October 2, 2026 • 39 mins
Will frontier AI labs really police themselves, or will the FTC pull rank? Reed Albergotti (Semafor), Margaux MacColl (Vanity Fair) and Nitasha Tiku (The Washington Post) dive into tech’s big week in Washington to tell us what to expect from evaluators, who the evaluators are, and which tech leaders came out on top at the boys’ lunch at the White House. Then, our nightlife correspondent reports back on the pro-data groups trying to party the woke out. Finally, remember Yahoo? Marissa Mayer is back (again!) with a new app that leverages your camera roll… but does anyone want it?
See omnystudio.com/listener for privacy information.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:11):
Natasha, Reid, Margot, we've talked about social media bans and technology.
And this is, of course, a podcast. So I was
absolutely charmed this week by a revelation that middle schoolers
are using the comments of random NPR podcasts on Spotify
to chat. Here's a clip of Ira Glass talking about it.
Speaker 2 (00:31):
Kids would name their playlist, chat here or something like that.
And then there'd be some podcast. And everybody would go
to that podcast and talk in the comments. She says
it was maybe 20 kids in the core group. Lots
of theater kids, she says. Mostly girls. Most of them
with strict parents who didn't let them on regular social media.
And can I ask, why did you guys pick NPR
shows to be the ones where you went to.
Speaker 3 (00:53):
In the comments?
Speaker 4 (00:56):
I think we just looked for podcasts that didn't have
many comments.
Speaker 3 (01:02):
I see.
Speaker 2 (01:02):
So you picked NPR because it didn't seem very popular.
Speaker 5 (01:07):
I've been feeling bad for Ira Glass all week because
of this.
Speaker 1 (01:13):
Welcome to Tech Stuff. I'm Oz Veloshian, and this is
The Week in Tech, where I'm joined by the world's
most plugged-in reporters to break down what's really happening in
tech right now. Today, we're joined by Natasha Tiku, tech
reporter at The Washington Post, Margot McColl, contributing writer at
Vanity Fair, and Reid Arbogati, tech editor at Semaphore. Welcome all.
Speaker 3 (01:33):
Great to be back.
Speaker 5 (01:34):
Hi, thanks for having me.
Speaker 1 (01:36):
Margot is a second-time Tech Stuff contributor, so special welcome
back to you.
Speaker 6 (01:39):
Thank you.
Speaker 1 (01:40):
So last week we were all fielding calls, or at
least a read was, from our mothers about the likelihood
of the AI apocalypse. I think the take of the
panelists was we're safe through the end of Q4 at least.
Dari Amode and others are persisting in their campaign to
get the government to intervene to pace the frontier, but
falling increasingly on deaf ears. The president has been averse
(02:02):
to regulating AI, or superintelligence as he prefers to call it,
And last Tuesday, he gathered the heads of the frontier
labs and tech firms in Washington for a kind of
working lunch that looked uncannily like the Last Supper. Natasha,
what happened?
Speaker 5 (02:19):
Well, everyone had lunch in the East Room.
Speaker 7 (02:22):
We got a readout beforehand of the layout of where
everyone was sitting.
Speaker 5 (02:26):
They discussed super intelligence.
Speaker 7 (02:28):
And then we had, I think, like one of the
world's most interesting films on the North driveway afterward, where
you saw various male egos in repose.
Speaker 5 (02:39):
And in flux.
Speaker 7 (02:41):
It was like Dario Amadei, Jensen Huang, Mark Zuckerberg, Elon Musk,
David Sachs. And they signed an accord, a White House super,
you can't see my scared clothes, but a White House
super intelligence accord where they agreed to some beautiful self-regulation.
including third-party audits. Yeah.
Speaker 1 (03:03):
Tremendous self-policing, in other words.
Speaker 7 (03:05):
Yep, exactly.
Speaker 3 (03:06):
It was more like a friend DA, right?
Speaker 6 (03:09):
What did he say? It was like morally enforced? He's like, no, guys,
they're going to do it.
Speaker 3 (03:13):
It's a gentleman's agreement.
Speaker 1 (03:15):
It was literally a gentleman's or a men's agreement. They're
boy brothers now. Natasha, I want to ask you a
bit more about the content. But first, I mean, take
us through the dynamics of the lunch. Who is that where?
And then the jostling on the lawn afterwards.
Speaker 5 (03:31):
I guess I was most fixated on on the lawn afterwards.
Speaker 7 (03:35):
I mean, there was just this like one minute segment
I was semi obsessed with because, you know, all the
questions are going to Dario, who's.
Speaker 5 (03:43):
Like ready to hold court, you know, like as usual.
Speaker 7 (03:47):
Then he realizes Trump is on his other side and
he's realizing, oh, and Trump is not liking all the
questions are going to him. So he He literally does
the like Homer Simpson into a Bush meme like behind
some assorted Republicans. And Mark Zuckerberg and David Sachs are
like smirking and laughing to each other in the background.
(04:07):
Like Mark Zuckerberg looked so happy to be part of
the cool kids club. I've just never seen him so
like in his element and pleased.
Speaker 5 (04:17):
And then he did the synopsis of the accord.
Speaker 7 (04:20):
And I mean, there were some great bits later as well,
like Trump saying that nobody's ever heard of Sundar Pichai.
Speaker 5 (04:26):
And he was like, his name is Sundar. That's all
you need to know. He's a monster.
Speaker 7 (04:32):
But it was really that Dario, Zuck, Trump, Sachs nexus
that was, I don't know, I thought it was very
rich text.
Speaker 3 (04:40):
Well, I saw some reporting on what was discussed at lunch. And,
you know, Trump apparently likes Dario now. He's cool with Dario.
Speaker 1 (04:48):
We had dinner, right?
Speaker 3 (04:50):
We had dinner. Dario's great. Although Trump had to, like,
he started it off with, like, I made him wait
for me, basically. Like, you know, we were supposed to
have dinner at 7 and I arrived at 10. And, like,
But then he's like, of course, in typical Trump fashion,
like everybody he talks to, he likes it. He's like,
I like the guy. He's great. Like, not at all
like what he's portrayed like in the media, you know. Meanwhile, like,
(05:13):
it's literally you portraying him that way. But nobody has
reported on what they were stickering about during the famous video, Natasha,
have they? Like, what were they actually saying? That's what
I want to know.
Speaker 7 (05:25):
Oh, I'd have to check the TikTok lip readers. Like
they must be on it. I just assumed it was
about Dario trying to say, trying to like, you know,
walk that tightrope.
Speaker 1 (05:39):
That reminds me of being like a 16-year-old at a wedding.
It's just awkward. Yeah, you mentioned Homer Simpson, but I
kind of, the stumbling into the bush awkwardly is very resonant.
Speaker 3 (05:47):
I mean, I felt for him.
Speaker 5 (05:48):
It was relatable. Like he was like, uh-oh.
Speaker 1 (05:52):
Margot, did anything about the summit catch your eye?
Speaker 6 (05:55):
I don't know.
Speaker 4 (05:55):
I mean, I'm still fixated on the anthropic S1 that
leaked where it's like going through all the risk factors
that are like normal. And then it's like, and it
might blackmail governments and like destroy the world. Like, I
love that they get to have like a boys lunch
and like. Are they going to ruin the world?
Speaker 8 (06:12):
Maybe.
Speaker 6 (06:13):
But they also get to have friends for the first time,
and that made me really happy for them.
Speaker 3 (06:17):
When do we get to say they're wrong?
Speaker 1 (06:18):
Well, maybe the evaluators will answer that question. Natasha, what
are evaluators, and what have you been reporting about them?
Speaker 7 (06:25):
Yeah, so I've been reporting on this third-party evaluator-auditor ecosystem
that is all of a sudden, you know, like the
agreed-upon first line of defense system. for all the harms
we talked about last week, you know, from existential risk
to like teen mental health. So this is really interesting
(06:48):
because this ecosystem, I think, was already growing. A bit nascent,
but there was a lot of money and interest going
into these evaluation startups. Like now they're kind of national names, Meter, Apollo,
Redwood Research. But, you know, that Particular part of the
ecosystem was already getting good funding and it's like considered
(07:11):
very prestigious. I think, you know, if you're going to
leave a lab, a frontier lab, then like it's a
great place to go. But then there were also companies
like Val's AI, which has just raised from Andreessen Horowitz,
and they previously raised from 8VC. And I think they
have been thinking a little bit more about like public sector,
you know, working for the government, making sure they have
(07:33):
a bipartisan kind of appearance, working better with DC. And
then you have the like Accentures and the Ernst & Young's,
which I see are now advertising for jobs on 80,000 Hours,
which is an EA jobs board. So yeah, everyone is
just like, I think seeing that legislation likely won't get passed,
(07:54):
but this is something that the companies agree on and
that there's already funding for it. And there is a
deep need even just to like vet the basic claims
of the companies because otherwise it's like, yeah.
Speaker 5 (08:07):
It's a little.
Speaker 1 (08:07):
Venture backed compliance.
Speaker 3 (08:09):
It's very similar to the big security boom that happened
like a decade ago when, I don't know if you
remember this Natasha, there were just like 8 million security
startups that did like telemetry. They essentially like all did
the same thing. And then every company in the world
would like hire them, of course, just usually as a
pilot and not giving them very much money, but like,
(08:30):
then they'd get all this data, which turned out to
not really be that useful in cybersecurity. And a lot
of them just like disappeared. I feel like it's a
similar thing, except it took, it took like decades for
that to materialize after the internet and like in AI fashion,
everything happens so quickly that this is like popped up
overnight and it will probably lose its mystique very quickly
(08:51):
as well as my, my prediction.
Speaker 1 (08:53):
Natasha, why is Mita the kind of household name? And
then I'm going to ask you to imagine that we
are three venture capitalists. Like what, what is the investment
pitch to invest in evaluators?
Speaker 5 (09:04):
Okay, I'll answer the easier one first.
Speaker 7 (09:07):
Meter is the company that, the nonprofit that was like
approved by OpenAI to do the investigation into the Hugging
Face incident. They worked with Redwood Research, but it's really
been the folks at Meter who've been the most high profile, like,
you know, doing the three-hour.
Speaker 5 (09:22):
Dwarcash podcast, writing their own substacks.
