00:00I'll try not to I can spend the next 6
00:01hours just on this topic but I'm banging
00:03my shoe on the table but um why is it
00:05craft macaroni and cheese and not you
00:07know Jack Black macaroni and cheese
00:14provocative hello and welcome to this
00:18episode of the mark and Ben show where
00:19we're going to answer a bunch of
00:22questions um that you have submitted via
00:24Twitter via X um and the first question
00:31what is the best way to approach VCS for
00:34potential funding good Ben why did you
00:38start yeah well this is an interesting
00:40one because I I've answered this
00:42correctly before and people have gotten
00:45upset with me so I'm going to try and
00:47explain it in a little more detail so by
00:49far the best way so if you're an
00:53entrepreneur kind of getting the
00:56VC is a little bit of a test of
00:58Entrepreneurship because a big part of
01:01being a successful entrepreneur is being
01:04able to kind of network your way to the
01:07right person um and so by far the best
01:10way to kind of get to uh a VC is to kind
01:16of find somebody who knows them and I
01:19think that the good news is there are
01:23just a crazy number of people who know
01:26every VC because that's what VCS do they
01:28try to get to know every everybody um so
01:31it's pretty easy to find you know to get
01:33to whatever Six Degrees of Separation
01:35and chain your way in um you know the
01:38criticism is always well like you know I
01:41wasn't in that Circle um but you know
01:43look I didn't even know what a VC was
01:46really you know when when I first
01:48started raising money and I pretty sure
01:51Mark had no idea what a VC was so like
01:54you know that's actually helps you if
01:56you come if you can come from sort of
01:58outer space and work your way all the
02:00way to Venture Capital that's also
02:03that's more impressive than if you have
02:05a Stanford NBA or something like that
02:07and so you know you should just take
02:09that as a challenge but that's that I
02:11think is by far the best way to get
02:13started cold email is probably the worst
02:15way especially if you're guessing the
02:18address yeah and I i' bu I agree with
02:20all that and then I'd build on that to
02:21say look like um you know we I often
02:23refer to this as like the initial test
02:25like you know raising money from a VC as
02:26the initial test for a founder and then
02:28obviously meeting a VC as initial test
02:30for being able to raise money and I I
02:31describe those as the initial test
02:32because basically it's
02:35like those are the easiest parts of the
02:37entrepreneurial Journey um right yeah it
02:40turns out the rest is much harder
02:42exactly so um and you know it seems like
02:44it's daunting when you haven't done it
02:45before but like it really is the easiest
02:47part and the reason it's the easiest
02:48part is because VCS are literally in
02:49business to meet with startups and to
02:51give them money um like everybody else
02:54you're ever going to deal with is not
02:55actually in that business like everybody
02:57else is like busy right all the
02:58customers and all the potential like
03:00recruit you know employees you want to
03:01recruit and partners you want to have
03:02and everything else from here on out
03:04like they you know they're they're not
03:05actually just in the business of like
03:06paying attention to startups they have
03:07other things happening in their lives
03:09and so what you're doing when you learn
03:11how to network your way to a VC and when
03:12you learn how to raise money is you're
03:13basically learning a set of reusable
03:15skills that you will then use basically
03:17thousands and thousands of times over
03:18the course of the next 20 years to
03:20actually build a company um and for
03:22people who haven't and by the way I was
03:23I was like this I didn't I didn't know
03:24any of this when I started but like in
03:26sort of you know professional in sort of
03:28professional business World professional
03:30Tech World um you know post you know
03:33which is what you'll you'll find after
03:35you you get underway in sort of 100% of
03:37circumstances um the way people meet
03:40people is through connections um the way
03:41that people meet people is by finding a
03:43warm introduction somebody who knows
03:44somebody who can vouch for them that
03:45that that is how the business World
03:47operates and it always has how the Adult
03:50World operates and it it kind of always
03:51has been uh it turns out um and so you
03:54know as Ben said if you if you come from
03:56a sort of social background where that's
03:57a normal thing then you'll already
04:00if you come from outer space like I did
04:03um you know you you you'll learn it the
04:04hard way but it's a very important thing
04:06to learn because once you come to grips
04:07with that you'll realize okay this is
04:10actually how it works and so therefore
04:11what I need to do is I just need to get
04:12really good at that um and then once you
04:14get comfortable with that and then you
04:16apply yourself and you get good at that
04:17then all of a sudden you have this new
04:18superpower where you can you can reach
04:20lots of people and and and that's sort
04:21of the you know the sort of compounding
04:23ability to build a network and be able
04:25to reach lots of people is sort of a
04:26real key for for success in you know
04:30business yeah and in life by the way you
04:32know like to get anything done like it's
04:34a great skill to have yeah all right the
04:37next one from I'll not get this
04:40pronunciation right Gil
04:42giler um I always feel like somebody's
04:45gonna prank us one time and I'm gonna
04:47say something really hilarious on the
04:49name but giler says uh reflecting on the
04:53evolution of venture capital could you
04:55share which of your foundational
04:57investment dogmas have with stood the
04:59test of time and which have transformed
05:03throw out one that's definitely um which
05:06we got from John door which is uh really
05:11done well over time which is um the
05:15question is how High Is Up um and I'll
05:20just kind of leave it at that so A lot
05:22of times you you know people get very
05:25wrapped around the axle on everything
05:28from entry price to um you know all
05:32these questions but the the big question
05:35is is really like okay if it works then
05:38how big is it gonna work it's also hard
05:40to tell sometimes yeah so let's let's
05:42debate that for a second because you
05:43know we we talk about this a lot inside
05:44the firm um so you know I wrote this
05:47piece I wrote this blog post years ago
05:49um probably 15 years ago now called the
05:50only thing that matters uh or that one
05:53or a different one but it was it was the
05:54blog post where I talked about this sort
05:55of triangle the sort of classic Venture
05:57triangle of people product and Market
05:59Market um right and so this is sort of
06:01the Endless triangle that you kind of
06:03deal with in startups and in VC which is
06:04basically it's at the end of the day
06:06it's always basically who's the team
06:07what's the product and then what's
06:08what's the market um yeah and uh and so
06:12when I wrote that post years ago um I
06:14sort of based it on on uh actually on a
06:16different legendary vc. Valentine um
06:19yeah yeah yeah right he he always said
06:21the market yeah he always said market
06:22right so what a lot of VC's what a lot
06:24of D's peers and colleagues at that time
06:25said is it's either team it's the
06:27founders or they said it's basically the
06:28killer product and was always like no no
06:30it's the market and and Ben it goes to
06:31exactly what you just said it's like
06:33because what Don observed basically is
06:34you can have a great team with a great
06:35product and then they they enter a small
06:37market and then it just it never gets
06:38big you know it turns out how how High
06:40Is Up is just not very high um and so
06:42the whole thing is kind of a waste of
06:43time and then from a venture standpoint
06:45you just don't make any money because
06:46the outcomes are small from a from a
06:47startup standpoint you end up
06:49disappointed because you just you know
06:50you didn't build a very big business and
06:52and so he he always pointed a market
06:54size and then he he he took an of course
06:56an even more aggressive approach on that
06:57consistent with sequoia's kind of method
07:00um which was basically like if the
07:01Market's big enough like basically if if
07:03the market is really big then once the
07:05company kind of has prit Market fit you
07:07know the the team is somewhat fungible
07:09right you can kind of pull out the team
07:10you know Cisco you can fire the founders
07:12you can put in a professional CEO um and
07:14then the job of the professional CEO is
07:16to run and capture the market and his
07:17his his big thing always was look at the
07:19market first go to the big markets and
07:21then sort of everything else will will
07:23will take care of itself um you know
07:26well let me pause there like how Ben how
07:28would you reconcile that um with uh like
07:30is that the same thing that you're
07:31talking about from John or yeah so so I
07:34I actually think it's not quite the same
07:37thing in that um there there's kind of
07:41like if you were going backwards in time
07:44and you knew how big the market was then
07:47that would be kind of the answer to the
07:49question of how high is up but I think
07:51that you know when we evaluate deals
07:54what we always find is like the most
