Fresh back from London! In this episode, Eric Newcomer reunites with co-hosts James Wilsterman and Max Child of Volley to dive into the best moments from the Cerebral Valley AI Summit.
From buzzy startup founders to incumbent innovators, the London event showcased a rapidly evolving AI landscape. But what stood out most? Eric, James, and Max break down their favorite clips and debate the key questions driving the industry right now:
- Models vs. Applications: Are we back in a “models win” moment?
- The Uber Dilemma: Why did Uber spin off its self-driving technology instead of competing with Waymo? Dara Khosrowshahi’s reasoning sparks a heated debate about platform dynamics, marketplace power, and whether Uber made the right call.
- Figma’s IPO Moment: With their S-1 filing just days after the summit, Dylan Field defended Figma’s AI strategy and new product launches.
- The Science of Discovery: Can AI models trained on “boring rule followers” actually make Nobel Prize-worthy breakthroughs?
- The End of Reading?: Synthesia’s CEO made the boldest prediction yet — that kids won’t read anymore, and video will replace text entirely. The hosts wrestle with what this means for human intelligence and whether writing really is thinking.
The next Cerebral Valley AI Summit returns to San Francisco on November 12th!
Timestamps08:21 Uber’s Self-Driving Strategy and Market Positioning16:56 Figma’s IPO Bear and Bull case26:17 Harry Stebbings’ Interview Insights with Granola’s CEO32:11 Exploring AI’s Role in Scientific Discovery38:26 The Impact of AI on Reading and Writing
00:00:00
This episode is brought to you by Forethought.
00:00:03
Most companies build their customer experience in pieces.
00:00:06
Sales in one system, support in another, and onboarding
00:00:09
somewhere else. Forethought brings it all
00:00:11
together. Forethought is an AI system
00:00:14
made-up of advanced agents that handle sales, onboarding,
00:00:17
support, and retention. Each team manages its own
00:00:21
agents. The customer sees one unified
00:00:23
experience. Forethought powers over a
00:00:26
billion interactions every month for brands like Scale AI,
00:00:29
Cohere, Air Table, and Upwork. Learn more at Forethought dot
00:00:34
AI. Welcome back to America.
00:00:41
I'm here with Max Child and James Wilstroman.
00:00:45
Hey, guys. Hello, Eric.
00:00:47
Hey, Eric. We, we had a fun European
00:00:50
adventure, what, couple weeks ago with this Rebel Valley AI
00:00:55
summit in London. Yeah, we, we had a blast.
00:00:58
I don't know what. What did you guys think about
00:01:00
it? I thought it was tremendous
00:01:01
experience. I mean, it really drove home how
00:01:04
exciting the AI scene is in Europe, which I think you know,
00:01:08
in the sort of mobile and web eras, maybe the scene a little
00:01:11
bit beleaguered compared to America, but I felt like AI
00:01:14
startups and and big companies are totally popping off there.
00:01:17
I mean, we had granola, we had Lovable, we had Synthesia photo
00:01:21
room and then you know, on the big company side, obviously we
00:01:23
had Wave and Uber, which was super exciting and so.
00:01:26
Well, in classic Cerebral Valley, I feel like the the
00:01:29
buzzy startup founders are the real celebrities.
00:01:31
You know, it's like, I mean, people are excited about Dara
00:01:34
and we had Dylan Field at Figma, which, you know, what was like a
00:01:37
week away from filing its S1. But yeah, I feel like the
00:01:41
lovable guy and the granola guy, you know, we're definitely sort
00:01:45
of people of interest at the event.
00:01:47
My Spidey sense for for startup heat was was very was very
00:01:52
tingly around the lovable guy. He was he was Anton.
00:01:56
I interviewed him. But like, even just walking in,
00:01:57
he was being surrounded by like investors and partners and other
00:02:01
startups and you could just he was a look, get the vibe he's a
00:02:04
little mini celebrity in the the AI scene.
00:02:07
He's a He's a Unicorn now. He's a.
00:02:09
Unicorn now as of like today or yesterday?
00:02:12
Yeah, how how many? Like how many coding AI adjacent
00:02:16
companies are unicorns now? We've got like lovable replet
00:02:20
cursor windsurf. It's just like goes on and on.
00:02:24
And the narrative moves so fast. Are are is is it good to be a
00:02:28
coding startup now or are we already like oh man cost of
00:02:31
goods sold and everything are out of control?
00:02:33
And Windsurf was falling apart and luckily they were able to
00:02:36
save it. I don't know, right?
00:02:38
Since Cerebral Valley, though, we've had people like bouncing
00:02:41
between anthropic and back and to cursor and back, and we've
00:02:45
had this whole saga of windsurf getting chopped up.
00:02:49
I don't know if I had to stick my finger in the air right now.
00:02:53
I feel like the mood is it's good to be a model again.
00:02:56
I mean, you've got a a big sentiment that anthropic and
00:03:02
clawed are key to coding and you sort of rather be anthropic than
00:03:06
cursor. And obviously we're recording
00:03:09
this on Thursday, it's coming out on Friday.
00:03:11
I mean, open AI just get a live stream about agents and seems
00:03:15
like, oh, we're figuring out this agent thing.
00:03:17
I don't know, you know, they're they're at once, I think they're
00:03:19
trying to be supportive of MCP, you know, the sort of generic
00:03:23
open source interface across agents.
00:03:25
But then they're making the announcement that like, oh, that
00:03:28
thing you're trying to build a company around.
00:03:29
Yeah, we do. We do that too.
00:03:31
So I guess my take at the moment is models back in ascendancy,
00:03:35
which you'd hope so given their valuations.
00:03:39
Yeah. I mean, I think obviously we've
00:03:41
had this debate a million times, but I think ultimately it is
00:03:43
good to be the company that has put, you know, 5 to 10 billions
00:03:48
of dollars to work on something rather than the company.
00:03:52
So, yeah, yeah, yeah, that that then the company that is
00:03:56
slightly repackaging the thing the other guy did for five to
00:03:59
$10 billion. Like, I mean, there are
00:04:01
obviously I think application use cases here, but you know, in
00:04:04
the sort of. Bullish on applications, yeah.
00:04:06
Oh, no, I, I am bullish on applications, but I do think,
00:04:09
you know, anything that's sort of close proximity to the model
00:04:12
itself, which I think a lot of these coding startups are very
00:04:15
close proximity to the model itself in many ways is a
00:04:18
dangerous place to be because, you know, Dari, Dario and Sam
00:04:21
have got their eyes on that space.
00:04:23
It's like you don't want to be the most popular application
00:04:27
that makes the most money because then you're square in
00:04:29
the targets of the the foundation model labs.
00:04:33
It's like you kind of want to be like making good money, but not
00:04:36
like the number one use case of LLMS.
00:04:40
We're going to get into clips so this episode is really we're
00:04:42
going to engage with the key moments at this Rupal Valley AI
00:04:45
summit, which I think will be fun.
00:04:46
Well, you can go watch them all on YouTube, and I think they're
00:04:49
all still pretty current even though AI moves fast.
00:04:52
But we're bringing you our favorite clips so that if you
00:04:55
don't have the time, you can get the best of what happened there.
00:04:58
I wanted to say two different things.
00:05:01
One on London. I think one thing I really liked
00:05:03
about the event is that we took a sort of like, AI is global.