Speaker 7 (09:25):
In terms of the pitch, I think it is, okay,
there's really no way for anyone to know what they're
getting from these AI models. Like one of the folks
I talked to who's head of public sector for Val's AI,
I mean, he used to work at the Pentagon in
generative AI. And he was saying like, these are agencies
that have testing capabilities, but there's just, was nothing in
(09:48):
place for generative AI models. And because of everything happening faster,
people are buying these products without having really tested it
in Warfield to the point where this prioritizes company is
doing like a test of does Claude or other chatbot
models actually follow the rules of war. Like this is
not how things would normally be unfolding. There would be
(10:11):
like a very rigorous Pentagon testing process, but.
Speaker 1 (10:16):
Is it VALS, like V-A-L, like VALS a guy?
Speaker 6 (10:19):
VALS like evaluation.
Speaker 5 (10:21):
Like evaluation.
Speaker 6 (10:22):
Oh, I see.
Speaker 1 (10:23):
Okay, sorry. Was the substance of the joint commitment on
frontier responsibilities basically, we'll use evaluators? Is that a fair summary?
Speaker 7 (10:32):
No, I think they had four steps. It was basically
like internal controls and some kind of external evaluation that
your controls monitoring and like safety checks. methods are working
and there was something about internal.
Speaker 1 (10:48):
Risk is it conceivable that the lead investor of the
evaluated company and the frontier company could be the same
person called mark andreessen yeah aren't all.
Speaker 4 (10:57):
The employees these places also from the frontier labs like
isn't it a very like co-mingled uh employees.
Speaker 7 (11:04):
Yeah like there's the mark andreessen valves that's a very
small minority i would say most of it is uh
Nonprofits funded by open philanthropy, staffed by former frontier labs
like Meter, for example, was initially started by Paul Cristiano,
who is like an early and very influential open AI employee.
(11:25):
And it's his wife who did the investigation. And she
used to work for open philanthropy for a decade. You know,
so there's a lot of overlap. And Paul recently joined
the board of the OpenAI Foundation, something they had to
add to the report as a footnote. So that's really
been seized upon. Like the New York Post is having
a field day.
Speaker 5 (11:46):
I mean, they are eating well.
Speaker 1 (11:49):
I know we've got to come to the Post because
they've been really leading the story on tech regulation this week.
They had a huge scoop, an exclusive on the FTC
and AI regulation.
Speaker 5 (11:58):
Yeah, it was an exclusive.
Speaker 1 (12:00):
So what's the story and why did it end up
in the Post?
Speaker 7 (12:03):
Well, I think the Post has really been covering stories from,
I would say, the tech right. And this one was
from the Trump admin FTC, which said it is targeting OpenAI, Anthropic,
potentially some other companies, and Meter, according to the New
York Post. And, you know, the FTC confirmed this. But, yeah,
(12:25):
the timing of it was quite interesting because here they're
having their little accord, boys' lunch event.
Speaker 5 (12:32):
And then this news comes out.
Speaker 1 (12:34):
So this was already, was in the works. But what
exactly are the FTC doing? And are they opening a probe?
Are they using existing regulations? Are they writing new regulations?
Speaker 5 (12:44):
No, they're not.
Speaker 7 (12:45):
I think the idea was to demonstrate this talking point
that we can use existing regulations to make sure that
people are still kind of towing the line on that.
That's something we heard a lot from Jensen Huang. That's
something we heard from David Sachs.
Speaker 5 (13:01):
And others.
Speaker 7 (13:02):
And, you know, I think this is supposed to show like, hey,
the FTC could have done this with or without the accord.
It can do it with or without any new laws passing.
And so we don't know yet whether it's is it
an antitrust probe? Is it about some like collusion around
security or just details around concern for the public?
Speaker 5 (13:21):
We don't know yet.
Speaker 9 (13:22):
Yeah.
Speaker 3 (13:22):
But but there is a concern, I think, that these
companies all get together and they say, OK, let's all
agree to not push the frontier and Because we've got
a great business right here. This is a high cost
for us. So let's, under the guise of safety or
whatever it is, let's not spend a bunch of money
training some massive new model. We can make a bunch
(13:43):
of money on the existing ones. That's the kind of
agreement that I think people are worried about. And it
gets very complicated because you can make a safety argument
for doing that.
Speaker 1 (13:54):
Didn't the chairman of the FTC level that exact accusation
that the frontier labs are pleading for regulation now? to
insulate themselves from possible competition.
Speaker 7 (14:04):
I mean, Dario said that explicitly. He asked for a waiver,
an antitrust waiver, in his, like, Pacing the Frontier article.
Speaker 3 (14:12):
So many people have leveled that accusation, including in the
White House, like, on the record, off the record. I mean,
David Sachs is constantly saying it, you know, in his
X feed.
Speaker 1 (14:21):
You know, the chairman did say it in the article.
I want to close out this first segment with a
Tech Stuff exclusive, though. We have a forthcoming show at
Kaleidoscope called The Builders, hosted by Walter Isaacson, where Walter
is in conversations with some of the people we talk
about often on the show. One of whom was Peter Thiel,
(14:42):
whose comments we published earlier this week exclusively in the
Tech Stuff feed. I want you all to.
Speaker 3 (14:46):
React to it.
Speaker 8 (14:48):
There is a certain type of high-tech totalitarianism that's being
tried out in China. It's not as violent as Stalin
or Mao or or Hitler, you don't need to kill
as many people, but it can go to the inner
recesses of your thoughts and it knows more about you
than you know about yourself. And there's something about that
(15:10):
that we really don't want that to be the model
that wins.
Speaker 9 (15:13):
And your old friend, David Sachs, and of course, President
Trump are now just very recently saying, because of the
competition with China, stop talking about pausing AI, even if
you think it could be an existential threat. Do you
agree with that?
Speaker 6 (15:31):
Yes.
Speaker 8 (15:32):
I broadly agree with that. I've been friends with David
Sachs for many decades. The thing that makes me less
comfortable making those kinds of arguments publicly is that on
some level, it's hard when the big companies themselves are
telling you to stop and to slow down.
Speaker 9 (15:52):
So that's caused you to have some pause because you
were not in the pause button camp.
Speaker 8 (15:57):
Well, I'm still not in the pause button camp. And
I'm saying this on your show that I'm not in
the pause button. We need to hit the accelerator as
hard as possible.
Speaker 6 (16:06):
Guys, I have so many thoughts. I think it's like.
Speaker 4 (16:11):
Impossible to overstate how much men like Peter Thiel and
people in that circle believe China is like almost intrinsically evil.
Like when I talk to these men, in public and
in private they truly believe that like china like might
be the antichrist like i talked to one guy who's
building a fully autonomous fighter jet and he's like china's
(16:31):
the antichrist i'm making this for the end times.
Speaker 1 (16:34):
And you believe based on your conversations that this is
a deeply held belief rather than a convenient.
Speaker 4 (16:40):
Uh narrative i think that these men find a lot
of of god and meaning where money.
Speaker 3 (16:47):
Is do you know what i mean well i think
they're taught They think the Communist Party is evil, right?
Not like necessarily all Chinese people, right?
Speaker 4 (16:55):
Not necessarily all Chinese people, but I think with these, like,
you know, I go to these parties, these guys, and
they're quite racist, right?
Speaker 6 (17:01):
And so it's a little hard to disentangle it.
Speaker 1 (17:04):
This is a perfect segue. We're going to come to
your story after the break. When we come back, the
cool kids in Washington have hatched a new plan to
win hearts and minds over to the pro data center side.
One viral particle link at a time. Welcome back, and
(17:33):
welcome back, Margot, contributing writer of Antifair. Very happy to
have you back with us on Tech Stuff. We like
talking about what's going on in Washington and tech lobbying
on this show, but we don't often cover the party
scene in DC. What exactly is Data After Dark? And
is it as ominous as it sounds?
Speaker 6 (17:51):
Yeah, quite the weekend.
Speaker 4 (17:52):
There were two pro-AI parties thrown by very like tech,
what are tech bro, tech people. The big one was
Data After Dark. It was thrown by a Pirate Wires writer,
which is like a tech kind of conservative outlet, and
some like libertarian think tanky guys.
Speaker 6 (18:08):
And they threw this big party.
Speaker 4 (18:11):
The Partiful was talking about they need to fight the
bipartisan psychosis against data centers.
Speaker 6 (18:17):
And it was as fascinating as you might think.
Speaker 1 (18:20):
Well, tell us about the invitation, which had kind of
its own life before the party even started, right?
Speaker 4 (18:23):
Yeah, so it was recently just going to be a
house party for these, like, three guys that are friends.
And they had a Partiful that, again, talked about this
bipartisan psychosis. And Jonas from PirateWire tweeted it out.
Speaker 7 (18:37):
And.
Speaker 6 (18:37):
Went super viral and they're like, oh, we don't know
how it.
Speaker 4 (18:39):
Went so viral. And it's like this was a perfect cocktail,
especially for DC. I think if they had thrown this
in SF, it might have been like, this is just
how guys talk there and whatever.
Speaker 6 (18:51):
But to bring that bombasticness.
Speaker 4 (18:54):
To DC, which obviously normally has a bit more of
a measured tone, it caught fire.
Speaker 6 (19:00):
It went super viral.
Speaker 1 (19:02):
And tell us about the party itself. I mean, you
quoted someone, I think, with saying, the good guy with
a gun beats the bad guy with a gun, which
calls back a little bit to what we were talking
about before the break. Yeah, yeah.
Speaker 4 (19:13):
So I brought, important context is I brought Taylor Frazier McCollum,
who's a local organizer against data centers. Obviously being in DC,
we're right next to Loudoun County and counties that have
had just a ton of data center build out. And
the organizers told me ahead of time, they were like,
We want this to be fun, but also like educational.
Like we want there to be representatives from different companies.
(19:34):
One of the organizers was like, we're going to have
construction workers that are actually making a lot of money
from data centers and it's really revitalizing their business.
Speaker 6 (19:40):
And yeah, one of them was like, absolutely, we would
welcome you bringing her.
Speaker 4 (19:43):
And so I brought her and it was like incredibly
like Trumpy.
Speaker 6 (19:48):
I knew it was going to be kind of libertarian.
Speaker 3 (19:50):
Leaning.
Speaker 4 (19:51):
It was at the Bitcoin bar in D.C., which like
already from that starting point, you knew it was going
to be quite libertarian.
Speaker 1 (19:57):
Is it literally called the Bitcoin bar?
Speaker 4 (19:58):
It's called PubKey, but it was founded by a big
Bitcoin guy. It's become like an event venue basically for
a lot of libertarian Maggie stuff.
Speaker 6 (20:08):
I find myself.
Speaker 1 (20:09):
Sadder place in the last few weeks, no doubt.
Speaker 6 (20:11):
Yeah, yeah. I find myself at the Bitcoin bar for
work often.