07:57impossible thing to evaluate is the the
07:59size of the market particularly if the
08:00market hasn't been If the product is
08:02going to create the market um and so
08:06then you're a little bit back to how far
08:09can this entrepreneur go and how big a
08:12breakthrough is this idea is a a little
08:15bit more how I would interpret how High
08:17Is Up um so is this something that's
08:21basically if it works it's going to
08:23affect everybody um and you know and
08:26this this the entrepreneur that is going
08:28to make it work is kind of the more
08:31relevant question to me than like what
08:35is the market size you know particularly
08:38for things what was the size of the
08:40electric car market when Elon started it
08:43seemed like it was Zero yeah yeah uh but
08:46you know now it seems like it's you know
08:49may be potentially bigger than the
08:51entire auto market combined and
08:54so you know that that I think is a bit
08:57of a trap to really try and size the
08:59market in a kind of business school
09:01methodology yeah Elon um so for people
09:04who don't know the history when when
09:05Elon started Tesla the the market for
09:07electric cars was so small was viewed as
09:09so small by everybody at the time except
09:11for except for Elon and his his
09:12colleagues um it was viewed as Tiny and
09:14and the and and it was actually said
09:16give me idea of how prevailing The View
09:17was and it was small was um you know the
09:19the main electric car up until that time
09:21in the modern era was called the the ev1
09:23and it was a it was a car made by
09:24General Motors probably 20 years ago uh
09:2725 years ago and um and it basically a
09:30handful of people bought them you could
09:31You' see them occasionally Silicon
09:32Valley that yeah they killed it like
09:35they they killed you haven't hurt yet
09:36people listen to podcast I'm sure I
09:37haven't even heard of it because they
09:38they killed it um and and it was such a
09:40flop um that they actually made a do
09:42there was actually a documentary movie
09:44called whatever Happ Who Killed the
09:45Electric Car who killed the electric car
09:47and literally if you watch the
09:48documentary it goes through basically
09:50the complete comprehensive failure of
09:51the electric car and basically and the
09:53big conclusion is oh the market there's
09:55just only the only people want electric
09:56cars are like a very small number of
09:57Rich hippies and like that's it
09:59um right and you can't build a business
10:01on that and then you know of course Elon
10:03and his colleagues came along and like
10:04completely redefined the market right
10:06and so yeah and you redefine the market
10:09by building the right product right so
10:11that's right exactly so um and then you
10:13know Ben this goes to you know kind of a
10:14contradictory piece of of guidance that
10:16I think has served a lot of
10:17entrepreneurs well um you know and let's
10:20talk about how we can reconcile that
10:21which is what Paul Graham says is sort
10:23of very very Contra you know that that
10:25maybe the John Dorp um or Don Valentine
10:27view of things what Paul Graham says is
10:28his famous advice to Founders is quote
10:30do things that don't scale um and you
10:33know what he means by that is like look
10:34like basically prove the prove the
10:36product Market fit up front um with you
10:39know by hand right like like you know
10:41his classic example here would be like
10:43something like Airbnb for example where
10:44it's like you know you have this like
10:45fundamental liquidity issue you have no
10:47Supply you have no demand you have no
10:48Marketplace you have no matching you
10:50know nobody's listing their houses for
10:51renters who don't exist and nobody's
10:53renting on on Airbnb if there's no
10:54Supply and so right so so you know his
10:56advice to like the founders of Airbnb
10:58would be something like like go out by
11:00hand and sign up a bunch of places right
11:02to uh list you know rooms to rent right
11:04bootstrap it bootstrap it and then and
11:06then go out by hand and find a group of
11:08people who might want to rent a room and
11:09that the particular breakthrough for
11:10Airbnb right actually was around design
11:13conferences right so they they
11:14specifically Target a design conference
11:15because they were designers and so they
11:17literally marketed to the people who
11:18were coming to San Francisco for a
11:20design conference here's you know here's
11:21how to rent a room you can St and then
11:23they they went out and like signed up I
11:24think basically by hand they signed up
11:26rooms you know available to rent in
11:27people's houses in San Francisco but but
11:30I bring this up because like do things
11:31that don't scale sounds antithetical to
11:34you know basically how High Is Up um and
11:37yeah I think that's right and I think
11:38kind of it goes with the other thing he
11:41says which is um make something that
11:43people want um and I think that's
11:47the that's kind of the the key unlock on
11:51the whole thing um but you know there
11:54are things that that people want that
11:57have uh not that much potential to them
12:00as well so and and and it is tricky it's
12:03much easier to go okay that's something
12:05that people want rather than um you know
12:08and this is exactly how many people what
12:09they'll be willing to pay for it that's
12:11a harder thing to figure out certainly
12:13in the kind of seed a stage yeah that's
12:16maybe the final final point I make maybe
12:18on this topic is um you know make
12:20something people want you know do a lot
12:21of people want this that's one question
12:23there there is a related and extremely
12:24important question which is will they
12:25pay for it yeah and how much how much
12:28much all you pay for it right exactly
12:30and so there there is something that you
12:31you see like look there have been some
12:33great Venture investments in things
12:34where the product started out as free um
12:37and you know we we can all L many
12:38examples of that you know are super
12:39cheap and then they walked up the price
12:40over time but you know you also do see
12:43cases um where you know it turns out
12:45there's something that people actually
12:46like a fair amount as long as they don't
12:47have to pay for it or or by the way as
12:49long as they don't have to pay for it at
12:50a rate that's actually both the cost of
12:52providing it right like you know some
12:54sometimes there are products that are
12:55quite simply just like subsidies right
12:57where you're just giving people a
12:58discount on something be buying anyway
12:59and turns out they want a lot of that it
13:00just turns out that can't be a business
13:02because it's just going to burn cash uh
13:04forever and so um yeah there is this
13:06related thing which is yeah what what
13:08will will people pay for it and how much
13:10will they pay for it um and of course
13:11that you know price pricing pricing as
13:14it turns out is its own science and its
13:15own art um and probably you know a whole
13:17could be topic of like many shows just
13:19just uh in in and of itself but but
13:21always worth thinking about even if
13:22you're not starting out charging all
13:24right the next question is from Kong
13:26famam and uh Kong fam's question is what
13:30are your thoughts on the Creator economy
13:33blending into startup territory examples
13:37festivals Lo Logan Paul with prime will
13:40it become the norm or is it just a fat
13:42yeah so I'll take a whack at it so um uh
13:47so this is basically I think that if I
13:48have the question right this is
13:49basically the phenomenon of sort of
13:51branded creators individuals uh people
13:53with individual Brands um who are kind
13:54of big you know either you know existing
13:56celebrities like movie stars or new
13:57celebrities like you know YouTube stars
13:59like Mr Beast um and then they basically
14:02have consumer product lines um and so Mr
14:04Beast you know Feasta BS is a is a food
14:06you know it's cookies and and candy bars
14:07and so forth um and then I think uh
14:10Logan Paul's I think is I think it's an
14:11energy drink if I if I recall properly
14:13yeah I think yeah yeah so there's a
14:15bunch of these look you know Hollywood
14:16is kind of very focused on this right
14:18now and so there's you know famously
14:19Ryan Reynolds has I think his his
14:21Aviation Jin uh George Clooney has a
14:23tequila label you know these um yeah
14:26that's in very well Cas Amigos cigos
14:29yeah so um yeah so there there's a big
14:31there's a big thing a big kind of thing
14:33on this um so you know there's two ways
14:35of looking at this I think one is you
14:37know these well by the way another one
14:39we should give you know huge credit for
14:40is Kim kashian with skims um oh yeah
14:43yeah that's right which is like a huge
14:45apparently like a very everything I'm at
14:46least hearing is a very big business now
14:48um and so there's kind of two ways of
14:51looking at this you know the historical
14:52way of looking at this I think would be
14:53sort of these are gimmicks um you know
14:56which is it's like you know it's like
14:57okay fans of somebody are going to buy
14:58the thing they recommend you know kind
14:59of for a while um but at the end of the
15:02day you know mainly you know these
15:03things are not you know most consumer
15:05markets are not this most you know most
15:07when you go to the grocery store most of
15:08the products on the Shelf are not this
15:09kind of thing uh but if you go to the
