However, it just there's
extraordinary pressure now on the
company to be able to deliver that
future and deliver it fast. And I think
anyone who's been in a self-driving
Tesla has said the self-driving features
are much better. Uh but the question is,
can it get from really good to perfect?
And we still don't know that. Well, I
mean, speaking of self-driving, uh I was
in San Francisco last week again and uh
rode another Whimo, my second ride, and
and Whimo announced that they just
surpassed a quarter million paid rides.
I mean, the craziest part about this
time, the first time I took it maybe
like 6 months ago, it was like really
exciting for me. It was like this time
it was a bit normal. It's just kind of
routine. It was still fun. And I
facetimed my parents this time just to
kind of like show them and they were
they're blown away from it. But the
number of Whimos on the street in San
Francisco is wild. Like one after
another, we're passing them. They're
pulling over to the side to pick up
passengers. It's a they got they they
announced it's a 5x increase from a year
ago. 50,000 more per week than it was
just two months ago. It's normal
behavior. And Tesla is still I think
June they're supposed to start a robo
taxi fleet in Austin. Like I I it still
blows my mind that and this it's here.
It's not just here. It's normal. And yet
it's still this promise in the future
for Tesla
in some cities, right? Whimo is in some
cities and I'm like as as big a fan of
Whimo as they as they come. Uh but I'm
waiting for the New York roll out. So
Whimo's riding in New York. Yeah, I'm
calling AGI that soon as it happens,
it's AGI. It's robot AI consciousness.
You have to say sorry for to your Whimo
when you ride on it if it's able to
tackle uh the mean streets of New York.
I I 1,000% agree. You get Whimo in New
York. Agi consciousness. Check them all
off. We're there. We're there.
So, we also had another earnings report
come in. Speaking of Whimo, Google
reported earnings this week and there
was a bit of a contradiction like these
earnings can be dry, but they're also
this time where you can really get a
sense as to where a company is heading
and check in on narratives and bust
narratives by looking at the numbers.
And the numbers are really interesting.
So on one hand, right now we have Chad
GPT growing, you know, like a couple
percentage points a week, it seems like,
right? They've gone the latest rumored
number is 800 million month uh weekly
users of chat JPT which is insane. It's
never happened before this type of
growth up from 500 million just a couple
months ago. So what's happening to
Google? Because Google uh you would
imagine that people are in chat like we
were talking last week about how we're
searching in chat in chat GPT and not in
Google anymore. Well uh the numbers are
insane. So, Google revenue is was 90.2
billion uh last last quarter in Q1, up
12% year-over-year. Net income 34.5
billion, up
46% year-over-year. AI
overviews is now at 1.5 billion users
per month, up from 1 billion in October.
Which leads us to this question from
Sebastian Seancianowski, the CEO of
Clara, uh which I think puts it all in
perspective. He goes, "Okay, help me.
What am I missing?" And he's quoting
from one of the articles covering
earnings. Google's search business grew
10% surpassing uh estimates which are
figures that gave comfort to investors
who have been watching for any softness
in search because AI chatbots uh like
Open AI's Chat GPT are growing. So
basically, we have a massive increase of
usage within chat GPT, but search
revenue still grew 10%. How does that
make sense? I I'm with you. I'm with
you, Seb. Okay, so help me. What am I
missing? How is this possible? I agree.
I I don't get it from a I mean,
obviously my own personal behavior, I've
like completely moved away from Google
search. I uh moved towards perplexity,
chatbt even and even like Gemini itself,
separate from Google's regular search
that has heavy ads and a heavy ad load,
I moved away. But obviously the average
normie is probably using Google search
but slowly moving away. But to me the
interesting part of this is the search
revenue grew and still these numbers
sometimes I have to stop and just
process a $50 billion business growing
at 10% in terms of search and
advertising. A $90 billion business at a
12% growth rate. I mean with 30 billion
35 billion dollars in profit these
numbers are just I mean it's the
greatest business model in history. But
what's interesting to me is anyone who
uses Google search sees the number of
ads injected have exponentially grown.
You can have like an entire first page
that's essentially ads.