Speaker 4 (20:15):
And yeah, you walk in, there's like a giant screen
playing clips of like Tony Stark and like clearly AI
generated clips of data centers looking cool or whatever. And
then all the speakers were either like kind of conservative
commentators or just AI company people.
Speaker 6 (20:33):
And it was basically just kind of like rallying cries.
Speaker 4 (20:35):
You know, they got up there and they were like,
It's the Chinese that is convincing people that data centers
are polluting our air and polluting and using up too
much water. Like it's it's Chinese propaganda. And you guys
are on the right side of history. And, you know,
this is where the future is.
Speaker 1 (20:49):
Natasha, would you have accepted an invitation to this event?
Speaker 9 (20:53):
Yeah.
Speaker 7 (20:53):
I mean, if it had been in San Francisco, I
definitely would have gone. You know, they host these events
because they want people to attend and to write about it.
Speaker 5 (21:00):
No, I'm just so fascinated.
Speaker 1 (21:01):
It's coming to San Francisco soon, I think.
Speaker 3 (21:03):
Is it?
Speaker 5 (21:04):
Okay, well, let's see if they let me in.
Speaker 3 (21:05):
I'll see you there, Natasha.
Speaker 7 (21:09):
Yeah, I just loved Margot's write-up. You know, you don't
always get good party reporting on tech meme. So I
was quite thrilled.
Speaker 4 (21:19):
They actually had, I didn't put this in the piece
because it was more of a cultural thing, but they
were packets lying around where they clearly had given some
sort of presentation ahead of time fundraising for like pro
data center lobbying just for the VIPs. Media was like
on the VIP list to get in, but you couldn't
go to the VIP room where they were handing out
these packets.
Speaker 1 (21:37):
Isn't that the experience of working in the media?
Speaker 4 (21:40):
Yeah.
Speaker 1 (21:42):
There's always a back room. Reid, what about you? Have
you been following this?
Speaker 8 (21:46):
Yeah.
Speaker 3 (21:46):
I mean, I'm fascinated by this anti-data center sort of movement,
it still is unclear to me whether it's a right
or a left movement. Although like reading Margo's stuff and
other stories, it kind of makes me think that it
is starting to, it is like maybe going to divide
along those lines. Like that's going to be the left
is anti-data center and the right is pro-data center or
(22:08):
pro-development or whatever. Did you guys agree with that? That's
my real question here.
Speaker 7 (22:14):
I feel like this actually might end up being kind
of a split between MAGA, right, and tech, right? Because
it's like literally in your backyard and there is at
least the polls show bipartisan animus towards data centers.
Speaker 5 (22:29):
I mean, I think it is really hard to tell, right?
Speaker 7 (22:32):
Because it's like, If you ask people who don't even
have data centers being built in their community, they are
still anti it. But like it shows high bipartisanship. I
think that's why we are all talking about it. And
I think that's why the industry like sees this as
an existential issue. They need to make sure that it's
not seen as like, that they are the enemy of
the American people, that it's seen as, like, misinformation or,
(22:56):
you know, some kind of NIMBY-type issue. Actually, that's another
thing I was curious about, Margot, was it all framed
as not just CCP propaganda, but, like, also just misinformation?
I saw the report from, I guess, Washington Sun, where
they were mentioning that, like, yeah, folks talked about it as,
it's just these people need to be educated.
Speaker 4 (23:16):
There was definitely some of that, um, And I think
you're totally right when you're talking about it being a
split between the tech right and like MAGA right. Because
to me, like being anti-data center from the people I
talk to, it feels very like the populism that Trump
kind of tapped into, right?
Speaker 1 (23:31):
Yeah, I really seized on what you said as well, Natasha.
I'm sorry to interrupt you, but there's a line in
your piece where you wrote, the activist you brought cares
about her immediate access to clean water, pollution-free air and
affordable electricity, while the sort of pro-data center party goers
speak in decades against forgiving the follies of today if
they pave the way for future abundance. And that kind
of made me think of the whole like globalist versus
(23:53):
Vance kind of thing as well. Like it's really interesting.
Speaker 4 (23:57):
Yeah, that was something I ran into a lot. And
especially I talked to a lot of local organizers for
this piece. And what was interesting was that like genuinely
a lot of the problems they have like will be solved.
The water issue, I think closed systems do largely solve that.
The power grid issues, I think nuclear power You know,
if we're able to build that out fast enough, we'll
(24:17):
probably solve that. And at the same time, you're having
these companies have to build them now. And people are
feeling these like incredibly negative immediate effects. And there was
just this like dissonance where I talked to people and
I'd be like, OK, well, this woman, her electricity bill
is higher and she hears a constant humming and she's
worried about water. and they may face rolling blackouts into
(24:39):
2027 if they keep building data centers.
Speaker 6 (24:41):
And they'll be like, well, all of that will be fixed.
And it's like, okay, but you're building them right now.
Speaker 1 (24:46):
Hello, we're going to put them in space.
Speaker 6 (24:48):
Yeah, literally.
Speaker 3 (24:49):
The humming is, that's been going on since before AI.
I actually think, have you done reporting this? What I've
heard is that the water-cooled, the liquid-cooled data centers being
built now do not have the same humming sound. Is
that right?
Speaker 4 (25:05):
Oh, interesting.
Speaker 6 (25:08):
I don't know.
Speaker 1 (25:09):
Reid, you may have been a better messenger than the
party organizers in terms of communicating some of.
Speaker 3 (25:14):
The benefits of data centers. Well, yeah, the water issue
is solved. Also, the energy issue is kind of a misnomer, too.
Energy prices are going up for lots of reasons, and
I don't think data centers are really one of them.
Speaker 6 (25:25):
No, I agree.
Speaker 4 (25:26):
Figuring out, I also looked into it because I was like,
there's not solid evidence that it's increasing energy bills. There
are counties where the bills have increased by a significant amount.
But it's hard to link that to the data centers.
And then also you're seeing like this happened with the
Elon data center in Memphis was kind of an egregious
(25:46):
example of this where they started bringing in portable natural
gas plants, I think, to not strain the electrical grid.
Speaker 5 (25:51):
Diesel generators. Yeah.
Speaker 1 (25:52):
Yeah.
Speaker 5 (25:53):
But I think it's not just in Memphis. I'm not
sure how. how prevalent it is.
Speaker 7 (25:58):
But like talking to Justin Peterson there, who's like a
local representative, I mean, it is like hugely polluting.
Speaker 5 (26:06):
As he told, he had a really good quote.
Speaker 7 (26:08):
He was like, I'm hearing this from people, you know,
that this is CCP propaganda from people who like never
cared to know me, you know, like didn't come down,
like just want to use me as a talking point.
Like the people that I live with are dealing with
this repercussions right now i.
Speaker 3 (26:22):
Mean there is you have to admit there is some
ccp propaganda it's just.
Speaker 5 (26:26):
Like no i do not have to admit that i'm not.
Speaker 3 (26:30):
Saying it makes that much of a difference or if
it's driving the issue but i mean it is there
i think i think it's like the 2016 election interference
with russia right it was like yeah i mean the
russians were doing stuff but like they were exploiting divisions
that already existed in our.
Speaker 7 (26:45):
Systems I mean, I haven't looked into this since I
wrote about it a few months ago when the CCP
allegations were very easy to beat. So maybe some new
data has come to light. But I also have been
reporting on this for like when you talk to people
like The data center organizers, in fact, like when I
was talking to people late last year in Virginia, and
(27:07):
I was like, oh, are they connecting it to like
Silicon Valley and the billionaires? And they were like, no, because,
you know, they come in with an NDA. They don't
even know who the company is. This is very much
like the county next to me has data centers. They're
not benefiting from it. Why would we say yes? You know,
you just kind of doge your way in. The EPA
restrictions are being rolled back. Like, why wouldn't I fight
(27:29):
against this?
Speaker 1 (27:30):
Reid, do you have any evidence of the Chinese propaganda?
Speaker 3 (27:33):
I mean, I'm not saying it's some grand scheme that's
changing America's opinions. I'm just saying, if you're going to
say confidently the Chinese are not going to exploit this
in any way.
Speaker 1 (27:48):
My hunch would be that any social division in the US,
the Chinese Communist Party... internet activists will be.
Speaker 4 (27:54):
Yeah, like insofar as it's always happening, you.
Speaker 6 (27:57):
Know what I mean?
Speaker 3 (27:57):
Yeah, this is always happening. It always will. Like, is
it actually like changing the debate in some significant way?
Probably not. Because like you said, Natasha, like we were
writing about this issue in Virginia, like over two years ago. And,
you know, we were warning that this was going to
be a problem for the AI companies long before like
(28:18):
anyone was talking about it. These issues predate chat GPT.
These data centers, they do make a loud humming noise
and they look really ugly. People don't like them.
Speaker 1 (28:29):
And they're scary.
Speaker 6 (28:31):
By the way, one of the companies there is solving
that by making them look like barns.
Speaker 1 (28:34):
Yeah, we wrote about that too.
Speaker 3 (28:35):
We wrote about the fact that these companies are hiring
architects to try to make cool looking data centers, right?
Speaker 5 (28:40):
Which was an idea they like batted down.
Speaker 7 (28:42):
Yeah, I think like the tech world maybe recently discovered,
but people have.
Speaker 5 (28:47):
Been reporting on this.
Speaker 7 (28:48):
I mean, I'm not saying that it's not, privy to influence.
But I would say like More Perfect Union has been much,
much more influential. They've been doing these like amazing videos
about local data center protests. I think it was Bernie
who helped connect what I could see happening. It was like, OK,
people are very mad about AI here and they're very
mad about data centers. And Bernie is the one who
(29:08):
like brought this together. And this is partly from More
Perfect Union.
Speaker 1 (29:13):
We've got to go to the break. But Margot, last
word to you. You raised the question in your piece,
is there no way to create a pro-tech movement that's
neither cruel nor uncool?
Speaker 9 (29:22):
Yeah.
Speaker 6 (29:23):
And I don't know the answer to that.
Speaker 4 (29:26):
I was struck by, I mean, again, their excuse now
is like on Twitter, the organizers are like, it was
a party, like everyone's like overreacting. And it's like, well,
when I talked to you guys beforehand, you were like
very measured and we had some great conversations and you
were like, we really want this to be a And
it's actually DC, half the events here are coalition building
to try to like give you the good talking points
(29:47):
for an issue.
Speaker 6 (29:48):
This party did not do that. This party was capitalizing on,
you know.