15:11clothing store most of the products are
15:12not this mostly it's you know most
15:14consumer products are provided by these
15:16you know giant branded product
15:17conglomerates like proor and gamble or
15:19Unilever craft Foods or you know Ral
15:22Lauren or you know what you know
15:23whatever right um uh and so that you
15:26know the C traditional view would be
15:27these are kind of gimmick sesh to kind
15:29of the mainline consumer products
15:30businesses um I think there's a more
15:32aggressive argument you could be made
15:34which is kind of where I am which is I
15:35think maybe the sort of influencer
15:37Creator branded kind of individually
15:39branded things I think this might be the
15:40future of consumer products generally um
15:43and the the the theory on that would be
15:45that you know why are there product
15:47Brands like why is it craft macran
15:49cheese and not either not and by the way
15:52not either generic macaroni and cheese
15:54um right or not you know whatever you
15:56know Jack Black macaroni and cheese
15:57right yeah um or whatever I just watched
16:00Tropic Thunder again so's he's on my
16:02um that's a great movie by the way I was
16:04trying to think of somebody who might
16:05Ador son mac and cheese I think it would
16:07be banned today though it could they
16:09make that movie today it is
16:10inconceivable um that they could they
16:13that movie holds up incredibly well and
16:16inconceivable um that made today but
16:18anyway so Jack Black was is a genius in
16:20it so it's on my mind but um so uh so
16:24why do these like why do these sort of
16:25consumer why do these nonpersonal
16:29nonhuman right like you know craft or
16:32whatever whatever you know tide or you
16:34know take your pick of any of these
16:35things uh you know Johnny Walker you
16:37know Scotch whatever you know why do
16:39these sort of nonperson non you know
16:41celebrity non Creator non influencer
16:42Brands exist and the reason for the
16:44reason I think it's because of the media
16:46of the era in which those Brands were
16:47created um oh right right right right
16:50centralized media leads to to Brands as
16:53opposed to people as opposed to people
16:55and so right exactly and so this is like
16:57you know in the Mad Men era right from
16:59like you know basically the the sort of
17:00mass mass media era call it from the
17:021930s to like the 1980s or something
17:04with radio and television you know
17:06basically the thing that you could do to
17:08build a brand was you could run TV you
17:10could hire Don Draper and you could run
17:11TV commercials um and you just want a TV
17:14commercial was like a 30 second mini
17:16drama mini comedy and you had a chance
17:17to get one point across and you weren't
17:19trying for the most part to Market a
17:20person you were trying to Market a
17:21product um and so you just had this like
17:24Super Hyper distilled like basically
17:25single shot uh to be able to kind of get
17:27you know CA established like or whatever
17:30it was um and you had celebrities in
17:32those days but like you they they
17:34weren't like front and center in this
17:35effort because just like you were just
17:37trying to get the basic me message of
17:38the of the product out for the for the
17:40most part um and so but but then that
17:42leads to this kind of very unnatural
17:43configuration where now you have
17:45individual consumers who like have a
17:47relationship with a brand right like
17:50like a relationship with a
17:52corporation with Bud Light with Bud
17:54Light yeah exactly like my relationship
17:55with Bud Light like why why do I have a
17:57relationship with a made up name for
17:59a Fortune 500 company like you know it's
18:02like okay if that's all I can have then
18:04okay fine but like really like that's my
18:06emotional Affinity like that's just how
18:07I'm going to kind of process things uh
18:09if if in the alternate I now have a
18:11different kind of relationship I can
18:13have and these are this is what they
18:14refer to as a psychologist call a
18:16parasocial relationship if I could
18:17actually know instead no like actually I
18:19have a relationship with you know I love
18:20Ry Reynolds movies or you I love
18:22Deadpool or I love Kim Kardash you know
18:24reality TV or I love you know whoever
18:26you know what whatever that is or I love
18:28Mr you know my my my kid loves Mr Beast
18:31right like he's like you know like
18:32number one role model in the world right
18:35um and so but yeah you know you can
18:37still say look is it a real relationship
18:38if it's not a two-e relationship but you
18:40could also say look it's like a person
18:42right um it's a relationship with a
18:44person and so maybe in the future you
18:46know maybe we're at the beginning of
18:47what is a monster wave and we'll you
18:49know be sitting here 20 years from now
18:50and it will turn out this was basically
18:52the great transition and in the future
18:53the brands will actually all be
18:54individually LED uh Ben what do you uh
18:57what do you think of that
18:58you know I I think that's probably right
19:00it just has me thinking back probably
19:02the original case of This was um the Air
19:05Jordans where you know uh sneakers were
19:09always Converse or uh Adidas or what
19:13have you and then Michael Jordan signed
19:16this contract with Nike which was kind
19:18of like a distant whatever third in
19:20basketball shoes or maybe fourth and you
19:24know that shoe he he negotiated his
19:28negotiated for the first time ever a
19:31piece of the back end in exchange for
19:33the shoe being called
19:36Jordan um which was you know it was
19:38Michael Jordan I I bought Michael Jordan
19:42and that worked amazingly and it's sort
19:45of you know it's been a slow Evolution
19:48massively accelerated by social media to
19:50get kind of all the way to what people
19:53really wanted which is a relationship
19:55with a person who wears the shoe not a
19:57re with the company that makes the shoot
20:00um and we see this on on social media in
20:03a in a pretty big way right in like
20:06andrees and Horwitz is are firm but like
20:08you and I have way more Twitter
20:10followers than andrees and Horwitz does
20:14because like like what is the
20:16firm you know what is that compared to
20:20you know who who do you want the actual
20:22relationship with is is kind of obvious
20:24I think that the exciting thing to me
20:26you know kind of getting back to the
20:27question question is oh now the
20:30economics are actually changing so
20:32rather than it being you know a Nike
20:34shoe with like a small percentage to
20:37Michael Jordan on the back end or or a
20:39good percentage at whatever it was right
20:41you know it's Tim Kardashian owns skimps
20:44like that's her company right um and so
20:48I think that's a you know quite a
20:50positive evolution in the way Brands
20:52work I think um just because it
20:56should you know the value should go to
20:58the person that I mean this is kind of
21:02true in all the kind of creative world
21:04you'd really like more of the value to
21:05go to the creative and less to
21:07manufacturing and distribution right um
21:10and historically all the money's gone to
21:12manufacturing and distribution and very
21:14little or a tiny amount to the creative
21:16and uh that you know you all kinds of
21:20stories about that but like this seems
21:21like a clearly Better World at least in
21:23my view yeah I think so yeah all right I
21:27was a very good question um next
21:30question is from Gabriel what happens
21:32when the cost of AI goes to zero will
21:36startups Ben benefit more than
21:38corporations this is a very good
21:41question one that we ask ourselves often
21:44yeah so maybe we could start just by
21:45saying look there's like a huge debate
21:47in the industry and in our firm um and
21:49you know other people may have figured
21:50this out but like we we still debate it
21:53um uh is you know does basically like is
21:57like there's one version of the world in
21:59which you have a small number of what we
22:01sometimes call God models um which are
22:03these like super big you know GPT you
22:05know visualized gp8 or something like
22:07that um where it's like this super model
22:10in the sky running on this super you
22:11know millions of processors you know
22:13with billions or hundreds of billions or
22:14trillions of dollars if capex and
22:16training you know investment kind of
22:17behind it um and you know there's like a
22:20monopoly or oligopoly of these companies
22:22and you know but everybody uses the god
22:24models just because they give better
22:26answers on everything um um and you know
22:29that that's like it's sort of the
22:30equivalent of the world of search that
22:31we've been in for the last you know
22:32basically 20 years um you know sort of
22:34an oligopoly thing where the you know
22:36the search companies extract a lot of
22:38the economics in the startup ecosystem
22:40through the advertising um so you know
22:42that's one scenario there's another
22:43scenario where basically the it's sort
22:45of a sort of what that you know
22:46sometimes call a race to the bottom
22:47which is no actually what happens is
22:49intelligence commoditized um and it just
22:51turns out that there are going to be a
22:53set of techniques that you use to you
22:55know get create large language models
22:56and there's going to be a set of
22:57techniques you use Ed to like you know
22:58gather and and uh and and and training
23:01data or there's going to be a set of