00:05:06
We're not getting dragged into like, which city is the best?
00:05:09
I mean, obviously London has deep mind and there are like,
00:05:11
good companies in Europe, but it's like, yeah, it's a global
00:05:14
world, all connected on the Internet.
00:05:15
They're great people everywhere. Like, I think like half the
00:05:18
audience was London and half came from elsewhere probably.
00:05:21
Like, I don't know, 1/4 US and a quarter rest of Europe, I don't
00:05:25
know. Yeah, decent amount of like
00:05:26
France and Sweden and things like that for.
00:05:29
Sure, though we had a train, yeah, Public transportation
00:05:32
infrastructure, despite being like much, you know, better on
00:05:36
its surface, uh, consistently we had transportation problems
00:05:40
across Europe. Eurostar was in trouble from
00:05:44
France, and then I had my flight deflayed out of France from an
00:05:47
air traffic controller strike. So I don't know.
00:05:50
They brag about public transit. There are transit and then they
00:05:53
can't keep it working. It's like like their healthcare
00:05:56
system or something you know works works great for most
00:06:00
people most of the time and then falls apart at the at the
00:06:05
limits. The other thing is we we have
00:06:07
not officially committed to going back to London.
00:06:09
This is, it's like we need, we need a sponsor bidding war
00:06:14
between New York and London, between which one we had for
00:06:19
first. So if you if you have a strong
00:06:21
horse in the New York versus London race, reach out about
00:06:24
sponsoring. I would just say in terms of
00:06:27
vibes at the event, the the London audience was dramatically
00:06:31
more engaged than the New York audience.
00:06:32
I would say they were excited to be there.
00:06:36
They stayed for the whole event. They they engaged with the
00:06:38
speakers. They wanted to chat and meet
00:06:40
afterwards, whereas well. Those New Yorkers, they're
00:06:42
grinding. They're like, I don't fucking
00:06:44
job like. New Yorkers are like, I've got a
00:06:46
happy hour at 3:45 that I've got to be at, so the 4:30 speakers
00:06:51
will not be. Fully attended.
00:06:53
New York invited. Before I go back to work.
00:06:55
Until, yeah, you live. In New York, you're a bit of a
00:06:57
booster. You did have a little bit of
00:06:59
like, you know, we had Goldman, we had Snowflake, it had a
00:07:02
little bit of like big company tries to like embrace AI,
00:07:06
whereas I do think it turned out that Europe had some of the
00:07:10
cooler up and coming startups. I mean, I think Rogo, you know,
00:07:14
which we had at the fintech event is buzzy New York
00:07:17
application company. I mean, there's certainly, I'm
00:07:20
sure if we dig dug into like finance AI startups, we'd get
00:07:24
more of the like application excitement in New York.
00:07:27
I flew back from San Francisco a couple weeks ago to be on a New
00:07:34
York Tech Week panel. And of course, we, you know,
00:07:37
from one of the New York boosters, Julie Samuels.
00:07:40
I don't know who it is. I forget what our group is these
00:07:42
days, but you know, does advocacy for New York's tech
00:07:45
scene was like, oh, what do you think about the New York's tech
00:07:47
scene? And I gave the most like, oh,
00:07:49
pro San Francisco, just like, you know, the beauty of New York
00:07:53
to me is that people come through.
00:07:54
It's everybody's like second favorite city no matter where
00:07:56
you are. But I'm not sitting here
00:07:58
claiming that New York is the center of the tech universe.
00:08:02
And they're like, you are not invited to next year.
00:08:05
I know, I just believe I've said this a million times, but I
00:08:10
agree with Ryan Hoover, the product hunt guy who said to me
00:08:15
during the pandemic, Silicon Valley is online.
00:08:17
Like come on, it's like it's on the Internet.
00:08:19
Like who cares where people are? All right, back to the present
00:08:23
moment. A lots happened we're in now
00:08:25
this like AI spending war for employees, obviously with Meta,
00:08:29
which is I think the most present happening.
00:08:32
Obviously scale deal just happened when we started the
00:08:35
London event Yeah, let's get into clips and let's react to
00:08:39
some specific lines from the most recent Cerebral Valley AI
00:08:42
summit. I mean in the Uber case, I mean
00:08:44
you came in, you did spin out Aurora like how did.
00:08:48
You we we merge ATG into Aurora. Yeah, and then became its own
00:08:52
company. How did you decide, I guess that
00:08:55
self driving wasn't there in 2017 or they this shouldn't be
00:08:59
something to invest in or how did you go into sort of all your
00:09:02
technologists saying, oh, we're just tomorrow, just tomorrow and
00:09:06
say I'm not convinced. Listen, I I think Erica was a
00:09:09
really tough decision at the time.
00:09:11
And one is we were deeply marked by the fact that we had an
00:09:15
accident, lost life there. And so the responsibility that
00:09:21
you hold in your hands when you're operating in the real
00:09:24
world hit us deeply. I think the other circumstance
00:09:30
that affected the ultimate judgement was that we, we ran
00:09:34
ATG very separately from mainline Uber because we wanted
00:09:39
the opportunity to partner with the rest of the ecosystem.
00:09:41
Again, like we, we don't just work with, you know, we don't
00:09:45
just have Priuses on our platform.
00:09:48
We got forth, we, we want to work with the entire ecosystem.
00:09:52
And when I went and talked with some of the other AV developers,
00:09:55
I'd say, you know, ATG, it's in the family, but it's separate.
00:09:59
We're going to treat you fairly, we're not going to share data,
00:10:00
etcetera. You know, when you're having a
00:10:02
conversation, actually, like I usually do with you and I'm
00:10:05
saying stuff and you don't believe a fucking thing I say,
00:10:07
pardon me, That's that was the response.
00:10:11
Like the we were having conversations, but I think
00:10:13
people didn't trust us. Yeah, because they also viewed
00:10:16
us, rightly so, as a competitor, right.
00:10:18
So we had a decision to make. Do we go with a proprietary kind
00:10:24
of vertical strategy or do we truly partner with the industry?
00:10:29
We couldn't have the best of both worlds and ultimately we've
00:10:33
started that a partial model was the right model going forward
00:10:36
and based on, you know, basically all of the leading
00:10:39
players sans Tesla, but everyone else is open working with us,
00:10:44
etcetera, I think it was the right decision to make.
00:10:46
I guess I don't really buy this explanation honestly, because
00:10:52
like I if you think of like Amazon today, right, they sell a
00:10:56
lot of their own, you know, Amazon branded products on their
00:11:00
marketplace. It pisses off their suppliers
00:11:03
and they do it anyway because it's a good, a good business
00:11:05
idea. Like I I just don't really buy
00:11:07
that this is a good enough reason not to invest in your own
00:11:09
self driving technology just because other providers on the
00:11:15
marketplace won't be happy about it.
00:11:17
Right. I mean, the premise of my
00:11:18
question was that I thought it was skepticism on the imminence
00:11:23
of the technology working and Uber's team particularly working
00:11:28
and Aurora, which is what they spun off has not really
00:11:31
delivered on a major successful. That might be the real answer I
00:11:35
but that was. So to me, I did think like a
00:11:38
call on the quality of the technology was the reason.
00:11:41
But I mean, I don't know, I take him in his face.