Speaker 4 (29:51):
MAGA contrarian edgelord culture to go super viral.
Speaker 6 (29:57):
And to me, it wasn't very compelling.
Speaker 1 (30:00):
When we come back, Yahoo's Marissa Meyer is back. Maybe
this time she's betting you're more likely to show a
genetic AI your camera roll than your inbox. Welcome back, Reid.
(30:25):
You have spent some time in your career following the
former Yahoo CEO, Marissa Meyer. And she's actually back in
the news this week with a new venture called Dazzle,
which may answer some of Natasha's security fears about Muse.
But what is Dazzle and how does it work?
Speaker 3 (30:42):
Yeah, Dazzle. I mean, well, you mentioned covering her. I mean,
I've talked to her and covered her previous starts. But
when I moved out here like over 10 years ago,
she was like the big name in Silicon Valley. I mean,
she was at Yahoo and Yahoo was still, if some
of our listeners might not even remember Yahoo, but it
was a very hot technology company. She played a big
(31:04):
part in like the early years of Google. And, you know,
I think what's sort of cool about her is that
like a lot of these people, like they just want
to be like CEO or move on to like the bigger,
bigger title or become like a big VC or something
like that. Like she just wants to like be a
startup founder and like start a company in her garage,
even though, you know, she's this.
Speaker 1 (31:26):
Even though her garage probably has a Gulfstream jet in it.
Speaker 3 (31:29):
Yeah, yeah, yeah.
Speaker 1 (31:30):
She's the 20th employee at Google.
Speaker 3 (31:30):
Totally. Actually, she's known for like her Halloween parties that
she like decorates the entire neighborhood.
Speaker 7 (31:35):
I was going to say, I have been in her
garage to get the Halloween candy. You know, she has
like movie size Halloween candy. That's what she gives away.
Speaker 1 (31:42):
What do you mean movie size?
Speaker 3 (31:44):
Like the one you would buy in a theater, you know,
like you're not getting like the mini Snickers bar.
Speaker 7 (31:48):
And there were people like stocking shelves.
Speaker 5 (31:50):
Anyway, sorry.
Speaker 3 (31:51):
Go ahead. What I respect about her is like she
just wants to like start little companies, like do startups. And,
you know, her last one was this like photo sharing app,
which I tried. I thought it was kind of a
cool idea, but like was never going to be some
huge success.
Speaker 7 (32:06):
Shine. Yeah.
Speaker 3 (32:07):
And like this. This concept for this one is like,
instead of reading all your emails and texts and whatever,
you know, your calendar, it's just going to look at
your photos and glean all the information it needs from
your photos, which I don't know, to me doesn't sound,
I mean, there's probably like, Natasha, you can go into
like the privacy implications here, but like, it does actually
technically doesn't sound like it's going to work, especially for
(32:29):
people like me. I'm not taking a photo of like
every single thing that I witnessed.
Speaker 9 (32:34):
And, you know.
Speaker 1 (32:35):
What should I do to have fun? Go to the
beach once a year. Go cycling.
Speaker 3 (32:42):
I'm finding the tools that actually read my texts and
emails and calendar very, very useful. I've been using since yesterday.
I used my dot from OpenAI that was announced earlier
this weekend. It's been incredibly useful for me already. I
just don't see how... There may be a larger play
here But at first glance, just my honest opinion is
(33:06):
I don't see this startup becoming a big success either.
But who knows? I could be wrong. What I will
say on the topic of these AI assistants like Muse
and others, OpenClaw, they're essentially the same thing, is this
is a wide open field. On consumer, I think no
one has figured this out yet. Someone will. And there
(33:27):
will be some massive hit. I've always thought Google has
the best opportunity to do this. for a lot of reasons,
but they haven't for some reason. And someone's going to
figure it out.
Speaker 1 (33:37):
I had dinner with a friend who's 85 years old
on Monday, who'd seen the front page story about Muse
on the New York Times, which we talked about last
week and was like, very persuaded by it. You know,
so I think something is happening. I know there was
some sense that maybe Muse wasn't all it was cracked
up to be last week, but something is clearly going
on right now.
Speaker 3 (33:54):
Muse is cool. It's not going to be the answer, though. Like,
I can already tell you this, like, Muse is not
going to be the consumer AI app. And actually, I
don't think the infrastructure is there. I think, unfortunately, we
need some more data centers. Like, there's not enough compute
right now for everyone to.
Speaker 4 (34:08):
But don't you think they have, like, I'm just saying,
you know, when I talk to my parents, they might
use Facebook's AI, which is, this is what Zuckerberg's been
banking on, right? Which is that, like, the average person's
not going to go and download ChatGPT. My mom doesn't
trust it. But if there's AI built into an app
that she's using every day, like, why wouldn't that be
the answer?
Speaker 3 (34:25):
I think that makes, that's a logical argument. It makes
total sense. I just, I feel like after covering this
for a while, I have like a good sort of
gut sense about this and like Muse is not it,
but maybe Facebook will come up with, or Meta will
come up with another one or Muse will evolve or
something like that. Like I would leave open that possibility.
Speaker 1 (34:42):
Well, I mean, Marissa Mai has said that part of
the reason for doing this is that the kind of
foundational AI that already exists isn't really working for normal people.
So the insight may be... Go ahead, Natasha.
Speaker 7 (34:53):
No, I actually love the idea of people going through
my camera roll or just like any camera roll technology.
It's like the closest I would come to a diary
in terms of the number of screenshots and photos and...
I just, I remember there was a startup a few
years ago that was about organizing it. And I just
feel like it's one of those things that is not
part of a tech dad presentation that would be at,
(35:16):
you know, Meta or Apple, but actually a lot of
people would love to have a way to- I feel
like as a tech dad, I just have to- I
feel like this is like a Gen Z product.
Speaker 6 (35:25):
I think young people.
Speaker 5 (35:26):
Yeah, but I mean, I'm.
Speaker 7 (35:27):
Not Gen Z, but I mean, yay for people like
looking and messing with my.
Speaker 4 (35:32):
I mean, Snapchat, when I talk to my like young cousins,
Snapchat's like the only app they use.
Speaker 6 (35:36):
They are only communicating via photo.
Speaker 9 (35:38):
Yeah.
Speaker 6 (35:38):
So I love that.
Speaker 5 (35:39):
And it takes away.
Speaker 7 (35:40):
I mean, I don't do anything wrong. So like maybe
if you're worried about like, who are you with in
your photos or anything?
Speaker 5 (35:49):
But I find that great.
Speaker 7 (35:50):
Like if you're not sharing my photos anywhere, you're just extracting,
you know, like. hey, I don't know, like your son
needs new sneakers that are obviously very beat up or
something like that.
Speaker 6 (36:00):
It's insulting your clothes?
Speaker 5 (36:02):
No, no, like, no, like you're running out of things.
Speaker 8 (36:04):
I don't know.
Speaker 5 (36:05):
I'm just saying.
Speaker 6 (36:05):
If my AI app is like, actually, your sneakers are busted.
Speaker 7 (36:11):
I just am saying I have seen zero innovation. Here
is a product, my camera roll that I am interacting with.
I don't know how many times a day. It's just
stored full of information, even if somebody could just organize
it better into folders than me, I mean, yes, please
put some innovation towards this. I cannot believe that there
(36:31):
hasn't been more technology around it.
Speaker 3 (36:34):
I mean, I agree with what you're saying, but what's
the utility there? When I use, for instance, this dot,
I'm saying like, oh, I've got to respond to this
calendar request for an event that other people in a
different time zone are wanting to dial into. And then
I have to deal with like my kids pick up
from their activities. And I literally told my dot yesterday,
(36:57):
like just a voice prompt. And it just did all
those things for me instantly. Like, I don't think just
looking at your photos, like you have that kind of utility.
Speaker 1 (37:06):
I mean, Margot, I want to hear from you because
you're closest to the fabled Gen Z cohort. I am
Gen Z. I'm like the oldest Gen Z.
Speaker 4 (37:14):
But yeah, I was just going to say, like, I
don't think it would be for that, right? Like, I
don't really see how it would like serve as like
a real personal assistant. I think though that at least
for me, like my photos are kind of the most
concentrated data of like what I care about.
Speaker 6 (37:29):
It's like my cats and the best meals I've had.
Speaker 4 (37:33):
And the places I actually want to travel or I
actually go to when I travel.
Speaker 3 (37:36):
But then what's the utility though?
Speaker 6 (37:38):
Well, I could see being like, I'm.
Speaker 4 (37:40):
Going to Croatia, plan me a trip. And it's like, okay,
you take pictures of. You know, when you went through Europe,
you took pictures of all the churches and you seem
to really like a great meal at 2 p.m. You're
eating at 2 p.m.
Speaker 6 (37:49):
There. Like, here's a great restaurant.
Speaker 3 (37:52):
It is a reservoir of data. I would guess that
it would get a lot of things totally wrong by
misinterpreting your photos because you're taking photos of random stuff
all the time. Mine would be like, I suggest you
go to vacation at Home Depot. You seem to spend
a lot of time there.
Speaker 7 (38:04):
Well, honestly.
Speaker 5 (38:11):
But I would be curious to see what they would
find out about me from that. It's not known.
Speaker 7 (38:17):
I just feel like it's the anti tech dad. It's
the anti like make my reservation for me.
Speaker 5 (38:23):
You know what?
Speaker 7 (38:23):
I know that I've taken screenshots of Instagram posts in Croatia.
If they could go through and be like, here's actually
all your aspirational posts.
Speaker 5 (38:32):
Instagram screenshots.
Speaker 3 (38:33):
Why is tech dad your denigration phrase?
Speaker 6 (38:37):
This is such an intergenerational fight.
Speaker 3 (38:40):
My wife loves this stuff. We're super busy parents. You
want stuff that's going to help you out. Why is
that a tech dad thing?
Speaker 5 (38:47):
Okay, you know why? Because I write about this stuff.
Speaker 7 (38:49):
I've sat through many demonstrations at google meta apple and
i do not find that they reflect like what i
as even as a parent like when they talk about
things like i'm not posting my data on facebook you
know or i don't either that's what i'm signing.
Speaker 1 (39:06):
For today we're gonna host hopefully another intergenerational fight on
text how soon maybe even live from croatia.
Speaker 3 (39:13):
I feel like this is just starting to get.
Speaker 1 (39:15):
Good thank you all so much For Tech Stuff, I'm
Oz Veloshin. This episode was produced by Eliza Dennis. It
was executive produced by me and Julia Nutter for Kaleidoscope.
Speaker 7 (39:29):
And.