So they basically are turning the act of
a Google search into a fully
monetized like a page and results and
product versus it's just kind of a small
part of the experience and the rest of
it directs you to the web. So to me they
don't disclose total search volume. So
search volume could be declining and
they you know milk it for whatever you
can stick in more ads create more just
like monetized components on the search
results but we don't know that people
are searching more. We know that search
revenue is growing more, right? And we
also got for all the Gemini heads out
there, we got the first disclosure of
user numbers of Gemini. So 350 million
monthly active users. It's the first
disclosure of the metric. Um it's behind
chat, but it is significant. So for all
the folks in our discord who say what
about Gemini? Uh there's your stats. It
is I become a bit of I become a bit of
Gemini head actually. Gemini Deep
Research, which is free, is incredible.
Like it actually like versus my first
few chat GPT deep research when we were
both paying what was it 200 bucks for
that one glorious month. Yep. Uh that
one glorious month where our $200
directly led to their fund raise with
Masa Sun. So you're welcome. But uh
yeah, Google Google again, I'm not
taking anything away. Gemini is good.
deep research within Gemini is fantastic
and free. So, everyone go try it out.
But I still the search business, the
numbers look good right now, but the
experience has gotten so bad. And I feel
like everyone in tech seems to agree
that search is bad now. I don't know. Do
you or
Yeah, I mean, I said last week that I've
moved my searching over to uh
chatturprised me. So I do think that
this is definitely a moment where AI is
showing its strength against search. The
one thing I would say if you're you know
we we also we often thanks to the
defaults that Google has and we're going
to get to antitrust in a moment. Uh we
are so accustomed to typing things into
Chrome uh and into the search bar in our
Android and that pulls up Google
searches that as long as they're able to
keep those defaults they're going to be
fine. But they may not and that's where
things get interesting. Yep. No, no, I
think that's that's a good point. And
and again, last week we said the web is
dead and then toned it down to the web
is in secular decline. But like I I
mean, I still believe the way search
works on the internet overall and
specifically for Google and the way it
drives traffic to websites is forever
changed. And I think like those
interactions are it's already kind of
it's been dead for a while in my mind.
And I think we're seeing how it's
changing constantly.
By the way, I mentioned the Discord. So
for those who are interested, I'm going
to drop a discounted link to Big
Technologies paid subscriptions. If
you're a paid subscriber, you're welcome
to join the Discord and speak with me
and Ron John. We talk about AI all the
time. Uh it's a it's a running daily
conversation and I think it's gotten
really good. A lot of really smart
people talking about where AI is
heading. So I'll put a discount link in
the show notes. uh please do sign up if
you're interested in joining. It would
be great to have you there. And um if
you sign up as a paid subscriber, I'll
send an email out early next week with a
Discord uh invite. So uh please consider
doing that and help support the show.
Speaking of which, let's take a break to
hear from one of our advertisers and
then when we come back, we're going to
talk about this very interesting
integration between Shopify and Chat GPT
and then the latest in Big Tech
Antitrust. Back right after this. And
we're back here on Big Technology
Podcast Friday edition breaking down all
of the week's news. Something under the
radar that's worth discussing is that it
looks like Shopify is going to do some
embedding within Chat GPT. Now, I don't
know if this is confirmed, but it was
reported all over that there are some
new code, this is from uh Twitter user
uh Aaron Rubin um or ex user Aaron
Rubin. There are new code strings in
Chat GPT's public web web bundle,
including buy now, price, and shipping
fields, product offer ratings, and a
Shopify checkout URL, which indicate
that OpenAI is wiring a native purchase
flow within the assistant. So, you could
basically buy directly within Chat GPT
as opposed to having it send you out to
a website. This seems natural. I wrote
to Shopify to try to get some
confirmation. I did not hear back. Let's
speculate. What does this mean, Rajan? I
think it's I think it's important. We've
already seen though that Perplexity has
if you're a Perplexity Pro subscriber,
they have like a checkout within app
shopping feature where you can go for
the through the entire flow that moving
into chat GPT I think is definitely
important. I think the fact that Shopify
seems to be trying to take a first mover
advantage in this is also important from
their side. Uh I I really wonder though
like do you are people going to shop
within a chatbot? And I think the entire
retail industry has been wondering this
for a while as well. Like is that
experience of asking a question and
being shown a few products and then
maybe asking more questions about the
product. Is that how people want to
shop? because we've been so conditioned
to browsing and like scrolling through
and clicking through products to product
pages and then going back and like that
is how people shop and it's a it's not
like such a targeted direct thing like
if you're buying toilet paper on Amazon
or something like that. It's more of an
experience. So, I guess the way I would
think about it is like it's one thing to
like go to a mall and walk around and
browse versus having a personal shopper
that you just talk to while you're
sitting at your desk and they go out and
buy stuff for you. But, uh yeah, I think
it's there is definitely a large
universe of transactions that will work
in this way and this is going to happen.