00:11:44
I mean, it's interesting to that his actual first answer was that
00:11:49
they had such a negative PR situation that could they really
00:11:53
do it? And I, I think it's easy to
00:11:56
forget how much Dara was brought in to like clean up the PR
00:11:59
situation. It's like we had the hack.
00:12:01
We and they had literally, you know, launched self driving cars
00:12:05
in the firestorm of their PR nightmare without the approval
00:12:10
of like San Francisco. It was like just brazen.
00:12:13
I mean, I do ultimately think that if Uber can sort of
00:12:16
maintain it's like marketplace, you know, middle man position
00:12:20
between riders and drivers, even if the there are no drivers and
00:12:24
then, you know, it's just way MO's or, you know, in the case
00:12:27
of Serbo Valley London, it was wave they were partnering with.
00:12:29
And it's plausible that for, you know, 120 different countries
00:12:33
around the world, there's going to be, you know, 60 different
00:12:35
self driving companies you need to ultimately partner with, you
00:12:38
know, to, to serve self driving cars to people.
00:12:41
And all these different locations that Uber may
00:12:43
ultimately be the sort of aggregator of this need for a
00:12:47
car, whether it has a person behind the wheel or not.
00:12:50
Right. And I think if that strategy
00:12:51
works, I think they're in an awesome position because kind of
00:12:54
being the middle man in this massive market of personal
00:12:57
transportation is, is really great.
00:13:00
I think what's what's the challenge is in San Francisco,
00:13:03
there's this sort of Waymo brand loyalty building among the tech
00:13:07
community. I think like people like James,
00:13:10
you know, are like, I basically only take Waymo's, even if
00:13:12
they're take longer and they're more expensive.
00:13:14
And I use the Waymo app, right? And so it's an interesting
00:13:17
question. Like for someone like James, I
00:13:19
would pose like if Waymo's were in Uber in any city you wanted
00:13:23
to visit, and you could just instead of clicking Uber X, you
00:13:26
click Waymo. Like would you just go back to
00:13:28
using the Uber app because like sometimes you want to use the
00:13:32
Uber X because it comes faster and sometimes you want to use
00:13:34
Waymo because you like it more like, or would you be like, no,
00:13:37
I'm always using the Waymo app, you know?
00:13:39
I do think a lot of Uber success here depends on their ability to
00:13:43
specifically convince Waymo that they should not deploy in every
00:13:47
city and need to participate in the platform.
00:13:51
I mean, you know, and they're running those experiments now.
00:13:54
So it'll be interesting to see where Waymo thinks the power
00:13:57
resides. Yeah, I think, I think I would
00:14:00
that to that point, like if, if I know there's a Waymo's in
00:14:05
certain cities that I can only access the Uber, it'll probably,
00:14:10
you know, switch the habit back to just using Uber everywhere,
00:14:12
including in San Francisco. If I can access Waymo's through
00:14:16
through the Uber app. All of this though is just, you
00:14:20
know, pointing back to my original point, which I just
00:14:22
think Uber would still be better off if they owned something
00:14:27
approximating Waymo right now. Like I just don't see a world
00:14:30
where they are worse off if it's.
00:14:33
Waymo Quality. If it's Waymo, sure.
00:14:36
Yeah, yeah, yeah. And I who knows if they could
00:14:38
have gotten there. Right, Amazon knows that they
00:14:40
can deliver. You know, batteries just as well
00:14:43
as Duracell or whatever, Yeah. And Costco and places they they
00:14:46
pick lanes where they see where they can offer something.
00:14:50
I'm not saying it was like a bad decision to spin this off
00:14:53
originally, I just think disingenuous to say that he's
00:14:57
only doing this for platform neutrality.
00:15:00
Well. Literally like a day or two
00:15:02
after the event, there was reporting in the New York Times
00:15:04
that Uber might be working with Travis Kalanick to like acquire
00:15:08
a Chinese self driving car startup.
00:15:10
So it's, it's very possible this is on Dara's mind that we,
00:15:14
they're all these Chinese startups that have actually
00:15:17
caught up really well in self driving.
00:15:19
And we spun off the one that wasn't succeeding and cost us a
00:15:23
ton of money. Now everybody's ripping each
00:15:25
other off and they're all going to catch up to each other just
00:15:28
like the language models. And so then we have our white
00:15:31
label version, so nobody can push us around too much Like,
00:15:35
yeah, I think that's very plausible.
00:15:37
And they skip a lot of the needless money losing that's
00:15:41
hard to do on the public markets.
00:15:43
It's like Google Pixel or something, right?
00:15:45
Yeah. You want the flagship that just
00:15:47
kind of shows how to do this and you want to learn even if
00:15:50
you're, you know, also distributing Android to other
00:15:54
phone makers. I do think, yeah, to Eric's
00:15:56
point, Waymo is so head and shoulders above the field right
00:15:59
now that you would just be terrified that you're going to
00:16:01
lose Waymo. And so you got to balance these
00:16:02
considerations of like, is it going to piss off Waymo if we
00:16:06
work with this Chinese startup? Like, but what if Waymo just
00:16:09
walks away tomorrow? Like we better have a backup
00:16:11
plan, you know, so it's, it's a challenging place to be and.
00:16:13
I think the Waymo brand is so good right now.
00:16:16
You got started in like Jaguars. You have none of the baggage of
00:16:20
like bad Uber rides. It's just like it's such a good.
00:16:25
The MPs on Waymo has got to be like great relative for actual
00:16:29
like riders. Yeah.
00:16:31
Anyway, it's uh, it's, I mean, it's like a quintessential
00:16:34
technology debate question, which is just like, what are the
00:16:37
platform dynamics? Who actually has leverage?
00:16:39
And it'll be interesting to see it play out.
00:16:41
But it sounds like we all agree that it would serve Uber to have
00:16:45
some credible self driving car technology of its own, at least
00:16:49
to understand the fundamentals and have a little bit of a
00:16:53
competitive edge against potentially bullying partners.
00:16:57
All right, well, we did Uber. Let's do the other big headliner
00:17:00
from the event, Dylan Field, the CEO of Figma, who literally like
00:17:05
he seemed very busy when we saw him and I can see why.
00:17:10
I think they literally flipped their S1 filing the next week.
00:17:14
So very appreciative that he made it to London at sort of the
00:17:18
busiest possible moment for a pre IPO company.
00:17:22
I wanted to talk about sort of the company building piece of
00:17:24
this for a second. I mean, you're, you're clearly
00:17:28
sort of moving towards going public.
00:17:31
I'm curious like what your decision making process has
00:17:34
been. We're in this moment where
00:17:36
Stripe, SpaceX, like it feels like they can stay perpetually
00:17:40
private. What's sort of the impetus to
00:17:43
potentially move Figma towards the public markets?
00:17:46
Yeah. I mean I can talk like I told
00:17:48
you about anything regarding like that process, but I think
00:17:53
that just the important thing that I've been like making sure
00:17:58
that I'm grounded in everyday. Whatever happens with going
00:18:02
public, not going public is this is like the moment to make sure
00:18:07
that we are all in on product, whether that's me, our team,
00:18:13
like there is so much to build, so many workflows changing, so
00:18:16
much happening and we have such opportunity to help our
00:18:19
customers. So that's what we're pushing for
00:18:22
every day. It's funny to hear that played
00:18:24
back, like as an interviewer, him being like, I told you, I
00:18:27
couldn't talk about this, like seared in my mind in a way that
00:18:31
watching it was like, OK, he said that and he moved on.