Speaker 1 (39:29):
Katrina Norvell for iHeart Podcasts. Kyle Murdoch mixed this episode
and wrote our theme song. A special thank you to
Margot McColl, Natasha Tiku, and Reid Albigotti. Please check out
all the work they put out into the world. We're
lucky to call them friends of the pod.
Natasha, Reid, Margot, we've talked about social media bans and technology.
And this is, of course, a podcast. So I was
absolutely charmed this week by a revelation that middle schoolers
are using the comments of random NPR podcasts on Spotify
to chat. Here's a clip of Ira Glass talking about it.
Speaker 2 (00:31):
Kids would name their playlist, chat here or something like that.
And then there'd be some podcast. And everybody would go
to that podcast and talk in the comments. She says
it was maybe 20 kids in the core group. Lots
of theater kids, she says. Mostly girls. Most of them
with strict parents who didn't let them on regular social media.
And can I ask, why did you guys pick NPR
shows to be the ones where you went to.
Speaker 3 (00:53):
In the comments?
Speaker 4 (00:56):
I think we just looked for podcasts that didn't have
many comments.
Speaker 3 (01:02):
I see.
Speaker 2 (01:02):
So you picked NPR because it didn't seem very popular.
Speaker 5 (01:07):
I've been feeling bad for Ira Glass all week because
of this.
Speaker 1 (01:13):
Welcome to Tech Stuff. I'm Oz Veloshian, and this is
The Week in Tech, where I'm joined by the world's
most plugged-in reporters to break down what's really happening in
tech right now. Today, we're joined by Natasha Tiku, tech
reporter at The Washington Post, Margot McColl, contributing writer at
Vanity Fair, and Reid Arbogati, tech editor at Semaphore. Welcome all.
Speaker 3 (01:33):
Great to be back.
Speaker 5 (01:34):
Hi, thanks for having me.
Speaker 1 (01:36):
Margot is a second-time Tech Stuff contributor, so special welcome
back to you.
Speaker 6 (01:39):
Thank you.
Speaker 1 (01:40):
So last week we were all fielding calls, or at
least a read was, from our mothers about the likelihood
of the AI apocalypse. I think the take of the
panelists was we're safe through the end of Q4 at least.
Dari Amode and others are persisting in their campaign to
get the government to intervene to pace the frontier, but
falling increasingly on deaf ears. The president has been averse
(02:02):
to regulating AI, or superintelligence as he prefers to call it,
And last Tuesday, he gathered the heads of the frontier
labs and tech firms in Washington for a kind of
working lunch that looked uncannily like the Last Supper. Natasha,
what happened?
Speaker 5 (02:19):
Well, everyone had lunch in the East Room.
Speaker 7 (02:22):
We got a readout beforehand of the layout of where
everyone was sitting.
Speaker 5 (02:26):
They discussed super intelligence.
Speaker 7 (02:28):
And then we had, I think, like one of the
world's most interesting films on the North driveway afterward, where
you saw various male egos in repose.
Speaker 5 (02:39):
And in flux.
Speaker 7 (02:41):
It was like Dario Amadei, Jensen Huang, Mark Zuckerberg, Elon Musk,
David Sachs. And they signed an accord, a White House super,
you can't see my scared clothes, but a White House
super intelligence accord where they agreed to some beautiful self-regulation.
including third-party audits. Yeah.
Speaker 1 (03:03):
Tremendous self-policing, in other words.
Speaker 7 (03:05):
Yep, exactly.
Speaker 3 (03:06):
It was more like a friend DA, right?
Speaker 6 (03:09):
What did he say? It was like morally enforced? He's like, no, guys,
they're going to do it.
Speaker 3 (03:13):
It's a gentleman's agreement.
Speaker 1 (03:15):
It was literally a gentleman's or a men's agreement. They're
boy brothers now. Natasha, I want to ask you a
bit more about the content. But first, I mean, take
us through the dynamics of the lunch. Who is that where?
And then the jostling on the lawn afterwards.
Speaker 5 (03:31):
I guess I was most fixated on on the lawn afterwards.
Speaker 7 (03:35):
I mean, there was just this like one minute segment
I was semi obsessed with because, you know, all the
questions are going to Dario, who's.
Speaker 5 (03:43):
Like ready to hold court, you know, like as usual.
Speaker 7 (03:47):
Then he realizes Trump is on his other side and
he's realizing, oh, and Trump is not liking all the
questions are going to him. So he He literally does
the like Homer Simpson into a Bush meme like behind
some assorted Republicans. And Mark Zuckerberg and David Sachs are
like smirking and laughing to each other in the background.
(04:07):
Like Mark Zuckerberg looked so happy to be part of
the cool kids club. I've just never seen him so
like in his element and pleased.
Speaker 5 (04:17):
And then he did the synopsis of the accord.
Speaker 7 (04:20):
And I mean, there were some great bits later as well,
like Trump saying that nobody's ever heard of Sundar Pichai.
Speaker 5 (04:26):
And he was like, his name is Sundar. That's all
you need to know. He's a monster.
Speaker 7 (04:32):
But it was really that Dario, Zuck, Trump, Sachs nexus
that was, I don't know, I thought it was very
rich text.
Speaker 3 (04:40):
Well, I saw some reporting on what was discussed at lunch. And,
you know, Trump apparently likes Dario now. He's cool with Dario.
Speaker 1 (04:48):
We had dinner, right?
Speaker 3 (04:50):
We had dinner. Dario's great. Although Trump had to, like,
he started it off with, like, I made him wait
for me, basically. Like, you know, we were supposed to
have dinner at 7 and I arrived at 10. And, like,
But then he's like, of course, in typical Trump fashion,
like everybody he talks to, he likes it. He's like,
I like the guy. He's great. Like, not at all
like what he's portrayed like in the media, you know. Meanwhile, like,
(05:13):
it's literally you portraying him that way. But nobody has
reported on what they were stickering about during the famous video, Natasha,
have they? Like, what were they actually saying? That's what
I want to know.
Speaker 7 (05:25):
Oh, I'd have to check the TikTok lip readers. Like
they must be on it. I just assumed it was
about Dario trying to say, trying to like, you know,
walk that tightrope.
Speaker 1 (05:39):
That reminds me of being like a 16-year-old at a wedding.
It's just awkward. Yeah, you mentioned Homer Simpson, but I
kind of, the stumbling into the bush awkwardly is very resonant.
Speaker 3 (05:47):
I mean, I felt for him.
Speaker 5 (05:48):
It was relatable. Like he was like, uh-oh.
Speaker 1 (05:52):
Margot, did anything about the summit catch your eye?
Speaker 6 (05:55):
I don't know.
Speaker 4 (05:55):
I mean, I'm still fixated on the anthropic S1 that
leaked where it's like going through all the risk factors
that are like normal. And then it's like, and it
might blackmail governments and like destroy the world. Like, I
love that they get to have like a boys lunch
and like. Are they going to ruin the world?
Speaker 8 (06:12):
Maybe.
Speaker 6 (06:13):
But they also get to have friends for the first time,
and that made me really happy for them.
Speaker 3 (06:17):
When do we get to say they're wrong?
Speaker 1 (06:18):
Well, maybe the evaluators will answer that question. Natasha, what
are evaluators, and what have you been reporting about them?
Speaker 7 (06:25):
Yeah, so I've been reporting on this third-party evaluator-auditor ecosystem
that is all of a sudden, you know, like the
agreed-upon first line of defense system. for all the harms
we talked about last week, you know, from existential risk
to like teen mental health. So this is really interesting
(06:48):
because this ecosystem, I think, was already growing. A bit nascent,
but there was a lot of money and interest going
into these evaluation startups. Like now they're kind of national names, Meter, Apollo,
Redwood Research. But, you know, that Particular part of the
ecosystem was already getting good funding and it's like considered
(07:11):
very prestigious. I think, you know, if you're going to
leave a lab, a frontier lab, then like it's a
great place to go. But then there were also companies
like Val's AI, which has just raised from Andreessen Horowitz,
and they previously raised from 8VC. And I think they
have been thinking a little bit more about like public sector,
you know, working for the government, making sure they have
(07:33):
a bipartisan kind of appearance, working better with DC. And
then you have the like Accentures and the Ernst & Young's,
which I see are now advertising for jobs on 80,000 Hours,
which is an EA jobs board. So yeah, everyone is
just like, I think seeing that legislation likely won't get passed,
(07:54):
but this is something that the companies agree on and
that there's already funding for it. And there is a
deep need even just to like vet the basic claims
of the companies because otherwise it's like, yeah.
Speaker 5 (08:07):
It's a little.
Speaker 1 (08:07):
Venture backed compliance.
Speaker 3 (08:09):
It's very similar to the big security boom that happened
like a decade ago when, I don't know if you
remember this Natasha, there were just like 8 million security
startups that did like telemetry. They essentially like all did
the same thing. And then every company in the world
would like hire them, of course, just usually as a
pilot and not giving them very much money, but like,
(08:30):
then they'd get all this data, which turned out to
not really be that useful in cybersecurity. And a lot
of them just like disappeared. I feel like it's a
similar thing, except it took, it took like decades for
that to materialize after the internet and like in AI fashion,
everything happens so quickly that this is like popped up
overnight and it will probably lose its mystique very quickly
(08:51):
as well as my, my prediction.
Speaker 1 (08:53):
Natasha, why is Mita the kind of household name? And
then I'm going to ask you to imagine that we
are three venture capitalists. Like what, what is the investment
pitch to invest in evaluators?
Speaker 5 (09:04):
Okay, I'll answer the easier one first.
Speaker 7 (09:07):
Meter is the company that, the nonprofit that was like
approved by OpenAI to do the investigation into the Hugging
Face incident. They worked with Redwood Research, but it's really
been the folks at Meter who've been the most high profile, like,
you know, doing the three-hour.
Speaker 5 (09:22):
Dwarcash podcast, writing their own substacks.
Speaker 7 (09:25):
In terms of the pitch, I think it is, okay,
there's really no way for anyone to know what they're
getting from these AI models. Like one of the folks
I talked to who's head of public sector for Val's AI,
I mean, he used to work at the Pentagon in
generative AI. And he was saying like, these are agencies
that have testing capabilities, but there's just, was nothing in
(09:48):
place for generative AI models. And because of everything happening faster,
people are buying these products without having really tested it
in Warfield to the point where this prioritizes company is
doing like a test of does Claude or other chatbot
models actually follow the rules of war. Like this is
not how things would normally be unfolding. There would be
(10:11):
like a very rigorous Pentagon testing process, but.