23:02techniques you know that you maybe use
23:03to synthesize training data um right and
23:06so maybe you have like the old AI train
23:08the new AI um and then you know in that
23:10scenario it's like a basically a race to
23:12zero uh so the the the cost of AI
23:15basically drops to zero and you know and
23:17the analogy there would be maybe what's
23:18happened in the microchip market right
23:20which is um you know it like what does a
23:22microchip cost you know well a really
23:24good one costs you know a few thousand
23:25dolls but you know there's companies
23:27like r that sell them for pennies um and
23:29you can put a microchip in anything and
23:31it's super cheap and everybody does it
23:33um or the internet or the internet the
23:35internet access is very cheap for
23:37everybody yeah that's right like what
23:38ises right what does it cost to put what
23:39does it cost to put tcpip into something
23:41right um and have it be on the internet
23:43and it turns out the answer is basically
23:44pennies um right um and I mean to the
23:47point where like you know literally I
23:48now have coffee mugs that are internet
23:49connected right like you know you know
23:52like right and so um you know look and
23:55and there's there's like this like
23:56amazing there's all these like amazing
23:58philosophical questions kind of wrapped
23:59in that around like what is the nature
24:00of intelligence and and there's this how
24:02High Is Up question which is how smart
24:03can can things get but um you know and
24:06so forth but um you know it it like you
24:08know push comes to shove my guess is
24:10it's that second scenario my guess
24:12basically is what we're about to see is
24:13something a lot like the microchip or
24:14the internet where you're going to have
24:15actually you know billions of AIS
24:17running around uh sort of with you know
24:19distri like a pyramid you know industry
24:21configuration where you know you'll have
24:23just a you'll have a giant plurality of
24:25AIS that are on the smaller cheaper end
24:27that will be doing all kinds of you know
24:28kind of localized functions um and
24:30you'll be able to you know basically
24:32summon intelligence at some level into
24:34any application you need for you know
24:36basically very cheap or ultimately
24:37basically the same as free um and then
24:40you know you'll have the god models in
24:41the background when you need like the
24:42Super Genius but like for most things
24:44you don't actually need a Super Genius
24:46um and so I I think I I do think it's
24:48actually like a re I have it's go it's a
24:51reasonable expectation um that basically
24:53intelligence is going to be something
24:54that's going to get built into
24:55everything um and then why don't you
24:58speculate from there on then what that
24:59means for like the economics yeah you
25:01know it's interesting I think it's
25:04um technological question in that right
25:09what is the ASM toote like how like when
25:12things that get to what point of
25:13intelligence does the um technology just
25:17commoditize and be available to everyone
25:19and then you know and then kind of
25:22around the edges is it like one
25:24intelligence or is it many many
25:25different kinds of and different models
25:27like right now we have very different
25:29models for generating images than we do
25:31for uh language and and you know other
25:35things so you know does that stay
25:37fragmented or does it consolidate around
25:39a single architecture I think those are
25:41important technological questions but
25:43then there's an industry structure
25:45question um which you know I think is
25:48very important which is you know is open
25:51source legal um does you know does uh
25:55crypto and blockchain help decentralize
25:57the data question um are there kind of
26:01forces what I would call forces for good
26:03forces for Humanity um that can
26:06counteract the forces for giant
26:08corporations being like our literal Tech
26:12overlords you know as they are on you
26:14know some of the current Internet
26:15services and uh and I think that you
26:18know that's probably the most important
26:19policy question that people are getting
26:22caught up in this other question of uh
26:25you know safety and alignment the more
26:28important question I think is probably
26:30open source and decentralization yeah so
26:33a cynic a cynic would say a synic would
26:36say that the a number one reason why a
26:38big company would be lobbying
26:40aggressively to try to get basically
26:41government regulatory protection um is
26:43if internally they are convinced that
26:45it's going to be rised to the
26:46bottom yep yep cynically they would
26:49cynically cynic would say that right
26:51because because if you knew you had it
26:52right if you knew you were headed to a
26:54Google you know sort of a search
26:55industry outcome thing where you were
26:56going to have like an Monopoly or lopoly
26:58you would need the government protection
27:00but if you knew over time that it was
27:01going to commoditize and everything was
27:02going to go to zero then you'd be racing
27:03to try to get a regulatory barrier
27:05constructed right now to prevent You'
27:06run to the government to to preserve
27:08your Monopoly right that's
27:11right yep um okay so this one I think is
27:16more appropriate for you which is how
27:18will generative AI change public
27:23education this is a long well we need
27:25public education it's a long and
27:27complicated topic um well so public I
27:30think public education right is I'll try
27:32not to I can spend the next six hours
27:33just on this topic but um uh banging my
27:35shoe on the table but um I mean look the
27:37thing is like education in our society
27:39is as mult is an overloaded concept
27:40right it has it has multiple it plays
27:42multiple roles right um and so there's a
27:46you know at the base level there's just
27:47like a daycare role right which is just
27:49the place where kids go so the parents
27:50can go to work um you know there's like
27:52a socialization role it's the place
27:54where kids learn how to you know deal
27:55with other kids For Better or For Worse
27:57um um uh you know there's there's a uh
28:00there's a there's a learning component
28:02to it um I I am told allegedly um
28:04although evidence on that is actually
28:05hard to come by um so Brian Kaplan wrote
28:08a great book on that um uh uh and then
28:11there's um you know sort of uh
28:14certification um you know kind of aspect
28:16which is like basically can you know
28:18have you have you demonstrated a track
28:19record of you know sort of sustained
28:21effort such that somebody would want to
28:23hire you um right um and there's there's
28:26there's an enormous like what's called
28:27sheep skin premium in the in the in the
28:28employment markets which is um somebody
28:30who's a semester away from graduating
28:32high school or college does not make you
28:34know 78s of what a graduate makes you
28:36know um he makes half of what a graduate
28:38makes right and so you know there's this
28:40is clearly the bar thing I graduated
28:41high school I graduated college which
28:43which seems to count a lot for employers
28:45um and then I would say fifth there's
28:46like a social ethical moral patriotic
28:50you know there's like a programming you
28:52know there's the in cult oh right right
28:54right right an indoctrination compon
28:56indoctrination comp comp the M component
28:58and you know when I was in school it was
29:00you know basically like you know there
29:02was probably more Patriot you know
29:03probably more like you know America and
29:04like here South American government
29:05works here's why democracy is good
29:07Pledge of Allegiance Pledge of
29:08Allegiance like all that stuff and then
29:10you know more recently maybe they're
29:11focused on other things but you know
29:12these days they would you know use use
29:14other have other Concepts other terms um
29:16but it's this kind of like and you know
29:18and again without without applying a
29:19value judgment here it's like basically
29:21what how yeah basically what how do we
29:22how are we acculturating how are we
29:24basically training up uh uh you know
29:27um of our country um to basically be be
29:31good citizens for whatever the
29:32definition is at that time of what it
29:33means to be a a good citizen and so that
29:36it's like this bundled set and so it's
29:38the the thing with the question of like
29:39how's AI going to affect it is it's
29:41because the answer you'd like to give is
29:42oh AI is going to make it so much easier
29:43for everybody to learn things but it but
29:45it actually the more humble way to
29:47approach the question is like it turns
29:48out learning things is actually a like
29:50not necessarily that large of a
29:52bundle yeah right um and so you know is
29:56before that reason that you would
29:57probably say wow it's unlikely that
29:59there's like some big fundamental
30:00restructuring of Education because even
30:02if you're even if AI is like a dramatic
30:04change in how people learn things like
30:06even if everybody in the future has an
30:07AI tutor and they don't need to wait for
30:08their teacher to explain things anymore
30:10like that still doesn't that doesn't
30:12affect that doesn't necessarily affect
30:13the other parts of of uh uh of the
30:16bundle uh now yeah I think what you
30:18could say is like look on the margin
30:20like you know this should be good for
30:22students and that all of a sudden you
30:23know and I you know you see this with