00:18:34
I do think he could have been slightly less defensive.
00:18:37
To me, it's always better just to be like, I, they ask
00:18:40
questions they feel like they have to ask and I can say
00:18:43
whatever the fuck I want that you could say something totally
00:18:45
different. He could have pretended like I
00:18:47
didn't even ask the question, you know?
00:18:48
Anyway, besides that little like bristle at, at my just doing my
00:18:53
job and asking a question, it's clear that they, they feel the
00:18:56
need to have like a great sort of pipeline of products heading
00:19:00
into an IPO and they, they literally are presenting
00:19:03
themselves as having launched what 4 fresh products after
00:19:07
having four products just this year.
00:19:09
My take is that he like actually believes that they're in a
00:19:15
disruptive moment. We, we, we talked about in the,
00:19:18
in on the podcast before where technology is changing and will
00:19:22
affect them in many ways through AI.
00:19:25
And then also, you know, he it's a message to his team to say
00:19:29
like, you know, even though we are going public and this is an
00:19:32
amazing milestone for the company, like don't get
00:19:35
distracted because we actually have to fight those battles and
00:19:39
build product. Yeah.
00:19:42
I think it's interesting given our experience with these vibe
00:19:46
coding tools over the last couple months, James that like
00:19:49
to me it increasingly seems like the bottleneck for vibe coding
00:19:53
or AI assisted coding or whatever you want to call it is
00:19:56
the quality of the design. I would say that like the
00:19:59
designs you get out of these tools, whether it's like.
00:20:03
Cursor or lovable or Claude code or whatever or just garbage like
00:20:07
really bad. And it's a sort of interesting
00:20:10
question if like Figma is extremely well positioned in
00:20:13
that case, because they have all the designers, like all the good
00:20:17
designers in, you know, in software, essentially use Figma.
00:20:21
And they can actually convert this sort of group of people who
00:20:24
actually know how to make things that look and feel really great
00:20:27
and polished into becoming creators and developers.
00:20:30
Or if you go the other way, which is say like, Oh no,
00:20:33
everyone else is going to like actually grab these designers
00:20:37
and they're going to start understanding how to use cursor
00:20:39
and Claude code. And then once cursor and Claude
00:20:41
code level up on their quality of design by either training on
00:20:45
new data or just something else, then all these designers are
00:20:48
just going to be using Claude code or cursor, which is kind of
00:20:51
what we experienced in the company in the last few weeks
00:20:53
when we trained everyone on Claude code.
00:20:54
It's like all of a sudden every every person in the company,
00:20:56
whether in design or product or data or marketing can can now
00:20:59
build apps. They all just look like garbage
00:21:02
because the models aren't there yet.
00:21:03
I I think Figma is in a funny situation launching Figma make,
00:21:08
which is their sort of lovable, you know what have you vibe code
00:21:13
platform, vibe code platform is that what?
00:21:16
We're calling the category. I mean, I don't know what
00:21:17
they're calling it, but that's what I'm calling it.
00:21:19
Yeah, I mean. It puts them in a weird position
00:21:21
where they have to sort of defend the category, which I, I
00:21:25
think is still like messy and broken.
00:21:27
I, I think they, they created like Figma.
00:21:30
I mean, a lot of what Figma is doing is like sort of you're a
00:21:33
marketer or you're a website builder.
00:21:36
We're going to give you sort of a lightweight version of Figma.
00:21:38
So it's easier for you to get in and do what you want.
00:21:41
And so they created Figma sites for web creation and that has
00:21:45
like a Figma make integration. And in some ways I think that
00:21:50
makes more sense with Figma's like long term plan where it's
00:21:54
like deeply integrated with their design elements.
00:21:57
And Figma make feels a little bit like, all right, there's
00:22:01
this sexy thing that everybody has.
00:22:02
It's just sort of like a wrapper over models.
00:22:05
If there's going to be so much zeitgeist around it, let's have
00:22:08
it too. But I agree that the real
00:22:10
opportunity for Figma is that like they have all your design
00:22:13
assets. They have all your design
00:22:15
assets. It's a moment when context is
00:22:18
seen as like extremely important and Figma has all your context
00:22:22
on how you design things. And so clearly the more they can
00:22:26
make this make thing much more like related to the actual
00:22:29
design tools and assets that Figma has, it'll be much more
00:22:34
valuable. And then it'll be truly
00:22:35
differentiated from Lovable and everybody.
00:22:38
Else, I mean, for the record, like I don't know if they're
00:22:40
gonna do this, if they have the talent to do this or if there's
00:22:42
it's a focus of theirs, but they have this tremendous opportunity
00:22:46
to essentially like trained as a, you know, the first good
00:22:50
model in terms of design for, you know, for vibe coding or for
00:22:55
creating software, right? I mean they have again basically
00:22:59
every like good design of software you know is on Figma
00:23:04
for the last like I. Don't know they can actually use
00:23:06
that data to train, but yeah. I mean, they should, I don't
00:23:09
know. They should.
00:23:10
Their lawyers should try to find some loopholes on this.
00:23:12
Yeah, I guess. What?
00:23:13
Would you be outputting? You'd just be outputting the
00:23:15
layers themselves, not the code you're saying like.
00:23:18
I don't know how it would work exactly.
00:23:20
Maybe you create your own version of Claude code.
00:23:22
Maybe you create your own version of, you know, a super
00:23:24
enhanced version of love of lovable that just makes really
00:23:26
nice looking websites instead of pretty crappy looking websites
00:23:29
like. But I'm just saying, you know,
00:23:31
to Eric's point about like context is king, you know, IE
00:23:34
like the data on how people, you know, use these tools or create
00:23:37
things with these tools. Like Figma has the greatest
00:23:40
goldmine of design context like in the world.
00:23:43
And I don't know how they capture that or if they're
00:23:45
thinking about it. But to me, it feels like this,
00:23:47
this massive upside opportunity to be the the creation tool that
00:23:51
actually makes things that look nice and work.
00:23:53
Well, the, the other thing Dylan bristled at me about was that I
00:23:57
said, you know, oh, when I use these like tools, I just like
00:24:01
fucking copy and paste, you know, sub stack or whatever and
00:24:04
say like, build me sub stack. And he's like, well, you're, I
00:24:08
don't know, it's sort of like you're lazy.
00:24:09
Not everybody else does that. I don't know.
00:24:11
But I do think like a lot of what works about these is you
00:24:15
start with some visual design that you like and sort of
00:24:19
replicate it. I don't.
00:24:20
Know no, I, I, I'm fully on your team here.
00:24:23
I feel like most design is most design is, is remixing, you
00:24:27
know, to sort of the remixing is the kind of like generous
00:24:30
version and I think the ungenerous version would be a
00:24:32
lot of copies. Design.
00:24:34
I think Figma is a tool where you're trying to build it up on
00:24:36
the pixel level, but a lot of these like text based almost.
00:24:40
It's either you're asking the thing to just arbitrarily come
00:24:45
up with something you don't know how, or sort of like copy and
00:24:48
pasting, putting things in. I, I realize you can put in your
00:24:50
own stuff, but I, I do think there's a lot of like copying
00:24:54
that's going on. I think a lot of this is take
00:24:57
right now is going to age like milk and in a year we're going
00:25:00
to be looking back on this in a world where the foundation
00:25:03
models have just chosen to train more on on good design.