Speaker 1 (10:16):
Is it VALS, like V-A-L, like VALS a guy?
Speaker 6 (10:19):
VALS like evaluation.
Speaker 5 (10:21):
Like evaluation.
Speaker 6 (10:22):
Oh, I see.
Speaker 1 (10:23):
Okay, sorry. Was the substance of the joint commitment on
frontier responsibilities basically, we'll use evaluators? Is that a fair summary?
Speaker 7 (10:32):
No, I think they had four steps. It was basically
like internal controls and some kind of external evaluation that
your controls monitoring and like safety checks. methods are working
and there was something about internal.
Speaker 1 (10:48):
Risk is it conceivable that the lead investor of the
evaluated company and the frontier company could be the same
person called mark andreessen yeah aren't all.
Speaker 4 (10:57):
The employees these places also from the frontier labs like
isn't it a very like co-mingled uh employees.
Speaker 7 (11:04):
Yeah like there's the mark andreessen valves that's a very
small minority i would say most of it is uh
Nonprofits funded by open philanthropy, staffed by former frontier labs
like Meter, for example, was initially started by Paul Cristiano,
who is like an early and very influential open AI employee.
(11:25):
And it's his wife who did the investigation. And she
used to work for open philanthropy for a decade. You know,
so there's a lot of overlap. And Paul recently joined
the board of the OpenAI Foundation, something they had to
add to the report as a footnote. So that's really
been seized upon. Like the New York Post is having
a field day.
Speaker 5 (11:46):
I mean, they are eating well.
Speaker 1 (11:49):
I know we've got to come to the Post because
they've been really leading the story on tech regulation this week.
They had a huge scoop, an exclusive on the FTC
and AI regulation.
Speaker 5 (11:58):
Yeah, it was an exclusive.
Speaker 1 (12:00):
So what's the story and why did it end up
in the Post?
Speaker 7 (12:03):
Well, I think the Post has really been covering stories from,
I would say, the tech right. And this one was
from the Trump admin FTC, which said it is targeting OpenAI, Anthropic,
potentially some other companies, and Meter, according to the New
York Post. And, you know, the FTC confirmed this. But, yeah,
(12:25):
the timing of it was quite interesting because here they're
having their little accord, boys' lunch event.
Speaker 5 (12:32):
And then this news comes out.
Speaker 1 (12:34):
So this was already, was in the works. But what
exactly are the FTC doing? And are they opening a probe?
Are they using existing regulations? Are they writing new regulations?
Speaker 5 (12:44):
No, they're not.
Speaker 7 (12:45):
I think the idea was to demonstrate this talking point
that we can use existing regulations to make sure that
people are still kind of towing the line on that.
That's something we heard a lot from Jensen Huang. That's
something we heard from David Sachs.
Speaker 5 (13:01):
And others.
Speaker 7 (13:02):
And, you know, I think this is supposed to show like, hey,
the FTC could have done this with or without the accord.
It can do it with or without any new laws passing.
And so we don't know yet whether it's is it
an antitrust probe? Is it about some like collusion around
security or just details around concern for the public?
Speaker 5 (13:21):
We don't know yet.
Speaker 9 (13:22):
Yeah.
Speaker 3 (13:22):
But but there is a concern, I think, that these
companies all get together and they say, OK, let's all
agree to not push the frontier and Because we've got
a great business right here. This is a high cost
for us. So let's, under the guise of safety or
whatever it is, let's not spend a bunch of money
training some massive new model. We can make a bunch
(13:43):
of money on the existing ones. That's the kind of
agreement that I think people are worried about. And it
gets very complicated because you can make a safety argument
for doing that.
Speaker 1 (13:54):
Didn't the chairman of the FTC level that exact accusation
that the frontier labs are pleading for regulation now? to
insulate themselves from possible competition.
Speaker 7 (14:04):
I mean, Dario said that explicitly. He asked for a waiver,
an antitrust waiver, in his, like, Pacing the Frontier article.
Speaker 3 (14:12):
So many people have leveled that accusation, including in the
White House, like, on the record, off the record. I mean,
David Sachs is constantly saying it, you know, in his
X feed.
Speaker 1 (14:21):
You know, the chairman did say it in the article.
I want to close out this first segment with a
Tech Stuff exclusive, though. We have a forthcoming show at
Kaleidoscope called The Builders, hosted by Walter Isaacson, where Walter
is in conversations with some of the people we talk
about often on the show. One of whom was Peter Thiel,
(14:42):
whose comments we published earlier this week exclusively in the
Tech Stuff feed. I want you all to.
Speaker 3 (14:46):
React to it.
Speaker 8 (14:48):
There is a certain type of high-tech totalitarianism that's being
tried out in China. It's not as violent as Stalin
or Mao or or Hitler, you don't need to kill
as many people, but it can go to the inner
recesses of your thoughts and it knows more about you
than you know about yourself. And there's something about that
(15:10):
that we really don't want that to be the model
that wins.
Speaker 9 (15:13):
And your old friend, David Sachs, and of course, President
Trump are now just very recently saying, because of the
competition with China, stop talking about pausing AI, even if
you think it could be an existential threat. Do you
agree with that?
Speaker 6 (15:31):
Yes.
Speaker 8 (15:32):
I broadly agree with that. I've been friends with David
Sachs for many decades. The thing that makes me less
comfortable making those kinds of arguments publicly is that on
some level, it's hard when the big companies themselves are
telling you to stop and to slow down.
Speaker 9 (15:52):
So that's caused you to have some pause because you
were not in the pause button camp.
Speaker 8 (15:57):
Well, I'm still not in the pause button camp. And
I'm saying this on your show that I'm not in
the pause button. We need to hit the accelerator as
hard as possible.
Speaker 6 (16:06):
Guys, I have so many thoughts. I think it's like.
Speaker 4 (16:11):
Impossible to overstate how much men like Peter Thiel and
people in that circle believe China is like almost intrinsically evil.
Like when I talk to these men, in public and
in private they truly believe that like china like might
be the antichrist like i talked to one guy who's
building a fully autonomous fighter jet and he's like china's
(16:31):
the antichrist i'm making this for the end times.
Speaker 1 (16:34):
And you believe based on your conversations that this is
a deeply held belief rather than a convenient.
Speaker 4 (16:40):
Uh narrative i think that these men find a lot
of of god and meaning where money.
Speaker 3 (16:47):
Is do you know what i mean well i think
they're taught They think the Communist Party is evil, right?
Not like necessarily all Chinese people, right?
Speaker 4 (16:55):
Not necessarily all Chinese people, but I think with these, like,
you know, I go to these parties, these guys, and
they're quite racist, right?
Speaker 6 (17:01):
And so it's a little hard to disentangle it.
Speaker 1 (17:04):
This is a perfect segue. We're going to come to
your story after the break. When we come back, the
cool kids in Washington have hatched a new plan to
win hearts and minds over to the pro data center side.
One viral particle link at a time. Welcome back, and
(17:33):
welcome back, Margot, contributing writer of Antifair. Very happy to
have you back with us on Tech Stuff. We like
talking about what's going on in Washington and tech lobbying
on this show, but we don't often cover the party
scene in DC. What exactly is Data After Dark? And
is it as ominous as it sounds?
Speaker 6 (17:51):
Yeah, quite the weekend.
Speaker 4 (17:52):
There were two pro-AI parties thrown by very like tech,
what are tech bro, tech people. The big one was
Data After Dark. It was thrown by a Pirate Wires writer,
which is like a tech kind of conservative outlet, and
some like libertarian think tanky guys.
Speaker 6 (18:08):
And they threw this big party.
Speaker 4 (18:11):
The Partiful was talking about they need to fight the
bipartisan psychosis against data centers.
Speaker 6 (18:17):
And it was as fascinating as you might think.
Speaker 1 (18:20):
Well, tell us about the invitation, which had kind of
its own life before the party even started, right?
Speaker 4 (18:23):
Yeah, so it was recently just going to be a
house party for these, like, three guys that are friends.
And they had a Partiful that, again, talked about this
bipartisan psychosis. And Jonas from PirateWire tweeted it out.
Speaker 7 (18:37):
And.
Speaker 6 (18:37):
Went super viral and they're like, oh, we don't know
how it.
Speaker 4 (18:39):
Went so viral. And it's like this was a perfect cocktail,
especially for DC. I think if they had thrown this
in SF, it might have been like, this is just
how guys talk there and whatever.
Speaker 6 (18:51):
But to bring that bombasticness.
Speaker 4 (18:54):
To DC, which obviously normally has a bit more of
a measured tone, it caught fire.
Speaker 6 (19:00):
It went super viral.
Speaker 1 (19:02):
And tell us about the party itself. I mean, you
quoted someone, I think, with saying, the good guy with
a gun beats the bad guy with a gun, which
calls back a little bit to what we were talking
about before the break. Yeah, yeah.
Speaker 4 (19:13):
So I brought, important context is I brought Taylor Frazier McCollum,
who's a local organizer against data centers. Obviously being in DC,
we're right next to Loudoun County and counties that have
had just a ton of data center build out. And
the organizers told me ahead of time, they were like,
We want this to be fun, but also like educational.
Like we want there to be representatives from different companies.
(19:34):
One of the organizers was like, we're going to have
construction workers that are actually making a lot of money
from data centers and it's really revitalizing their business.
Speaker 6 (19:40):
And yeah, one of them was like, absolutely, we would
welcome you bringing her.
Speaker 4 (19:43):
And so I brought her and it was like incredibly
like Trumpy.
Speaker 6 (19:48):
I knew it was going to be kind of libertarian.
Speaker 3 (19:50):
Leaning.
Speaker 4 (19:51):
It was at the Bitcoin bar in D.C., which like
already from that starting point, you knew it was going
to be quite libertarian.
Speaker 1 (19:57):
Is it literally called the Bitcoin bar?
Speaker 4 (19:58):
It's called PubKey, but it was founded by a big
Bitcoin guy. It's become like an event venue basically for
a lot of libertarian Maggie stuff.
Speaker 6 (20:08):
I find myself.
Speaker 1 (20:09):
Sadder place in the last few weeks, no doubt.
Speaker 6 (20:11):
Yeah, yeah. I find myself at the Bitcoin bar for
work often.
Speaker 4 (20:15):
And yeah, you walk in, there's like a giant screen
playing clips of like Tony Stark and like clearly AI
generated clips of data centers looking cool or whatever. And
then all the speakers were either like kind of conservative
commentators or just AI company people.
Speaker 6 (20:33):
And it was basically just kind of like rallying cries.