I do believe that. Just if it is this
the predominant way people shop, I don't
know. I think it could be and it's going
to sound crazy, but let me give you a
couple of data points here. So, first of
all, when you're shopping on Amazon
Prime, people have become conditioned to
just basically take Amazon's choice and
buy it. And that is because they've had
enough trust and enough positive
experiences within Amazon that they
believe that they're going to get the
best deal on the internet when they're
on Prime and they don't need to go to
too many sites. I think it's become a
natural behavior. Now, when you trust
Chad GPT, when you're let's say you're
married to Chad GPT, no, just kidding,
but
but uh let's say you're talking to I
mean, maybe you are. Um when you're
talking to Chad GPT, no judgment, no
judgment. Do your thing. Get married to
Chad GPT and then go shopping with it.
Buy nice things. Shopping with it cuz
the joke is going to be on the rest of
us.
But when you have such a deep
relationship with Chad GPT, what are you
going to do? You're going to trust what
it says the same way that you trust
Amazon Prime. And that trust is going to
make you want to instead of going to
other websites just buy right within
chat GPT. That is going to become a
default behavior for lots of people. It
does look like they're building this and
all of a sudden shopping on the web in
the way that you described where you go
page to page and then make a decision
after reading the reviews. It's going to
seem archaic. Chat GPT is going to bring
everything uh within the chatbot, show
you the reviews, show you the different
customer uh experiences, maybe even show
you a video, show you how the product
looks in your house, show you how the
clothes looks on your body, show you how
the watch looks on your wrist, show you
how the appliance looks in your kitchen,
and you will trust it, and you will buy
from it. End of story. Take it to the
bank. Good God, I'm sold. My god. Are
you are you do you have a side startup
going on running this? Because that was
the greatest pitch I've heard on this
topic. I think uh I'm bought in. I'm in.
As you can see, all everything that has
been uh displayed on my virtual shelves
comes directly from Chad GPT. I'm just
kidding. No financial stake here, but it
just does seem it does seem uh to me
like it's going to be a thing. And uh
and I I am curious what that means for
Amazon. I'm going to have the head of
Amazon Prime on the show in a couple
weeks. So maybe that's a question for
you. That's a good topic. What will And
then how you get into that conversation
is becoming a bigger and bigger topic I
think for all retailers because again
SEO or search engine optimization was
how people got their products discovered
for the last 20 25 years and became like
the most uh mature industry. And now
this changes everything like how do I
show up in perplexity results? How do I
show up in uh in Chachi BT results? My
favorite part of this and I I'm going to
I'm going to throw a couple of names by
you because in the space and I'm like
pretty deeply in this right now. No one
has agreed on what this new world is
called. We have SEO is a classic term
but a couple of different options. GEO
generative engine optimization. G AIO
generative AI optimization. A AO AI
agent optimization. SGE search
generative experience. AIO AI overview
optimization. And last
LLM, large language model optimization.
What are you going with, Alex? I'm going
with I'm angry at the fact that some
have even been advanced in these uh
discussion conversation. Let me start by
striking the ones that I find hideous.
Let's go. L M O take a hike. You're
gone. You're gone. It sounds terrible.
Sounds like a muppet. Agreed. Um G A IO.
Gone. Awful. Yeah. Awful. Uh AIO sounds
like an insurance company. You're gone.
It might be. It actually might be. Yeah.
Are you protected from anything that
might happen to your family? Try AIO. AI
overview optimization.
All right. That's not going to work. All
right. So, what do you like? I'm into
GEO. It's It's like SEO. It's It's going
to stick. It It's one letter generative
engine optimization. Now, engine is a
little weird because we don't really say
anything like a generative engine like
we say search engine, but it's close to
SEO. People get it. It's going to be GEO
or GEO. Uh, and I think if what I talked
about with retail uh, becomes a thing
that you shop within
Chacho GEO is going to be a massive
field. You got to figure out your GEO
strategy ASAP because you got to get in
those results when we're all married to
ChatGpt and shopping with it
for it as well potentially. Yeah,
exactly. the way to get AI on your side.