00:25:07
You know, it won't be like Eric having to cut, copy paste
00:25:11
pictures of so stack to like copy it.
00:25:13
It'll just come out with better looking designs when you vibe
00:25:15
good, and I think that that's the future.
00:25:18
You can feel the AGI today, James.
00:25:21
Yeah, Yeah. I just, I just think the models
00:25:23
haven't like, you know, trained like the coding models haven't
00:25:28
focused on this yet and they will like it's just a matter of,
00:25:31
you know. Actually like a super short
00:25:34
figma case then I mean I'm just. Saying like.
00:25:36
I guess that's why. I'm saying they should be in
00:25:37
that race like I'm saying like they've got this goldmine.
00:25:40
They've got to get in here with these other models like, you
00:25:43
know, they could. They could be the leader.
00:25:45
I don't agree with the goldmine because they don't actually have
00:25:49
the connectivity to the code like at some point.
00:25:53
Like the way that clod code works, is it like generates
00:25:56
actual front end, you know, CSS? Yeah.
00:26:00
So I just think, you know, they, they have like a goldmine of
00:26:05
great looking imagery. All right, well, breaking news.
00:26:08
James's short Figma. James James's Shorting Figma I.
00:26:12
Predicted at the start of this episode that we would be high on
00:26:15
models today and that's that's bearing out.
00:26:18
All right Next up we've got Harry Stebbins, the only VC we
00:26:23
trusted to interview somebody. It helps that he has a podcast
00:26:27
of his own, so he knows what he's doing.
00:26:29
He interviewed the Granola CEO, Christopher Pedro Gall.
00:26:33
Here's the clip. My favorite quote is like will
00:26:35
the incumbent gain innovation before you know the innovator or
00:26:39
the startup gains distribution? How do you think about that
00:26:42
today, if you think about kind of an open AI's ability?
00:26:45
To do what? You do in some respects and have
00:26:47
distribution versus your ability to move fast.
00:26:50
Be very nimble, innovate. Yeah, I think my my view here is
00:26:57
that we are in you know, when the iPhone came out and you had
00:27:00
like the beer drinking apps and the lighter apps like I still
00:27:04
think we're in that stage of AI, which might sound ridiculous
00:27:07
because whatever ChatGPT has 500 million users and is clearly
00:27:10
useful, but I really think that is the case I don't think AI is
00:27:14
that useful. I think it's today and I think
00:27:17
it's mostly masquerading is useful, but you end up spending
00:27:20
a lot of time like rewriting or redoing it.
00:27:23
Another way to put that is it's going to be so much more useful
00:27:26
in one to two years. And and I think we have yet to
00:27:31
see truly AI native like interfaces and and products come
00:27:36
out. And I think that's the whole
00:27:38
question. Like I think the the it's not
00:27:41
about the products that exist today.
00:27:43
Those are. It's still like.
00:27:45
It's like the caveman days now. It's like who can get to the
00:27:49
truly AI native experiences and whoever builds those first and
00:27:54
gets distribution on that. That's that's my view.
00:27:57
And I think we're all everybody in the industry is pre that
00:28:00
right now. I basically agree with that, but
00:28:03
I'm not sure that means like bet on granola.
00:28:05
I don't know, it's like a little bit like I sort of
00:28:07
philosophically agree with him that there's a lot, you know, of
00:28:11
kind of latching AI onto existing products and workflows.
00:28:15
And even granola arguably, you know, is essentially just
00:28:17
latching voice AI onto a Zoom call essentially.
00:28:22
And the interface is, you know, not innovative.
00:28:24
It's like a notes app essentially with some folders.
00:28:27
It's a great product. For the record, I freaking love
00:28:29
granola, but I don't know if the the idea that like, oh, we're in
00:28:34
the beer drinking iPhone phase, so like air go, just wait one or
00:28:39
two years and then we'll see where the chips fall necessarily
00:28:41
makes me believe in someone other than opening eye, I guess.
00:28:45
I mean, it's a great. Fundraising posture because it's
00:28:47
like, it's cool now. Like, think about what?
00:28:50
Wait till next year. Really.
00:28:52
Yeah. I mean, I feel like yeah, the
00:28:54
the missing, yeah, the missing piece for me was the the
00:28:56
explanation of what the evolved version of his product looks
00:29:00
like, which maybe he doesn't want to say would have made to
00:29:02
that me that argument a lot more, right.
00:29:04
Yeah. I mean, it reminds me of like
00:29:06
the early social media days where, you know, first you have
00:29:10
all of these media companies just put their products on the
00:29:12
Internet, and then it turns out that the form factor that
00:29:17
succeeds dramatically is is Facebook or Twitter or
00:29:20
something, right? That's just like a totally
00:29:23
different product. But if we had interviewed the
00:29:25
Myspace CEO in 2003 and he had said, you know, we're not really
00:29:29
at the final form of social media, like, you know, it's
00:29:32
coming. And and then we're going to
00:29:34
really like unlock the value you get out of social media and
00:29:37
you'd be like, this dude's totally right.
00:29:38
But like, does that mean Myspace is going to win?
00:29:40
I mean, I don't know, it just it's sort of in, I feel like it
00:29:43
was a little bit of that kind of comment where it was like, so
00:29:45
you're saying like nobody's figured it out.
00:29:47
So ergo, like we're all just kind of killing time until the
00:29:50
models get better for. A year but but wouldn't the
00:29:52
wasn't Myspace like we're hot shit like I don't know.
00:29:55
I'm, I'm, I'm for the record, I'm not saying that Myspace CEO
00:29:58
would have said that. I'm just like.
00:29:59
Right. I'm just saying you'd prefer the
00:30:00
CEO who at least is like a realist about where we are if
00:30:04
he's correct. I guess if the Myspace CEO did
00:30:07
say that, they would be directionally correct in that
00:30:10
like, you know, most newspapers didn't win like incumbents,
00:30:15
right? But I guess you can argue like
00:30:17
this is where it gets to be like a tricky argument like open AI.
00:30:20
Are they really the incumbent or are they also just the, you
00:30:23
know, another startup that is, you know, in the game as much as
00:30:29
granola? I just don't get the I want.
00:30:32
I mean, I have a I want transcripts and I know they give
00:30:35
it to you, but like I don't want to query it for general
00:30:38
takeaways. Like I want it to give me
00:30:41
specific lines out of the conversation.
00:30:43
And so the whole thing sort of annoys me because it does the
00:30:47
opposite of what I want, which is you could give me a gloss
00:30:50
you. It shows you the whole
00:30:52
transcript, but I want it to deliver like here are the five
00:30:54
best quotes or like give me the specific verbatim thing that was
00:30:58
said and it's always trying to approximate sort of what was
00:31:01
said. I think the way that Harry asked
00:31:04
the question was like, how do you think of incumbents getting
00:31:07
to, you know, the product versus the innovator?
00:31:11
But like in this case, we're not, we're calling Open AI an
00:31:14
incumbent, not an innovator. I don't know.
00:31:15
They're innovating. That's funny.
00:31:17
I mean, that's sort of the role Figma played where it's like,
00:31:19
oh, you're the incumbent executor now, like right.
00:31:21
And Notion is a similar one. There's this whole set of
00:31:25
companies that like, oh, in any normal sense, you'd be like,
00:31:28
this is my innovator moment to be on the public markets.