30:24kids today all of a sudden every kid is
30:25going to have a tutor who knows
30:27basically everything about everything
30:28and then has infinite patience um you
30:30know to be able to um explain things
30:32I'll just give you for people who
30:33haven't tried this it's worth trying you
30:35know do something like the following is
30:37go to chat GPT and basically say um do
30:39the following queries um explain you
30:42know quantum mechanics to me like I'm a
30:44PhD in physics explain quantum mechanics
30:47to me like I'm an English major in
30:48college explain quantum mechanics to me
30:50like I'm in e8th grade explain quatum
30:52mechanics to me like I'm in third
30:54grade explain quatum mechanics to me is
30:57if I'm a toddler and and what you find
30:58is it actually can do all of those
31:00things right um and so um it can it can
31:04it can it's really really it's exactly
31:06the kind of thing that it's really
31:07really good at and so it it it it has
31:09the ability to meet you wherever you are
31:10in your Learning Journey right uh and
31:12there's going to you know there's KH
31:13Academy and others are working on this
31:14right now to appli to education there's
31:16going to be like amazing educational
31:17online tutoring and and learning you
31:19know kind of things that come from that
31:20you know so students are like at the
31:22very least students are going to have
31:23that a sort of a release valve if they
31:24find themselves in a setting where they
31:25don't understand what's happening or the
31:26teacher not good or the teacher is
31:28always distracted or whatever um so yeah
31:31I think that's my general answer on
31:32public education which is probably
31:33changes less than less than I would hope
31:35but look there's the other thing though
31:37which is like look there there there
31:38always are a small set of students who
31:40are like really um high potential um and
31:43you know in the old days you'd have a
31:45gift and intelligent program or you'd
31:46have AP courses you know the most
31:48Progressive school systems in the
31:49country right now are eliminating all
31:50that stuff for Equity reasons right um
31:53you know they're canceling calculus and
31:55algebra and AP courses like because of
31:57the disperate impact differential
31:58outcomes um and so um you know like the
32:03best and brightest kids like are going
32:05to need a different approach um and
32:07they're going to need a a different
32:09track uh and by the way historically the
32:11best and brightest kids have always had
32:12a different track there's always been
32:13some other track like uh the like
32:15members of the historical aristocracies
32:17didn't go to schools they they were
32:19tutored by by like like tutors right
32:21like Alexander the Great was tutored by
32:23Aristotle uh right um and so what's that
32:27good tutor yeah exactly well Nero Nero
32:29was tutored by senica um that one worked
32:31less well um but um the Alexander the
32:35Great one worked out so so um so uh you
32:39know look you know there are going to be
32:41some kids um you know who are like super
32:43high potential um who are in the middle
32:44of nowhere and don't have access to like
32:46a fancy prep school or whatever um who
32:48are going to be able to basically learn
32:50you know college level material starting
32:52when they're you know 10 or 12 or
32:53whatever and be able to like just kind
32:55of Raceway ahead and teach themselves
32:56that that's never going to be the vast
32:58majority of kids I don't think but like
32:59for the like super high potential kids
33:01like it may be that the outcomes their
33:04own personal education outcomes are like
33:07much much better uh on the other side of
33:09this yeah very interesting um and do you
33:15uh that AI could lead to the either
33:19unbundling or restructuring of the
33:21education bundle or that's less likely
33:23it could the the problem is like you
33:25know just take public education in the
33:26US like you know public C through 12
33:28like it's not it's not run primarily for
33:30the students right like it there it's
33:32the teachers you know mod I don't think
33:34it's controvers a modern C systems run
33:36primarily for the teachers and
33:37specifically for the teacher unions
33:39which are tremendously politically
33:40powerful you know to to the point where
33:42like you know School District you know
33:43famously like the New York City School
33:45District can't fire Bad Teachers right
33:46like they they they finally came up with
33:48this what they call the rubber room or
33:49whatever it is where they and they they
33:51shut down charter schools and all that
33:52kind of yeah absolutely yeah the
33:53teachers unions Mountain these war is on
33:55Charter Schools like they're just
33:56absolutely terrified of having
33:57competition and so there there's this
33:58there's this there's you know for better
34:00for worse whatever there's this there's
34:01this other driver out there um and so
34:04and then look like the daycare thing I I
34:06think is actually like you know
34:08especially you know you know if the if
34:10the sort of modern economy is going to
34:12be you know to paired incomes you know
34:13which is kind of what's happened over
34:14time um those are those are those are
34:17like you know those are real world
34:18problems that are not you know that are
34:20not directly addressed by technology so
34:22that that's why I'm probably I'm
34:24probably on the bearer side of like some
34:25sort of fundamental change there but
34:27um okay so Promethean
34:30cave asks how soon before AI advancement
34:35allows all people to access their
34:37greatest untapped creativity you know
34:39that is certainly kind of it's a great
34:42question and that that's going to be
34:43probably one of the big outputs of uh of
34:47AI is that um you know anybody will be
34:51able to create in many ways so if you
34:53have an idea I mean like the there's
34:55kind of this uh David Boe had this very
34:58interesting quote I watched a doc on him
35:01recently and he said I'm not very into
35:03virtuosity um meaning like I don't ever
35:07want to be like the greatest guitar
35:08player in the world that's not at all
35:09interesting to me I'm interested in
35:12creativity um and so AI kind of I think
35:15gives you in all kinds of Creative
35:18Dimensions the ability to tap a virtuoso
35:22to kind of express your idea um and that
35:24could be in you know music or writing or
35:27or or anything really um and so I I do
35:31think that that is um coming and
35:34probably in the relatively short term I
35:37would think in the next um several years
35:39where anybody with a very interesting
35:42creative ideas will be able to basically
35:46fantastic movie or song or kind of piece
35:50of art um and that's going to be uh you
35:54know that'll be a great transformation
35:58yeah you see there's a bunch of memes
35:59now um you know just with like DOL and
36:01mid Journey for image generation you
36:03know the the ability to use like
36:04basically art 2D art is like expressive
36:07humor um so the the meme in the last
36:09week that's taken off is um it turns out
36:11you can ask uh you can ask the AI image
36:13generators to basically do sort of like
36:16basically be more dramatic or more
36:19humorous in a certain direction and so
36:20there been a lot of examples of this but
36:21one was just like there's a guy sitting
36:22at a his laptop and and a computer and
36:25the thing is like have him get have
36:26image have him get like more and more
36:28excited about the potential for like
36:29civilian nuclear power um right and and
36:32then basically there's this little
36:33sequence of images or you could have you
36:35know for that matter you could have like
36:36somebody like you could have a
36:37bodybuilder and you know every every um
36:39you know have him you get 10% you know
36:40bigger and stronger each time until he's
36:42finally like the incredible hul whatever
36:44there's a ton of people have had a ton
36:45of funny ideas around this and and you
36:47can actually now you can actually
36:48generate you know you they actually
36:49generate the art which just like looks
36:51absolutely spectacular and it's like
36:52very entertaining you know for people
36:54with actually no visual skills um so
36:56that that's just this week's example but
36:57um but yeah look like what you're
36:59describing is already happening with
37:00visual art you know it's already
37:01happening with with uh with 2D image
37:03generation you know the 3D stuff is
37:05coming really fast now um and then um
37:08you know it started obviously text you
37:09know there's a lot of people now able to
37:11generate poetry uh with ch GPT who were
37:13certainly not able to before um yeah
37:16right much better tool than a thesaurus
37:18exactly yeah so it's yeah I I don't know
37:20you know you never know like there
37:21there's there's a very big there's a
37:24very big underlying question to all this
37:25right which gets into which is when I
37:27find myself talking to people like in
37:28the entertainment industry about which
37:29is because people in the entertainment
37:31industry like creators professional
37:33creators of like art or movies or books
37:34or whatever I'm very quick to say well
37:36of course hey I can never be creative
37:37like a person is and on the one hand I'm
37:40like yeah of course that's the case
37:41because like there are certain
37:42particular creative super Geniuses who
37:43clearly are doing something that like I