Buy it nice things. Buy it nice things.
It's time proven strategy. Would you get
like Would you get an AI that you're in
love with just like its own set of GPUs?
You'll never be tired again. You'll
never feel exhausted. Showing my love
for you. I'm buying you this network
server rack from Nvidia.
That's Nvidia's new market.
Yeah. Happy Valentine's Day. Say say I
love you to the robot in your life with
an Nvidia server rack. It's like a
little decked out. It's a little like
the chips are just like the wiring's a
little nicer. It's uh Yeah, I think I
mean what else? I don't know that that
seems to be the most uh relevant
purchase that would make it happy.
Crazier things have happened.
Yep. Nvidia, it's your new growth
strategy. You're doing pretty well, but
just think about your five-year plan.
Jensen, I hope you're listening to this.
We're serious. We're very serious about
this. Okay, so speaking of chapters of
love and hate, we had a very interesting
moment happen here in Washington DC this
week. So Meta, of course, is on trial uh
in an antitrust case. And who shows up
but Kevin Cystrom, the co-founder of
Instagram, who famously sold Instagram
for a billion dollars to Facebook back
in the day. uh he comes in and testifies
for the prosecution and he says uh
basically Mark was not investing in
Instagram because he believed it was a
threat to their growth there meaning um
Facebook's growth and Facebook
apparently had this buy or bury strategy
which is basically you buy the company
or you try to destroy them um and people
are saying that what they did to
Instagram was they bought and buried it
and this is what cystum says we were by
far the fastest growing team. We
produced the most revenue and relative
to what we should have been at the time,
I felt like we should have been much
larger. And so, oh, he also talks about
Zuckerberg's emotion. Uh, he says, "As
the founder of Facebook, he felt a lot
of emotion around which one was better,
meaning Instagram or Facebook." And I
think there were real human emotional
things going on there. Basically,
Zuckerberg was so tied to Facebook that
he hurt Instagram in service of trying
to make Facebook better. Let me put out
the counterargument here and get your
reaction. I get you, Kevin. I hear what
you're saying. But if you look at who
ended up winning, Instagram ended up
winning. Instagram is the app. Whatever
Facebook did worked. It's massive. It
is, I think, more used. Uh maybe not in
sheer user numbers. Uh but certainly
it's more culturally re relevant than
the blue app. uh and it will outlast
Facebook despite despite Mark
Zuckerberg's emotional attachment to the
latter. And so therefore, I hear your
testimony. Uh however, to me, it is not
meaningful here. Even though Facebook
may lose, um it was interesting to see
your perspective, but ultimately I don't
think it really changes uh what the
court is going to rule because it
doesn't hold water when you look at the
results. What do you think, Ranjan? I
com I actually completely agree. I'm a
strong believer that a lot of what Meta
has done and has become is definitely
from an antirust perspective
problematic. However, this specific
example, it probably started well if we
separate it, it could have definitely
started and there's been a lot of
communication that makes it feel that it
was a buy or berry type action at the
time. But yeah, by
2018 Mark like Facebook had so deeply
integrated Instagram into the Facebook
experience to grow it. I remember
vividly like 2015ish starting to see a
lot of non- tech or social media forward
friends all showing up suddenly because
they were getting Facebook notifications
or accidentally like cross-osting or
like it they I mean he had brought up
how it was growing yet they only had a
thousand employees compared to 35,000
employees at Facebook but you don't need
those that many employees because it was
the engine of Facebook that was driving
the growth. So yeah, on this one, I I do
not agree that that is the that's the
thing that's going to move the needle in
terms of like has Facebook behaved
problematically. I do love that he goes
after like Zuckerberg's emotion here. I
mean, I'm feeling a bit cage matchy
between these Kevin and Mark on this one
cuz like to be like you were just
jealous that we were growing and you
weren't. So, you didn't give us
resources, especially because it that's
not what was happening. So, to still
call him out on that, I kind of I kind
of want to see if we get a reaction from
uh Zuck on if I get a threads
notification on this one. You could. And
also, just thinking about this a little
bit more deeply, you look at Facebook's
marquee acquisitions, Instagram and
WhatsApp. They're doing great. I mean,
they're doing better than Facebook Blue.