00:31:32
And now it's like, no, you're, you're already the incumbent.
00:31:34
There's a new wave. I think that was like a big meta
00:31:37
theme of the event. Right.
00:31:39
And I'm just saying we need to be a little bit more precise
00:31:41
with the definitions here, like who are the incumbents?
00:31:44
I don't know. Incumbents, anyone who are they
00:31:47
thought they knew what their company was pre catching you 3
00:31:51
and everybody you know the the upstarts are people who said oh
00:31:56
man, I have to invent my company with catchy BT in mind.
00:31:59
I just say I, I basically what I'm seeing is that companies
00:32:02
like Notion and Figma are very much like a, like live players
00:32:06
in this game. And they're they're really
00:32:07
trying to avoid these disruptions from from below.
00:32:11
Our next clip is from the first panel of the day with
00:32:15
Microsoft's Bonnie Croft and Hugging faces.
00:32:18
Thomas Wolfe. I, I introduced it saying this
00:32:21
was my smart people panel. It was the question of whether
00:32:24
we'll find novelty in AI and where where true discoveries lie
00:32:30
with the AI systems of the day. Let's hear what they had to say.
00:32:33
So Thomas, you, you, you know, you wrote this great essay,
00:32:35
basically, you know, asking the question of will AI produce a
00:32:41
Nobel Prize? Will it create an Einstein?
00:32:43
Like what? What was your conclusion there?
00:32:46
How optimistic or not are you about the ability of the current
00:32:49
paradigm to produce a true revolutionary idea?
00:32:54
Yeah, I'm not super optimistic on this right now.
00:32:57
Not optimistic. You almost whispered it.
00:33:01
You're like, this is in the crowd to admit it.
00:33:05
In most part, I'm usually the optimist because I think there's
00:33:07
a lot of great stuff you can do with AI.
00:33:10
And as I mean, probably you got right, you got the idea right
00:33:13
here, which is you have two, two things with AI.
00:33:15
So you have the assistant that's really getting amazing.
00:33:18
And I think, I think training DfT models, training A4, these
00:33:22
are amazing assistants and they will definitely lead to some
00:33:25
interesting. Discovery, like we were
00:33:27
discussing yesterday where we started to talk about this panel
00:33:30
discovery, this discovery will come from still from this part
00:33:33
of creativity of a human saying I'm going to use alpha fault to
00:33:36
explore these proteins and that will be the thing.
00:33:39
And then Alpha Fault helps you find how it works.
00:33:42
And then you have the other thing which is taking this
00:33:44
trillion parameters model and ask them.
00:33:47
Invent something now invent something new.
00:33:49
We want to travel faster than light.
00:33:51
How should we do? Right?
00:33:52
If you ask that to chat DPT nowadays, the answer we get are
00:33:55
pretty bad. Well, I don't really.
00:33:56
I don't really understand his take away exactly.
00:33:59
Would it count if a scientist used ChatGPT or ChatGPT 5 to
00:34:05
like win a Nobel Prize? I mean, I think he's saying that
00:34:08
that is clearly going to happen. I think both panelists were
00:34:11
bullish on AI models expanding like the search space for
00:34:15
discovery or being like, we're going to be able to like I'm an
00:34:19
expert in material X and it will consider material Y, but it
00:34:24
doesn't think this sort of like contrarian, like think outside
00:34:28
the box, reinvent core premises that AI is capable of doing
00:34:33
that. I mean, I think he's saying like
00:34:35
if you just ask the big model like right off the shelf, go
00:34:38
invent something for me, it's not very good at that.
00:34:41
I agree with that. That could change, but that is
00:34:43
the current case. I guess that doesn't really
00:34:46
matter. I mean, like, we're gonna have
00:34:48
if, if, even if it's just, you know, humans have to like, tell
00:34:52
it to do something that's adjacent to their own research.
00:34:54
Like that would still be revolutionary.
00:34:57
Yeah, I'm definitely with jams here.
00:34:59
I mean, you can't just walk to a random human scientist and be
00:35:02
like, oh, go figure out, you know, the nature of the universe
00:35:04
or whatever. I mean, like they're it's like,
00:35:06
OK, well, give me 50 years and I'm gonna go do research and dig
00:35:09
into things, right? He's saying we're training these
00:35:11
models on the boring rule followers, which are him.
00:35:15
He includes himself in this and the vast majority of us.
00:35:18
But he's saying that the true novel discoveries are created by
00:35:21
independent thinkers, and that we are not training AI to be
00:35:26
that sort of revolutionary. I mean, I just don't believe
00:35:28
this. So obviously none of us are
00:35:30
scientists, but I do think don't.
00:35:32
Believe that there's this persona of.
00:35:36
Of course, I believe that being an independent thinker is
00:35:39
incredibly important to making novel discoveries.
00:35:41
But I think so much of the way novel discoveries come about is,
00:35:44
is not just this personality, but it's, it's searching the
00:35:47
sort of frontiers of, of what we know and, and being aware of,
00:35:52
you know, to go Rumsfeld here, like, what are the known
00:35:54
unknowns? What are the unknown unknowns?
00:35:56
You know, and, and even in that protein folding example, you
00:35:59
know, you could say, OK, well, here's like, you know, 10
00:36:03
proteins that, you know, I think might be interesting in terms of
00:36:06
like being an agonist for this receptor.
00:36:09
Like what are the top 100 you think I should look at or
00:36:11
whatever? And, and like, even just
00:36:13
narrowing down that list from 10 to 100, you know, using a
00:36:16
genius AI is like 100 X improvement in the efficiency of
00:36:20
this search space where you're trying to discover, you know,
00:36:23
new science, right? So like, I just don't, I don't
00:36:26
really believe that Like, you know, dude, sitting alone, like,
00:36:29
you know, coming up with crazy ideas is, is how most scientific
00:36:33
discovery happens. I think a lot of it's searching
00:36:36
these efficient frontiers that we've discovered so far and, and
00:36:39
trying to continue to push them forward in interesting ways.
00:36:41
And being 100X efficient is filtering ideas or remixing
00:36:45
ideas or sort of creating slight tweaks on what we know so far
00:36:48
is, is so much of I think discovery, which I think AI
00:36:51
would be tremendous at. Yeah.
00:36:53
Connecting ideas across disciplines, right?
00:36:56
Like you know, you, you know a lot about both biology and
00:36:59
chemistry and physics or something, right?
00:37:01
Which I think the models are clearly best in the world at in
00:37:05
terms of depth of knowledge in every scientific discipline.
00:37:08
So I think if we believe that connecting the dots across these
00:37:13
fields is important to discovery like I think we're, you know,
00:37:16
tracking in a great way there for for the models.
00:37:19
I think the defense is 1 like the models are pretty capable
00:37:23
now and like even in the sort of like, OK, we're not
00:37:26
revolutionary thinkers, like find the cool stuff.
00:37:28
I feel like it's a little underwhelming.
00:37:30
I mean, I I still don't think it's like delivering me scoops.
00:37:33
So, but, but putting putting that aside, I, I think we're all
00:37:36
somewhat optimistic that like sort of rule follower, consensus
00:37:41
associative thinker, we'll be able to find a bunch of stuff
00:37:44
that we just haven't put together and there'll be this
00:37:46
period of a lot of value. To sum up my point here, models
00:37:50
are sycophants. Sycophants don't win Nobel
00:37:54
Prizes. You know, I think every great
00:37:57
man story in history is about sort of like rejecting the
00:37:59
consensus, I think. But my point is, yeah, they're
00:38:02
they're sycophants. The models are.