37:45can't do right I mean just take your
37:46what whoever your favorite musician
37:48David Bowie like I you know there's no
37:49way that you or I are ever going to be
37:50able to do what he was able to do and so
37:52there's clearly something really special
37:53about the best creative Geniuses in a
37:55field and I would not be so presump this
37:56the same technology can just quote
37:58unquote do that look on the other hand
38:00it's like okay what exactly is human
38:01creativity right um and like how are
38:04humans creative and how much of
38:06creativity is based on actually learning
38:07a lot about all the people who came
38:08before you um and then figuring out to
38:10kind of mix and mat you know mix and
38:12match and you know create mashups and
38:14and and synthesize new versions of the
38:16things that came before you um and you
38:19know how much of that can the machine do
38:20and so and then it's like okay maybe you
38:22can't be as good a musician as David
38:24Bowie but like most m music in history
38:27have never been as good as David Bowie
38:28maybe it's like 99% of the of the field
38:31is something that I can do yeah and so
38:34there's all these like really
38:35interesting kind of philosophical
38:36questions around the nature creativity
38:37which I think we're going to is going to
38:39be a very good re reason to think about
38:42those yeah yeah yeah we're gonna find
38:44out that's for sure that's the exciting
38:46part yes okay ferit I think that's how
38:50you pronounce it uh asks are data
38:54centers and other infrastructure
38:56structure in a position today to
38:58accommodate the growth of AI
39:00applications um I can start with that so
39:03I think that the that there's two really
39:06kind of important um things lacking in
39:09most modern data centers today uh one is
39:12power interestingly so the concentration
39:16power needed to kind of run like the to
39:21to basically fulfill the demand of AI in
39:24any given data center is just
39:27um by a fair amount and so there's a lot
39:29of work going into you know everything
39:32from you know portable nuclear power
39:35generation to like hold different power
39:38architectures and so forth so that's a
39:39big thing and I've heard that from all
39:41the kind of large um data center
39:45providers you know power is probably the
39:47first thing they're going to run out of
39:49the other one that that's come up and
39:51it's more in the kind of chip
39:53conversation so if you look at gpus you
39:56know Nvidia is clearly the leader but
39:58like one of their biggest advantages is
40:01they have a you know Nvidia also owns a
40:04networking architecture called
40:05infiniband and infiniband turns out to
40:08be like really really
40:10important um in AI uh you know in in in
40:14terms of having the kind of gpus being
40:16able to communicate at the at the kind
40:18of at an appropriate um with an
40:21appropriate latency and speed and so
40:23forth uh and that is not available you
40:29know with chips from providers like you
40:31know AMD doesn't have infinite bat and
40:34so forth so the people who are able to
40:36kind of best exploit the other chip chip
40:39architectures that are competitive with
40:41the Nvidia chip architecture are ones
40:45amazing uh data center networking
40:49technology and so that's companies like
40:51you know a Google or a meta or so forth
40:53but so that that's lacking in like your
40:56run of the mill data center for sure and
40:58uh so those are probably the two big
41:01things that that at least we talk about
41:03internally quite a bit is networking and
41:05power yeah Ben like what like what's
41:08your sense of like are are the data
41:11centers and you know corresponding
41:13components uh chips and network
41:15interconnects and so forth are are they
41:17under like do we have half of what the
41:19industry would be using right now if it
41:21could or a tenth or like what's the
41:24what's the mismatch right now I think
41:25it's part right because so the the first
41:27bottleneck is just can I get the gpus
41:31and so I think we don't know yet um but
41:33I think that is so basically what every
41:36everybody says is okay once we get the
41:38gpus then the next issue if you're not
41:40using Nvidia is going to be networking
41:42and then the next issue if you are using
41:44anything is going to be power so that's
41:47that's a little bit how I think of it so
41:48I don't think there's a great estimate
41:51um you know what we could be doing if
41:54those things were fixed because we're
41:55stuck on like we just don't have the
41:57gpus right now and then there's an
41:59implication to this I think which I
42:00think is actually a fairly bullish
42:02implication to it the implication is
42:03basically the AI tools and their
42:06capability that you experience today as
42:08a user is a fraction of what is actually
42:12possible technologically today in other
42:14words like the the bottleneck today for
42:16your experience using a lot of these
42:18systems is not the algorithms or the
42:19data the bottleneck ises quite literally
42:21the hardware and the power underneath it
42:23um and so and that's that that would be
42:25a bullish a of the state of AI as a
42:27field because basically says as these as
42:29more chips and data centers of Power are
42:31brought online all of a sudden these
42:33systems are just going to get much
42:34better right without any algorithmic
42:37changes I think that's that's certainly
42:38true yeah like there's there's basically
42:40improve there are functional
42:41improvements in the bag um uh even
42:43without additional technology
42:44breakthroughs um and then look we see
42:46this every day in startup world there's
42:47a lot of startups that just like quite
42:48literally can't launch like they can't
42:49like actually build their thing yet
42:51because they can't get these you
42:52physical components um but but but the
42:55the the nature of the chip industry and
42:57the nature of things like you know power
42:58data centers is this you know it's
42:59mostly money um and now there's a you
43:02motivation for a lot of people to invest
43:03a lot of money and resolving these
43:04shortages and so I I I I I would expect
43:06in the next few years there will be this
43:08period of time where all of a sudden
43:09it'll be like Christmas just for the
43:11industry it'll be like oh all of a
43:12sudden we all have all the you know sort
43:14of equipment power we need and then and
43:16then that's when we're really going to
43:17see even what these systems can do yeah
43:22um I I think that I think that's
43:28then it's hard to even imagine you know
43:30when when you get all the kind of gpus
43:33power networking on the one side and
43:35then you also have the algorithmic
43:37breakthroughs on the other side it's
43:38going to be yeah right super exciting
43:41time one of the most exciting times in
43:44technology this is a great I'm GNA go
43:46right to this one Zach asks could AI
43:49replace governments in the
43:58which which you want to bite that one
43:59off first it's very provocative well I
44:02mean I think it kind of gets to the
44:03question is like can AI replace kind of
44:12think governments is a is an interesting
44:15one because it's kind of like can AI
44:17govern people um or or or can people
44:21govern AI is also another interesting
44:23question well if I could maybe I could
44:24break the question though into two parts
44:25which is there there's part of
44:26government which basically is like the
44:28decisions being made by high officials
44:30um but there's a separate part of
44:31government which is just administration
44:34bureaucracy yeah well I think you so
44:37here's the okay so on that one um it's a
44:40little frustrating that AI has not
44:42replaced government on the bureaucracy
44:44side so like I'll just give you my
44:46favorite example which is you know
44:49recently uh you know we had this big
44:52ridiculous political debate about
44:54whether or not we should hire 8 ,000 new
44:57IRS agents um and the other way to solve
45:01that problem would be to basically hire
45:05six good you know AI engineers and
45:09basically be able to get perfect
45:10auditing of everybody's uh tax result
45:13and you know maybe even maybe even tell
45:17people what they owed as opposed to
45:19making them guess and then arresting
45:21them if they guess wrong uh so yeah I
45:24mean I think there's no question that we
45:26could improve the bureaucracy with AI
45:29yeah because it is weird like the
45:30concept of the government the concept
45:31that the Secretary of State and the
45:33person you know at the behind the
45:34counter at the DMV are in the same you
45:37know yes they both government is a
45:39little bit of a a little bit of a let's
45:42there will come a time when they'll be
45:43view as a historical audity um yeah yeah
45:47and look like I would just make to make
45:48a more more General observation which is
45:50just like look if you look at you can
45:51look at like measures of what you just
45:53might call administ like administrative
45:56um right which is basically well I take
45:58I'll pick on universities right so the
46:00the highend universities places like
46:02Stanford and so forth they now have more
46:04administrators than they have
46:06students right and so I don't stord I
46:08for the exact numbers it's something
46:09like 12,000 students and like 13,000
46:11administrators right and and if you
46:13chart if you chart the growth rate of
46:15students and administrators the student
46:16body has grown somewhat and the