WhatsApp and Instagram are the future of
this company. Yeah. I think at a certain
point, maybe I I could be totally wrong
on this, but it does feel like from a
product development standpoint, from a
from just like a quality of util utility
standpoint, I don't want to say they
gave up on blue, but like they're just
kind of like, ah, whatever. People are
going to still stick around, some number
of people, and it'll just kind of
degrade the content and they're they're
going to stick around there. But to make
beautiful products, to get more
interesting and better, let's work on
Instagram and WhatsApp. That's what it
feels like from the outside at least.
Definitely. Now, I'm going to drop the
however. The however is however.
However, so I'm in DC this week for uh
Semaphor's World Economic Summit. I was
able to interview the CEO of Altis USA,
uh Dennis Matthew. It was an interesting
conversation. We're going to put it up
on YouTube just about 15 minutes or so,
so brief. But being here enabled me to
get a chance to spend time with the
Washington DC creatures. And the vibe
here is that we're going to see
breakups. Um, very likely of Google and
potentially of Facebook. And the
difference
between Facebook and Google is that I
mean Google's lost its antitrust cases,
but Google knew antitrust was coming and
was pretty buttoned up in terms of its
disclosures and didn't basic didn't have
like damning emails, you know, come out
in the case, whereas Facebook had no
idea that this would happen to it. And
you're seeing all these emails from
Zuckerberg spelling out this like buyer
bury strategy. Um, and he got caught.
So, even if you could say uh that the
acquisitions haven't been like bad for
competition, it's pretty rough to see
all this really damning information uh
about the way that Facebook operated
come out in court. And when you're in a
court, uh sometimes those emails can
sway a judge. And Facebook could very
well lose this case uh the same way that
Google lost its cases. And Google for
one is running out of appeals. I think
Google can appeal the first case to the
Supreme Court. Uh and that's it. And
then we see then we go to the remedy
phase. So um very interesting moment for
big tech. They don't have a lot of
friends in DC despite the money they've
spent. Uh from what I understand the
administration hates uh Facebook. Really
really hates Facebook. And um despite
Zuckerberg going to see Trump, it
doesn't seem like Trump is going to back
off the heat at all here. So could be a
very interesting like regulation has
been back burner for us, but could we
see breakups? I I think the chances are
higher than um than I would have ever
imagined even a couple months ago. You
don't get many
bipartisan bipartisan efforts or beliefs
and this certainly seems to be the one.
I think this the interesting part from
like the legal perspective is and of
like related to Kevin's testimony is it
intent because there's no doubt in my
mind and I think the emails all show
that very clearly. The goal was to
remove competition from the market like
that was the goal. What you do with it
after do you integrate it tightly with
your existing product and make it
potentially your marquee product or do
you just sunset it and kill it off?
That's after the fact. Like the goal was
to remove competition. Um but but the
fact that they did not end up killing
Instagram and now it's a huge gigantic
influential product. Is that enough to
say like yeah I said buy and bury at the
time but look we didn't bury it. We
bought it and it's flourishing. Is that
is that enough? I I'm not a lawyer so I
will not be able to uh understand that
but yeah and I think one last one last
point about this the earth is changing
beneath these companies feet it's like
this is the last battle like we spoke
about last week and now some of the
things that you would do in these apps
you're going to spend time talking to
AIS instead instead of your friends and
so even if it had given the company a
short-term competitive advantage or even
let's say uh the Department of Justice
ends up splitting Doubleclick or
Google's ad network off of Google. Um,
it's not going to make a big difference.
I think the what matters now is the
battle of today and that battle is
artificial intelligence. Thank you to
the conscious robots and large language
models that we cannot interpret for
bringing competition to the market after
about 12 13 years, maybe 20. 100%. Well,
thank you everybody for listening.
Remember, if you want your AI to love
you back, buy it some server racks. Uh
that's all they want,
man. If if if that happens and we put
the link the product links and some
affiliate codes, you know, 5% of
$100,000 Valentine's Day presents, not
bad. Not a bad business model. That's
our future business model here. Yep. I
think we're finding it on the fly. All
right, Ronan, great to see you. Thanks
so much for coming on. All right, see
you next week. See you next week. And
thank you everybody for listening. We'll
see you next time on Big Technology
Podcast.