00:38:04
And like academics, you know, need to be free thinkers.
00:38:08
And until we can figure out how to train a model to really
00:38:11
resist core premises, we're reinforcing the status quo.
00:38:15
And so it might discover things that we would have already
00:38:18
concluded, but to really reject human thinking it, it feels like
00:38:22
that's that's not how it's trained.
00:38:23
That's that's how I see the argument.
00:38:25
I don't know if I totally agree with that.
00:38:27
Moving on to what I think will be our last clip of the day,
00:38:31
perhaps the most provocative, we have a great line from the CEO
00:38:35
of Synthesia, Victor RIP. Our belly definitely designed
00:38:38
for peak spice. Give it a listen.
00:38:41
I guess, you know, last question for you, Victor, like within a
00:38:43
company similarly, like what percentage, you know, how many
00:38:47
times a day am I going to be watching an AI generated video
00:38:50
that's explaining something to me like 5 years from now?
00:38:53
Is it daily, hourly, you know, every 5 minutes, everything I
00:38:56
do? Like what percentage of
00:38:57
corporate, the corporate experience is going to be
00:39:00
consuming AI generated video, you know, five years down the
00:39:02
road? I think I'll probably mirror
00:39:04
your personal life, right. So, I mean, I don't know, don't
00:39:06
know that well, but when my own like media mix of like what I
00:39:09
consume, yeah, I consume mostly podcasts, YouTube videos.
00:39:12
Sure. Tick tock videos, Instagram
00:39:14
reels, Instagram photos. I also do read, but way less
00:39:17
than I did five years ago. And I think that'll be kind of
00:39:19
mirrored in the workplace. It could be hard to imagine
00:39:21
exactly what it's going to look like, but I truly think that
00:39:23
that is clearly what people prefer when no one just tells
00:39:26
them what to do. And so I think that's what we
00:39:28
want to kind of see in the in the in the workplace as well.
00:39:30
And I have this, you know, often very provocative to people idea
00:39:33
that, you know, your kids, kids are not going to be reading and
00:39:35
writing. They're going to be watching and
00:39:36
listening almost exclusively. And I think that is going to
00:39:38
happen. I'm teaching my daughter to read
00:39:40
right now. But I think at some point I
00:39:43
think text will just feel a bit like vinyls or something, right?
00:39:47
Or like, we'll do it sometimes because it's like a fun
00:39:48
nostalgic thing, but for actual information dissemination, it's
00:39:53
going to be like a very low bandwidth compared to those.
00:39:55
Where's Paul Graham? That's that's my first reaction.
00:39:57
Paul Graham has been beating the drum.
00:39:59
Writing is thinking. And so I guess my biggest
00:40:02
defense of writing and why this terrifies me the most is less
00:40:05
that like everybody, I mean, it's good.
00:40:08
Reading is how you learn to write and writing is how we
00:40:11
think. And so if we don't have reading
00:40:12
and writing. Human intelligence as we know it
00:40:14
today collapses. That's not necessarily defense
00:40:17
to the lure of video, which I agree is very strong, but
00:40:21
definitely this vision is to me one of much stupider human
00:40:26
beings. Well.
00:40:27
Do you feel like right now in this podcast, you're not
00:40:30
thinking, you're not learning like I do?
00:40:32
I feel like I have to like, you know, think about what you guys
00:40:34
are saying and concoct my, you know, responses like on the fly.
00:40:39
I just think like, if, if I could constantly talk to other
00:40:43
humans, you know, whether they're actually AI's or not,
00:40:48
like, I don't, I think I'd still be learning, right?
00:40:50
Like, so I guess I kind of understand what he's saying.
00:40:53
Especially does both podcasts are in an intellectually lazy
00:40:56
medium compared to writing. Writing you really have to
00:40:59
figure out your argument. Like, I could just be I'm most
00:41:01
fun on a podcast being loud, boisterous, disagreeing,
00:41:04
slipping around. Like the All in guys have the
00:41:06
most popular podcasts in the world.
00:41:08
Writing requires you actually, like, organize your thoughts.
00:41:10
Somebody could breakdown an argument.
00:41:12
No, no, I Yeah. I think podcasting is a much
00:41:16
lazier way to engage with ideas than writing.
00:41:20
What about like a one-on-one tutor?
00:41:22
Just chatting, You know, one-on-one with mean?
00:41:24
Socratic, you know, if you really push it, but it's just so
00:41:27
easy. Conversation lends itself to
00:41:31
mean social and like slipping to the next thing.
00:41:33
And like This is why this whole, you know, there's a whole
00:41:36
Silicon Valley way of, you know, argumentation that's sort of
00:41:39
like we, we're following each other's ideas and like jumping
00:41:41
around. That's that's not how you would
00:41:43
write like a proof. And like, I just think if you
00:41:46
want, you know, argument and reasoning in sort of like proof
00:41:50
form writing is much better than talking.
00:41:53
Yeah, I just think we're talking, we're basically talking
00:41:55
about like the average person growing up in America, like how
00:41:59
will they learn in 12 to 15 years?
00:42:02
Like I definitely think there will be a lot more like
00:42:05
personalized tutoring that basically feels more like
00:42:09
working one-on-one with a human, you know, assistant, right.
00:42:14
And, and so less, less writing. I, I agree with that.
00:42:17
I don't know if it means no writing, but.
00:42:19
But is this fatalist? I I, I sort of agree with that
00:42:22
we are going to be pulled to less writing.
00:42:24
I'm just saying, do you think we'll be Dumber if we write
00:42:26
less? I don't know, man.
00:42:28
I think the percentage of Americans that like writes and
00:42:33
at all like, you know, essentially like rounds to 0
00:42:36
for, I would argue probably 98% of the population like.
00:42:40
And so I would sort of argue like you're already there in
00:42:43
this dystopia if you believe this is a dystopia.
00:42:46
And then on the other hand, yeah, I'm kind of more team
00:42:48
James here, which is I agree, writing is a form of thinking.
00:42:52
And I think having to prepare your thoughts in written format
00:42:55
allows you to distill and improve them.
00:42:57
But I also don't think it is the only way to think.
00:43:01
I mean, like thinking is thinking as well.
00:43:03
I mean, there is a whole, yeah, there's a whole genre of, you
00:43:06
know, shower thoughts, right? I mean, like Elon Musk always
00:43:09
says he comes up with his best rocket ideas while he's
00:43:12
showering and stuff. Like, is he writing them down on
00:43:14
the, you know, the squeegee, like, you know, on the wall of
00:43:17
the shower? I mean, like, it is pot.
00:43:19
I have always thought like, my best thoughts came from like,
00:43:21
exercising or like walking around or something.
00:43:24
You know, it's not writing, right?
00:43:25
And so I'm just not sure IA believe like, first of all, that
00:43:29
nearest makes no difference. Basically nobody writes in, you
00:43:32
know, the United States, I would argue and then BI think there
00:43:35
are other ways to think. So I, I sort of believe Victor's
00:43:38
thesis and I, I think it's a little bit less dystopian, but,
00:43:41
you know, you guys are both, I think better writers than I am,
00:43:43
so that you might find more. But you write company memos.