46:17administrator growth rate has just been
46:19off the charts like it's they never had
46:21this many in the past like it is this is
46:23unprecedented and the same thing is true
46:25of like and yet they've added zero
46:27capacity for students exactly right um
46:30and so and so the running joke is of
46:32course you know Stanford can now afford
46:33to you Stanford with the Staffing model
46:35could switch and just have a personal
46:36coner for each student
46:39yeah the way to organize University um
46:42but um and that's just I I pick on them
46:45as a local local example but like that
46:46that's true of many many bureaucracies
46:48like so for example that's the story of
46:50a lot of the increase in the cost of
46:51healthcare uh over the last 20 years A
46:53lot of it is just the same thing has
46:55happened in hospital
46:56right it's not doctors and nurses it's
46:58administrators it's administrators
46:59exactly and so and and you know you
47:01could look at that a couple different
47:02ways you could look at that as just
47:03saying look I don't know the world you
47:04know I don't know the world the economy
47:05gets more mature and you just want
47:07things to run better and you're just
47:08you're going to have a lot more
47:09paperwork and it's just going to be the
47:11way it is um you know and then the other
47:12way you can look at it is well you know
47:14no the the administrators are a class
47:16you know you could you can apply like
47:17Marxist analysis you can say that the
47:19the administrators are a class in and of
47:21themselves they have a class interest uh
47:24right the class interest is in the
47:25propagation of administrative jobs um
47:27right and so you know they're like coat
47:29hangers in the closet they reproduce in
47:30the middle of the night like and you're
47:31just going to like left unchecked you're
47:34basically going to have the entire
47:35economy the entire world someday
47:36everybody's just going to have
47:37administrative jobs uh or you know what
47:39pejoratively sometimes call email jobs
47:42um and you know look like email jobs are
47:44great for the people who have them it's
47:45always less clear that they're great for
47:47the customers um right because of just
47:49this enormous cost blo that kind of
47:51comes with it um and so you know look I
47:54I well like I know for sure like a a lot
47:55of you know for-profit corporations are
47:57already looking at this you know you
47:59know increasingly aggressively which is
48:00like okay how much of this
48:02administrative blo you know how much you
48:03know like even we can reverse even if we
48:05could use just reverse the
48:06administrative bloat back to 10 years
48:08ago right there would be the opportunity
48:10for like pretty big change in both
48:11Staffing models and then in in you know
48:13you could literally you know you could
48:14drop the cost of a lot of products and
48:15services if you could take it back 20 or
48:1730 or 40 years even better and if you
48:19could rethink the whole thing from
48:20scratch and maybe you just don't even
48:21need all these people doing all these
48:23you know administrative jobs because you
48:24have a doing it then all of a sudden you
48:25know you can imagine a future in which a
48:27lot of products or Services the costs
48:29you know become you know a tenth of what
48:30they are today um because you're just
48:32taking out all this overhead um and so
48:34that's a uh you know that's that that
48:36for sure I think is one of the journeys
48:37that we're all going to be on is to try
48:39to is to try to figure that out because
48:40that that because that is that is the
48:42kind of work like can can AI do a better
48:44job than Secretary of State at like
48:45planning geopolitics I don't know um
48:47there are people I'm sure who would say
48:49yes I'm not goingon to you know who
48:51knows um you know can chair Point can
48:53can AI run the IRS better yeah almost
48:55certainly um and so you know why why not
48:58at least give it a try and maybe getting
49:00rid of the bureaucratic state or
49:02replacing it or improving it would be
49:04the key to dealing with our other like
49:06colossal problem which is the deficit
49:09issue which we're going to have to
49:10address at some point or or not or it'll
49:13address us we'll we'll see yes
49:16um Okay Ross Elliot asks uh what are
49:20your thoughts on AI being used to create
49:22new songs by mimicking deceased artists
49:26such as Kirk Cobain or Amy we housee
49:29yeah so you're the guy on this
49:30one yeah so you know that's very
49:33interesting question so I think on you
49:34know like on the one hand there's the
49:37copyright issue where does um you know
49:41where does a money go does it go to the
49:43estate and so forth and like who
49:44controls the copyright and how that
49:46works and I think those like I think
49:48that I think AI needs to be like within
49:50the law on that um in a very kind
49:54of I would say strict way you have to
49:56respect the deceased artist on the other
50:00um you know like it's I think it'd be
50:03pretty neat for an artist to be so kind
50:06of profoundly important that they become
50:08immortalized and you know Amy wiin house
50:12becomes you know the equivalent of a
50:14trumpet or like you know she becomes an
50:16instrument that many people can use into
50:19the future um that would be like an
50:22amazing tribute to her and then also
50:24like pretty exciting for creatives
50:29uh you know I I I kind of hope that we
50:32find a way to do that in a way that's
50:35you know respectful and gives credit
50:37where credit is due I think it'd be very
50:39bad if um you know people just stole her
50:41voice and use it for f so let me ask you
50:44the downside version of the question the
50:45dystopian View and this is this is sort
50:47of this idea that the uh that the or my
50:49friend the Lindy man has Paul scall
50:51which is this idea of stuck culture um
50:54and so like in the obs basically is like
50:56in the movie industry it's like there
50:57are basically no new movie stars you
50:59know past a certain point like right
51:01they would just sit on the old ones yeah
51:03yeah well like Tom K you know Tom Cruz
51:04is always a great example in this right
51:05because like he in the last you know two
51:07years he's had this amaz you know these
51:08two amazing huge movies one in which
51:10he's a fighter pilot and the other in
51:11which he's like a secret agent and like
51:13you know look he's great in both of them
51:14but like he's 60 right like and like you
51:16know like in the old days right you know
51:18you would have had 20 20y olds and 30-y
51:20olds in those roles and now you've got a
51:2160-year old and like he's going to keep
51:22making these movies as long as people go
51:23to them and so it's like and roll it's
51:25like rock and roll even with just you
51:27know like the Rolling Stone still you
51:28know you can go see MC Jagger in age 80
51:30right um you know is that like
51:32substituting for like going to see
51:34somebody you know like M MC Jagger when
51:36he was 20 was not competing with 80y
51:38olds right um but the next new rockar is
51:42competing with with with 8-year-old MC
51:43Jagger so anyway so the the the the
51:45dystopian kind of question would be
51:46would would version of the question
51:48would just be like look if you can
51:49basically if you can literally like re
51:51you know you see this today with like
51:52the Michael Jackson you know hologram
51:54concert and the abob you know kind of
51:55recreation concert and so forth Frank
51:58Sinatra like if if you can literally
52:00like you know reanimate through
52:01Holograms and voice synthesis and you
52:03know 3 you know 3D VR experiences and
52:06you know all these things uh visuals
52:08with impeccable you know production
52:10quality and you could have you know
52:11Frank Sinatra and Michael Jackson and
52:13Abba for the next hundred years um with
52:16like enormous you know money and
52:17resources being put at kind of doing new
52:19creativity under their names and with
52:20their faces and likenesses and voices
52:22like are are you actually like is that
52:23the end of creativity and music because
52:25like at that point you're it's just
52:27basically everything becomes a Nostalgia
52:29forever yeah so I don't think so because
52:34um you know like look if you don't have
52:36a new novel way to do something or
52:41um or create art then then sure that's
52:45going to be better than what you do but
52:47hip-hop you know they kind of brought
52:49James Brown back to life um you know
52:53with samples and but the music was all
52:55new and it created all kinds of new
52:58stars and and interestingly they weren't
53:01even the James Brown records that were
53:03hits they were all the stuff that he did
53:05kind of post his Peak um ended up being
53:08the things that that people wanted to
53:10kind of reuse and in the samples so it
53:12was both kind of a rebirth of you know a
53:17lot of the James Brown music that nobody
53:19had ever heard and then it was you know
53:21a whole new class of Music a whole new
53:27think I guess I would say I think that
53:30this idea is kind of independent of the
53:32Dead culture idea which um uh you know
53:36and I I I would be like I would
53:37definitely take a more positive view I
53:39think I think it could be actually kind
53:41of exciting postmodern a new postmodern
53:45uh artistic Endeavor yes awesome all
53:50um oh this is a good we have a techno
53:53Optimist question from Wayne
53:56Alman is the issue of the a growing