Speaker 4 (20:35):
You know, they got up there and they were like,
It's the Chinese that is convincing people that data centers
are polluting our air and polluting and using up too
much water. Like it's it's Chinese propaganda. And you guys
are on the right side of history. And, you know,
this is where the future is.
Speaker 1 (20:49):
Natasha, would you have accepted an invitation to this event?
Speaker 9 (20:53):
Yeah.
Speaker 7 (20:53):
I mean, if it had been in San Francisco, I
definitely would have gone. You know, they host these events
because they want people to attend and to write about it.
Speaker 5 (21:00):
No, I'm just so fascinated.
Speaker 1 (21:01):
It's coming to San Francisco soon, I think.
Speaker 3 (21:03):
Is it?
Speaker 5 (21:04):
Okay, well, let's see if they let me in.
Speaker 3 (21:05):
I'll see you there, Natasha.
Speaker 7 (21:09):
Yeah, I just loved Margot's write-up. You know, you don't
always get good party reporting on tech meme. So I
was quite thrilled.
Speaker 4 (21:19):
They actually had, I didn't put this in the piece
because it was more of a cultural thing, but they
were packets lying around where they clearly had given some
sort of presentation ahead of time fundraising for like pro
data center lobbying just for the VIPs. Media was like
on the VIP list to get in, but you couldn't
go to the VIP room where they were handing out
these packets.
Speaker 1 (21:37):
Isn't that the experience of working in the media?
Speaker 4 (21:40):
Yeah.
Speaker 1 (21:42):
There's always a back room. Reid, what about you? Have
you been following this?
Speaker 8 (21:46):
Yeah.
Speaker 3 (21:46):
I mean, I'm fascinated by this anti-data center sort of movement,
it still is unclear to me whether it's a right
or a left movement. Although like reading Margo's stuff and
other stories, it kind of makes me think that it
is starting to, it is like maybe going to divide
along those lines. Like that's going to be the left
is anti-data center and the right is pro-data center or
(22:08):
pro-development or whatever. Did you guys agree with that? That's
my real question here.
Speaker 7 (22:14):
I feel like this actually might end up being kind
of a split between MAGA, right, and tech, right? Because
it's like literally in your backyard and there is at
least the polls show bipartisan animus towards data centers.
Speaker 5 (22:29):
I mean, I think it is really hard to tell, right?
Speaker 7 (22:32):
Because it's like, If you ask people who don't even
have data centers being built in their community, they are
still anti it. But like it shows high bipartisanship. I
think that's why we are all talking about it. And
I think that's why the industry like sees this as
an existential issue. They need to make sure that it's
not seen as like, that they are the enemy of
the American people, that it's seen as, like, misinformation or,
(22:56):
you know, some kind of NIMBY-type issue. Actually, that's another
thing I was curious about, Margot, was it all framed
as not just CCP propaganda, but, like, also just misinformation?
I saw the report from, I guess, Washington Sun, where
they were mentioning that, like, yeah, folks talked about it as,
it's just these people need to be educated.
Speaker 4 (23:16):
There was definitely some of that, um, And I think
you're totally right when you're talking about it being a
split between the tech right and like MAGA right. Because
to me, like being anti-data center from the people I
talk to, it feels very like the populism that Trump
kind of tapped into, right?
Speaker 1 (23:31):
Yeah, I really seized on what you said as well, Natasha.
I'm sorry to interrupt you, but there's a line in
your piece where you wrote, the activist you brought cares
about her immediate access to clean water, pollution-free air and
affordable electricity, while the sort of pro-data center party goers
speak in decades against forgiving the follies of today if
they pave the way for future abundance. And that kind
of made me think of the whole like globalist versus
(23:53):
Vance kind of thing as well. Like it's really interesting.
Speaker 4 (23:57):
Yeah, that was something I ran into a lot. And
especially I talked to a lot of local organizers for
this piece. And what was interesting was that like genuinely
a lot of the problems they have like will be solved.
The water issue, I think closed systems do largely solve that.
The power grid issues, I think nuclear power You know,
if we're able to build that out fast enough, we'll
(24:17):
probably solve that. And at the same time, you're having
these companies have to build them now. And people are
feeling these like incredibly negative immediate effects. And there was
just this like dissonance where I talked to people and
I'd be like, OK, well, this woman, her electricity bill
is higher and she hears a constant humming and she's
worried about water. and they may face rolling blackouts into
(24:39):
2027 if they keep building data centers.
Speaker 6 (24:41):
And they'll be like, well, all of that will be fixed.
And it's like, okay, but you're building them right now.
Speaker 1 (24:46):
Hello, we're going to put them in space.
Speaker 6 (24:48):
Yeah, literally.
Speaker 3 (24:49):
The humming is, that's been going on since before AI.
I actually think, have you done reporting this? What I've
heard is that the water-cooled, the liquid-cooled data centers being
built now do not have the same humming sound. Is
that right?
Speaker 4 (25:05):
Oh, interesting.
Speaker 6 (25:08):
I don't know.
Speaker 1 (25:09):
Reid, you may have been a better messenger than the
party organizers in terms of communicating some of.
Speaker 3 (25:14):
The benefits of data centers. Well, yeah, the water issue
is solved. Also, the energy issue is kind of a misnomer, too.
Energy prices are going up for lots of reasons, and
I don't think data centers are really one of them.
Speaker 6 (25:25):
No, I agree.
Speaker 4 (25:26):
Figuring out, I also looked into it because I was like,
there's not solid evidence that it's increasing energy bills. There
are counties where the bills have increased by a significant amount.
But it's hard to link that to the data centers.
And then also you're seeing like this happened with the
Elon data center in Memphis was kind of an egregious
(25:46):
example of this where they started bringing in portable natural
gas plants, I think, to not strain the electrical grid.
Speaker 5 (25:51):
Diesel generators. Yeah.
Speaker 1 (25:52):
Yeah.
Speaker 5 (25:53):
But I think it's not just in Memphis. I'm not
sure how. how prevalent it is.
Speaker 7 (25:58):
But like talking to Justin Peterson there, who's like a
local representative, I mean, it is like hugely polluting.
Speaker 5 (26:06):
As he told, he had a really good quote.
Speaker 7 (26:08):
He was like, I'm hearing this from people, you know,
that this is CCP propaganda from people who like never
cared to know me, you know, like didn't come down,
like just want to use me as a talking point.
Like the people that I live with are dealing with
this repercussions right now i.
Speaker 3 (26:22):
Mean there is you have to admit there is some
ccp propaganda it's just.
Speaker 5 (26:26):
Like no i do not have to admit that i'm not.
Speaker 3 (26:30):
Saying it makes that much of a difference or if
it's driving the issue but i mean it is there
i think i think it's like the 2016 election interference
with russia right it was like yeah i mean the
russians were doing stuff but like they were exploiting divisions
that already existed in our.
Speaker 7 (26:45):
Systems I mean, I haven't looked into this since I
wrote about it a few months ago when the CCP
allegations were very easy to beat. So maybe some new
data has come to light. But I also have been
reporting on this for like when you talk to people
like The data center organizers, in fact, like when I
was talking to people late last year in Virginia, and
(27:07):
I was like, oh, are they connecting it to like
Silicon Valley and the billionaires? And they were like, no, because,
you know, they come in with an NDA. They don't
even know who the company is. This is very much
like the county next to me has data centers. They're
not benefiting from it. Why would we say yes? You know,
you just kind of doge your way in. The EPA
restrictions are being rolled back. Like, why wouldn't I fight
(27:29):
against this?
Speaker 1 (27:30):
Reid, do you have any evidence of the Chinese propaganda?
Speaker 3 (27:33):
I mean, I'm not saying it's some grand scheme that's
changing America's opinions. I'm just saying, if you're going to
say confidently the Chinese are not going to exploit this
in any way.
Speaker 1 (27:48):
My hunch would be that any social division in the US,
the Chinese Communist Party... internet activists will be.
Speaker 4 (27:54):
Yeah, like insofar as it's always happening, you.
Speaker 6 (27:57):
Know what I mean?
Speaker 3 (27:57):
Yeah, this is always happening. It always will. Like, is
it actually like changing the debate in some significant way?
Probably not. Because like you said, Natasha, like we were
writing about this issue in Virginia, like over two years ago. And,
you know, we were warning that this was going to
be a problem for the AI companies long before like
(28:18):
anyone was talking about it. These issues predate chat GPT.
These data centers, they do make a loud humming noise
and they look really ugly. People don't like them.
Speaker 1 (28:29):
And they're scary.
Speaker 6 (28:31):
By the way, one of the companies there is solving
that by making them look like barns.
Speaker 1 (28:34):
Yeah, we wrote about that too.
Speaker 3 (28:35):
We wrote about the fact that these companies are hiring
architects to try to make cool looking data centers, right?
Speaker 5 (28:40):
Which was an idea they like batted down.
Speaker 7 (28:42):
Yeah, I think like the tech world maybe recently discovered,
but people have.
Speaker 5 (28:47):
Been reporting on this.
Speaker 7 (28:48):
I mean, I'm not saying that it's not, privy to influence.
But I would say like More Perfect Union has been much,
much more influential. They've been doing these like amazing videos
about local data center protests. I think it was Bernie
who helped connect what I could see happening. It was like, OK,
people are very mad about AI here and they're very
mad about data centers. And Bernie is the one who
(29:08):
like brought this together. And this is partly from More
Perfect Union.
Speaker 1 (29:13):
We've got to go to the break. But Margot, last
word to you. You raised the question in your piece,
is there no way to create a pro-tech movement that's
neither cruel nor uncool?
Speaker 9 (29:22):
Yeah.
Speaker 6 (29:23):
And I don't know the answer to that.
Speaker 4 (29:26):
I was struck by, I mean, again, their excuse now
is like on Twitter, the organizers are like, it was
a party, like everyone's like overreacting. And it's like, well,
when I talked to you guys beforehand, you were like
very measured and we had some great conversations and you
were like, we really want this to be a And
it's actually DC, half the events here are coalition building
to try to like give you the good talking points
(29:47):
for an issue.
Speaker 6 (29:48):
This party did not do that. This party was capitalizing on,
you know.
Speaker 4 (29:51):
MAGA contrarian edgelord culture to go super viral.
Speaker 6 (29:57):
And to me, it wasn't very compelling.
Speaker 1 (30:00):
When we come back, Yahoo's Marissa Meyer is back. Maybe
this time she's betting you're more likely to show a
genetic AI your camera roll than your inbox. Welcome back, Reid.