00:43:47
You can, Yeah, but things and but I don't.
00:43:48
Necessarily think that's the my best thinking.
00:43:51
You know, I'm not like looking back on my company memos and
00:43:53
being like, oh, those are much better thinking than when I have
00:43:55
a good conversation with James. I mean, this is a, this is a
00:43:57
deep sort of question, but I think about a fair bit.
00:44:01
I, I, and I'm a pretty auditory person, like love audiobooks and
00:44:06
everything. But I think in terms of writing,
00:44:09
you can have great ideas in the shower, great ideas on a
00:44:11
podcast, But like, what's unique about writing is the emphasis on
00:44:16
revision. And I think what writing has is
00:44:19
that you have to sort of go back and say, Oh, does that sort of
00:44:22
make sense? And so, you know, I think, you
00:44:25
know, smart people, you know, like the sort of like first
00:44:30
burst of like this great idea and like that is like really fun
00:44:33
on a podcast and like a first draft of writing.
00:44:36
But I think like a tight argument is often in revision
00:44:39
and writing is unique there where you have to sort of face
00:44:43
your ideas and and sort of beat them down and reform them and
00:44:46
make them work. Back to Polygram, you know Y
00:44:50
Combinator has basically a two phase application process, right
00:44:53
The first is super writing heavy you you answer all these
00:44:58
questions about your company I think Max and I found that to be
00:45:01
a very helpful process in some ways where it kind of forces you
00:45:05
to like refine your startup idea.
00:45:07
But the actual like interview process is 10 minutes of them
00:45:11
asking you questions in a conversation, right?
00:45:13
And I also found that to be like a, a very thought provoking
00:45:17
experience. So I don't know what whether you
00:45:20
need both or not, but they it seems like we're definitely
00:45:23
gravitating to a world where conversation is gonna be more
00:45:28
prominent and and you'll be able to do that like sort of Socratic
00:45:32
style of. Pushing.
00:45:33
I agree, I think this like Socratic dialogue is definitely
00:45:36
another sort of valid and interrogating line of thought
00:45:39
that is a little different than podcast scene in podcasting.
00:45:42
I think there's often an agreeableness and it's sort of
00:45:45
like we're all going in the same place and like sort of like
00:45:47
improv or something, whereas like we're a true like interview
00:45:51
Socratic dialogue. And that's I think where
00:45:53
teaching can fit in where you're really like challenging that
00:45:58
that can help. The last thing I'll say is like,
00:45:59
you know, you can write with Chachi BT.
00:46:01
So I, I don't know that I necessarily think what's
00:46:04
happening with AI is going to totally hurt writing.
00:46:07
It's making a great writing tool that really, you know,
00:46:10
interrogates you much, much more available.
00:46:13
Yeah, it does seem like people are probably writing more than
00:46:15
they were in the past now with ChatGPT, because you have to
00:46:19
write full sentences, you know, you know, Google was just about
00:46:21
keywords. You end up in these longer
00:46:23
dialogues. So I wouldn't be surprised if
00:46:25
like we're on like a slight uptick in in writing.
00:46:28
But I think the elites are probably in a downward trend.
00:46:31
I don't know. You read old, old like famous
00:46:34
people letters to their friends. I mean, I just feel like they're
00:46:37
like these long sort of reflections and some of these.
00:46:40
Right. But but I mean to come back to
00:46:42
the the overall point like was the quality of their thinking
00:46:45
like superior to ours? I just don't believe that.
00:46:47
Like, I think that if you look at any metric you can measure
00:46:50
overtime, you know, there's this concept of like, the Flynn
00:46:52
effect, which is that IQ keeps going up on, you know, on
00:46:55
average over the decades, right? So like, yeah, sure.
00:46:58
I mean, John Adams and, like, Abigail Adams, like, wrote some
00:47:01
amazing letters. But like the, you know, the at
00:47:03
least the average IQ in their society was much lower than
00:47:06
today, even though, you know, the elites were all writing much
00:47:09
more. So I'm just not kidding the
00:47:10
elites. Are way less literary today.
00:47:13
I mean you. Sure, that for sure.
00:47:14
Of course, because that was kind of the only medium for
00:47:18
expression right now. Yeah, and we're going to movies.
00:47:21
Yeah, Yeah, Movies contain a lot of information, but I I remain,
00:47:27
you know. Pessimistic about the world,
00:47:29
basically. You remain, I know I think AI is
00:47:33
going to democratize a lot of tools.
00:47:35
I think we're still in this period where this is allowing
00:47:38
much more people to get access to education than before.
00:47:42
So I think it'll be net good, but I think it's encouraging
00:47:45
some of us, like super elites, to be lazier than we should be.
00:47:50
Writing is a good way to think. This is a super elite podcast
00:47:54
for the record. Obviously I mean super elites,
00:47:57
yes. I just admit it.
00:48:01
Like nobody wants to see their elite.
00:48:02
Anymore like I plan to be a Davos this year with my fellow
00:48:06
super elites. Oh my God, I'm not trying to be
00:48:09
indulgent. I'm just trying to be honest
00:48:11
that we don't reflect like, you know, the average reading
00:48:14
writing consumption Max, you know, you're, you're teaching
00:48:18
your daughter to read. That's sort of an inspirational
00:48:20
story. What is your?
00:48:21
What's your? What the method to teaching or
00:48:23
display? Oh, I mean, this is this is the
00:48:25
beauty of it is it's an iPad game.
00:48:28
It's an interactive iPad game that's actually made by a
00:48:31
startup in up in Seattle. I believe it's called Mentava.
00:48:36
And this is not a. Paid ad This is not a.
00:48:38
Paid ad and IT. Costs a fortune you're somehow
00:48:40
getting it for. Free.
00:48:41
I think getting it for free because my daughter is becoming
00:48:43
an influencer, which is also an important skill, they're making
00:48:46
videos of her playing the app. But yeah, I mean, it's awesome
00:48:51
because I think the screen time has become such a meme among
00:48:55
parents that it's sort of objectively bad to have any
00:48:58
screens in your kids life, especially in the Super elite
00:49:01
circles that I roll in. And this app is teaching my
00:49:06
daughter to read it at, you know, years ahead of when she
00:49:09
would be learning these things in school.
00:49:10
And so I think it's just evidence to come back to the
00:49:12
more optimistic perspective on software and AI in general that,
00:49:16
you know, interactive experiences, even if they're
00:49:19
sort of purely educational, can be a really positive thing for
00:49:23
for kids on screens and using software.
00:49:26
And so I actually think that makes me get very long like AI
00:49:29
education and sort of software driven human improvement or
00:49:33
augmentation or whatever you want to call it, because she's,
00:49:36
you know, making progress at just a crazy rate compared to
00:49:38
school. So whether or not, you know,
00:49:41
videos or games are the way she learns how to read or learns
00:49:44
math eventually, I think there's clearly much better ways for
00:49:48
kids to learn things and, and in my opinion, a lot of it involves
00:49:51
software and AI. That's a much happier note to
00:49:55
end on. Thank you very much.
00:49:56
Alright. Well, we'll see you guys ahead
00:49:58
of the next Cerebral Valley AI Summit.
00:50:01
It'll be in San Francisco on November 12th.
00:50:04
Sure, we'll have a lot to catch up on in late October or
00:50:07
November. Alright, we'll see you later.