(30:25):
You have spent some time in your career following the
former Yahoo CEO, Marissa Meyer. And she's actually back in
the news this week with a new venture called Dazzle,
which may answer some of Natasha's security fears about Muse.
But what is Dazzle and how does it work?
Speaker 3 (30:42):
Yeah, Dazzle. I mean, well, you mentioned covering her. I mean,
I've talked to her and covered her previous starts. But
when I moved out here like over 10 years ago,
she was like the big name in Silicon Valley. I mean,
she was at Yahoo and Yahoo was still, if some
of our listeners might not even remember Yahoo, but it
was a very hot technology company. She played a big
(31:04):
part in like the early years of Google. And, you know,
I think what's sort of cool about her is that
like a lot of these people, like they just want
to be like CEO or move on to like the bigger,
bigger title or become like a big VC or something
like that. Like she just wants to like be a
startup founder and like start a company in her garage,
even though, you know, she's this.
Speaker 1 (31:26):
Even though her garage probably has a Gulfstream jet in it.
Speaker 3 (31:29):
Yeah, yeah, yeah.
Speaker 1 (31:30):
She's the 20th employee at Google.
Speaker 3 (31:30):
Totally. Actually, she's known for like her Halloween parties that
she like decorates the entire neighborhood.
Speaker 7 (31:35):
I was going to say, I have been in her
garage to get the Halloween candy. You know, she has
like movie size Halloween candy. That's what she gives away.
Speaker 1 (31:42):
What do you mean movie size?
Speaker 3 (31:44):
Like the one you would buy in a theater, you know,
like you're not getting like the mini Snickers bar.
Speaker 7 (31:48):
And there were people like stocking shelves.
Speaker 5 (31:50):
Anyway, sorry.
Speaker 3 (31:51):
Go ahead. What I respect about her is like she
just wants to like start little companies, like do startups. And,
you know, her last one was this like photo sharing app,
which I tried. I thought it was kind of a
cool idea, but like was never going to be some
huge success.
Speaker 7 (32:06):
Shine. Yeah.
Speaker 3 (32:07):
And like this. This concept for this one is like,
instead of reading all your emails and texts and whatever,
you know, your calendar, it's just going to look at
your photos and glean all the information it needs from
your photos, which I don't know, to me doesn't sound,
I mean, there's probably like, Natasha, you can go into
like the privacy implications here, but like, it does actually
technically doesn't sound like it's going to work, especially for
(32:29):
people like me. I'm not taking a photo of like
every single thing that I witnessed.
Speaker 9 (32:34):
And, you know.
Speaker 1 (32:35):
What should I do to have fun? Go to the
beach once a year. Go cycling.
Speaker 3 (32:42):
I'm finding the tools that actually read my texts and
emails and calendar very, very useful. I've been using since yesterday.
I used my dot from OpenAI that was announced earlier
this weekend. It's been incredibly useful for me already. I
just don't see how... There may be a larger play
here But at first glance, just my honest opinion is
(33:06):
I don't see this startup becoming a big success either.
But who knows? I could be wrong. What I will
say on the topic of these AI assistants like Muse
and others, OpenClaw, they're essentially the same thing, is this
is a wide open field. On consumer, I think no
one has figured this out yet. Someone will. And there
(33:27):
will be some massive hit. I've always thought Google has
the best opportunity to do this. for a lot of reasons,
but they haven't for some reason. And someone's going to
figure it out.
Speaker 1 (33:37):
I had dinner with a friend who's 85 years old
on Monday, who'd seen the front page story about Muse
on the New York Times, which we talked about last
week and was like, very persuaded by it. You know,
so I think something is happening. I know there was
some sense that maybe Muse wasn't all it was cracked
up to be last week, but something is clearly going
on right now.
Speaker 3 (33:54):
Muse is cool. It's not going to be the answer, though. Like,
I can already tell you this, like, Muse is not
going to be the consumer AI app. And actually, I
don't think the infrastructure is there. I think, unfortunately, we
need some more data centers. Like, there's not enough compute
right now for everyone to.
Speaker 4 (34:08):
But don't you think they have, like, I'm just saying,
you know, when I talk to my parents, they might
use Facebook's AI, which is, this is what Zuckerberg's been
banking on, right? Which is that, like, the average person's
not going to go and download ChatGPT. My mom doesn't
trust it. But if there's AI built into an app
that she's using every day, like, why wouldn't that be
the answer?
Speaker 3 (34:25):
I think that makes, that's a logical argument. It makes
total sense. I just, I feel like after covering this
for a while, I have like a good sort of
gut sense about this and like Muse is not it,
but maybe Facebook will come up with, or Meta will
come up with another one or Muse will evolve or
something like that. Like I would leave open that possibility.
Speaker 1 (34:42):
Well, I mean, Marissa Mai has said that part of
the reason for doing this is that the kind of
foundational AI that already exists isn't really working for normal people.
So the insight may be... Go ahead, Natasha.
Speaker 7 (34:53):
No, I actually love the idea of people going through
my camera roll or just like any camera roll technology.
It's like the closest I would come to a diary
in terms of the number of screenshots and photos and...
I just, I remember there was a startup a few
years ago that was about organizing it. And I just
feel like it's one of those things that is not
part of a tech dad presentation that would be at,
(35:16):
you know, Meta or Apple, but actually a lot of
people would love to have a way to- I feel
like as a tech dad, I just have to- I
feel like this is like a Gen Z product.
Speaker 6 (35:25):
I think young people.
Speaker 5 (35:26):
Yeah, but I mean, I'm.
Speaker 7 (35:27):
Not Gen Z, but I mean, yay for people like
looking and messing with my.
Speaker 4 (35:32):
I mean, Snapchat, when I talk to my like young cousins,
Snapchat's like the only app they use.
Speaker 6 (35:36):
They are only communicating via photo.
Speaker 9 (35:38):
Yeah.
Speaker 6 (35:38):
So I love that.
Speaker 5 (35:39):
And it takes away.
Speaker 7 (35:40):
I mean, I don't do anything wrong. So like maybe
if you're worried about like, who are you with in
your photos or anything?
Speaker 5 (35:49):
But I find that great.
Speaker 7 (35:50):
Like if you're not sharing my photos anywhere, you're just extracting,
you know, like. hey, I don't know, like your son
needs new sneakers that are obviously very beat up or
something like that.
Speaker 6 (36:00):
It's insulting your clothes?
Speaker 5 (36:02):
No, no, like, no, like you're running out of things.
Speaker 8 (36:04):
I don't know.
Speaker 5 (36:05):
I'm just saying.
Speaker 6 (36:05):
If my AI app is like, actually, your sneakers are busted.
Speaker 7 (36:11):
I just am saying I have seen zero innovation. Here
is a product, my camera roll that I am interacting with.
I don't know how many times a day. It's just
stored full of information, even if somebody could just organize
it better into folders than me, I mean, yes, please
put some innovation towards this. I cannot believe that there
(36:31):
hasn't been more technology around it.
Speaker 3 (36:34):
I mean, I agree with what you're saying, but what's
the utility there? When I use, for instance, this dot,
I'm saying like, oh, I've got to respond to this
calendar request for an event that other people in a
different time zone are wanting to dial into. And then
I have to deal with like my kids pick up
from their activities. And I literally told my dot yesterday,
(36:57):
like just a voice prompt. And it just did all
those things for me instantly. Like, I don't think just
looking at your photos, like you have that kind of utility.
Speaker 1 (37:06):
I mean, Margot, I want to hear from you because
you're closest to the fabled Gen Z cohort. I am
Gen Z. I'm like the oldest Gen Z.
Speaker 4 (37:14):
But yeah, I was just going to say, like, I
don't think it would be for that, right? Like, I
don't really see how it would like serve as like
a real personal assistant. I think though that at least
for me, like my photos are kind of the most
concentrated data of like what I care about.
Speaker 6 (37:29):
It's like my cats and the best meals I've had.
Speaker 4 (37:33):
And the places I actually want to travel or I
actually go to when I travel.
Speaker 3 (37:36):
But then what's the utility though?
Speaker 6 (37:38):
Well, I could see being like, I'm.
Speaker 4 (37:40):
Going to Croatia, plan me a trip. And it's like, okay,
you take pictures of. You know, when you went through Europe,
you took pictures of all the churches and you seem
to really like a great meal at 2 p.m. You're
eating at 2 p.m.
Speaker 6 (37:49):
There. Like, here's a great restaurant.
Speaker 3 (37:52):
It is a reservoir of data. I would guess that
it would get a lot of things totally wrong by
misinterpreting your photos because you're taking photos of random stuff
all the time. Mine would be like, I suggest you
go to vacation at Home Depot. You seem to spend
a lot of time there.
Speaker 7 (38:04):
Well, honestly.
Speaker 5 (38:11):
But I would be curious to see what they would
find out about me from that. It's not known.
Speaker 7 (38:17):
I just feel like it's the anti tech dad. It's
the anti like make my reservation for me.
Speaker 5 (38:23):
You know what?
Speaker 7 (38:23):
I know that I've taken screenshots of Instagram posts in Croatia.
If they could go through and be like, here's actually
all your aspirational posts.
Speaker 5 (38:32):
Instagram screenshots.
Speaker 3 (38:33):
Why is tech dad your denigration phrase?
Speaker 6 (38:37):
This is such an intergenerational fight.
Speaker 3 (38:40):
My wife loves this stuff. We're super busy parents. You
want stuff that's going to help you out. Why is
that a tech dad thing?
Speaker 5 (38:47):
Okay, you know why? Because I write about this stuff.
Speaker 7 (38:49):
I've sat through many demonstrations at google meta apple and
i do not find that they reflect like what i
as even as a parent like when they talk about
things like i'm not posting my data on facebook you
know or i don't either that's what i'm signing.
Speaker 1 (39:06):
For today we're gonna host hopefully another intergenerational fight on
text how soon maybe even live from croatia.
Speaker 3 (39:13):
I feel like this is just starting to get.
Speaker 1 (39:15):
Good thank you all so much For Tech Stuff, I'm
Oz Veloshin. This episode was produced by Eliza Dennis. It
was executive produced by me and Julia Nutter for Kaleidoscope.
Speaker 7 (39:29):
And.
Speaker 1 (39:29):
Katrina Norvell for iHeart Podcasts. Kyle Murdoch mixed this episode
and wrote our theme song. A special thank you to
Margot McColl, Natasha Tiku, and Reid Albigotti. Please check out
all the work they put out into the world. We're
lucky to call them friends of the pod.
TechStuff News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.