53:59wealth Gap in society solved through
54:01techno optimism or is it possible that
54:05well positive sum and this is an
54:06interesting question um be just the way
54:10phrased while positive some
54:12technological innovations increase
54:15quality of life for everyone the wage
54:17Gap continues to grow or does it and
54:20this leads to serious issues later on
54:22question mark yeah so look a couple
54:25preconditions for having this
54:26conversation that are very important so
54:28um you know there's there's there's two
54:29concepts right there sort of uh absolute
54:32uh like uh income growth standard of
54:33living growth um and then there's
54:35relative right um and you know like like
54:39it's possible for technology to both be
54:41making everybody's lives better and yet
54:42also leading to increase inequality like
54:44like both of those are actually possible
54:46um and there's a version of the
54:47arguments that says at that point you
54:48shouldn't be worried about inequality
54:50because if everybody's lives are getting
54:51better then it's everything is better
54:52and you should just like keep everything
54:54better and then there's but you're
54:56dealing with humans so that that's
54:58that's a difficult thing precisely
55:00because you're doing human beings it
55:01turns out relative relative status
55:02matters and people get really mad about
55:04that so um so so that's one but it's
55:07important to think about um so that's
55:09one thing um uh just let's just set that
55:11as like a put put a pin in that because
55:13we'll talk about that um then there's
55:15this question of like okay that the
55:16actual nature of like wealth and the
55:18nature actual nature of sort of of
55:20income and and and the standard of
55:22living that kind of flows from income um
55:24right which is like income is a proxy
55:26for standard of living right income is a
55:27proxy for the things you can buy that
55:28make your life good uh you know for
55:30yourself and your family and your
55:31community so so so there's there so
55:33income is like the input and then
55:35standard of living is like the the
55:37output um the interesting thing there's
55:39one really interesting thing which is um
55:41the the cost of goods and services
55:43really matters um right because if if I
55:46if I have to spend $100 to buy something
55:48um and then um but then there's
55:50something that is as good or better that
55:52cost $10 a year from now like that's the
55:54equivalent to getting a $90 raise right
55:57because like the thing that I had to
55:58spend $100 on now I only have to spend
56:00$10 on now that frees up $90 of spending
56:02power and it's it's equivalent of a $9
56:05raise and so there basic I'm saying is
56:06there's like two ways to actually raise
56:07income and therefore raise Stander
56:09living one is to pay people more money
56:11by the way and I'm in favor of that and
56:13by the way we do a lot of that in our
56:15day jobs um and then but the other thing
56:18to do is to drop the cost of the goods
56:19and services that they're buying um
56:22which from a standard living standpoint
56:23is actually the exact same outcome um
56:25and so a a a a a major force and lever
56:29economic effect that Technology
56:31Innovation has on the economy um is
56:34driving down prices um and so and and we
56:37we could cite many many many examples of
56:38how that's the case either I mean
56:40there's just like really obvious ones
56:41like television sets that you can just
56:42pick on like very easily um you know and
56:44you can just like watch the price
56:45declines over time which are the direct
56:47consequence of technological innovation
56:48the same is true by the way for lots of
56:50it's been true of media and lots and
56:52lots of things that people buy uh over
56:54the last you know 20 30 years and so
56:55there there there's part of this which
56:56basically says you should be a tech
56:58Optimist even if the only outcome from
57:00it um is a systematic way to drive
57:02prices down um and so you know that
57:06that's one side of it and very important
57:07to think about um um and then um and
57:10then there's the other side of it which
57:11is the the question of how wages are set
57:13in the economy and um the the sort of
57:16the folk you know kind of standard
57:19default way that people think about that
57:20is wages are set in this kind of battle
57:23between the employee and the employer
57:25um and there's like this fixed amount of
57:26money right and either the workers get
57:28it or the or the capitalist get it right
57:30or either labor gets it or management
57:31gets it and you know this is kind of the
57:33point of view of every Union right it's
57:34just this it's fighting over this kind
57:36of zero some zero some pi and you know
57:38economists look at this with things like
57:40the they called like the labor to
57:41Capital ratio where they you know the
57:42percentage of of income that you know
57:44for a business that goes to labor versus
57:45capital and so forth um but that's not
57:48actually I mean that's how so this is
57:50how wages get set in Union negotiation
57:52this is not how wages get set in the
57:53economy broadly um the way wages get set
57:55in the economy broadly is as a function
57:57of What's called the marginal
57:58productivity of the worker uh it's a
58:00function of the economic value that the
58:02worker brings to an employer um and the
58:04reason why wages rise is not because
58:06employers are generous and they want to
58:07pay people more money the reason wages
58:09rise is because if you are an employer
58:11and you are underpaying a worker
58:13relative to the amount of economic value
58:14they're bringing to the business they
58:15will quit and they will go to an
58:16employer that will pay them fairly right
58:19U there there's an Arbitrage the other
58:20employer will hire them and pay them
58:22more because it's still worth hiring
58:23them because there was there was there
58:24was you know there was there were there
58:25was a $100 bill on the ground which is
58:27like a worker who was being underpaid
58:28relative to his contribution and so the
58:31the way wages get set is as a functional
58:32marginal productivity and then the Big
58:34Driver for how productive workers are is
58:35the technology that they use right and
58:38so an accountant who has a calculator is
58:40much more productive than accountant who
58:41doesn't an accountant who has a PC is
58:43much more productive than the accountant
58:45who has a calculator and then the
58:46accountant that has an AI is going to be
58:47much more productive than the accountant
58:49who just has a PC and so as technology
58:52gets introduced into Industries one of
58:54the really big things that it does is it
58:55drives up the marginal productivity of
58:57the workers and so technology is not the
58:59enemy of the worker in most of these
59:01cases in most of these cases it's
59:02actually the Ally of the worker in
59:03actually causing the worker's marginal
59:05productivity to rise therefore causing
59:07the the the worker's income to rise and
59:09so this is my argument of the in the in
59:11the Techno Optimus Manifesto which is
59:13these effects of Technology driving up
59:16uh the marginal productivity of the
59:17worker and therefore driving up income
59:19while simultaneously driving down the
59:20prices of goods and services is just
59:23these two factors are so m massive and
59:25increasing quality of life that all of
59:27the other concerns and debates around
59:29you know this and that inequality and
59:30fairness or whatever just kind of wash
59:31out because those those fundamental
59:33mechanisms are so powerful um that they
59:35actually that they actually drive most
59:36of the outcomes and I I would argue that
59:38you know we have 300 years of History to
59:39confirm this now and the reason that
59:41we're sitting here today in you know uh
59:432023 with like you know standards of
59:45living that are unparalleled by any kind
59:47of human precedent uh civilizational
59:49precedent um you know it sort of
59:50evidence that that that these mechanisms
59:52work this sounds too good to be true
59:54people kind of have a very kind of often
59:56visceral negative reaction to Good News
59:58um of this kind but you
01:00:01know like if this resonates with you at
01:00:03all as you're listening to it like this
01:00:05this is this is like standard economics
01:00:06like that this is the the the basically
01:00:09the great moral virtue of of of of free
01:00:11markets which is this is the mechanism
01:00:13by which actually um free markets are a
01:00:15non-zero sum uh non-zero sum game and
01:00:17and and and there actually is a rising
01:00:18tide and and and I think and then
01:00:20therefore more Technology Innovation
01:00:22will make all that work better as
01:00:26all right that is I'm I'm going to leave
01:00:28it at that because that was good um and
01:00:31then final techno Optimist question for
01:00:33us uh how do you assess the future
01:00:36economic and technological growth
01:00:39potential in the Middle East
01:00:40particularly in light of initiatives
01:00:43such as the line in Saudi Arabia Urban
01:00:46developments like Dubai and the progress
01:00:48in countries like Qatar
01:00:51uh so you know it's pretty interesting I
01:00:54think I think that the thing that we're
01:00:55seeing in the Middle East is a very very
01:00:59strong will among um the governments
01:01:03there to kind of want to build towards
01:01:05the future uh and particularly I think
01:01:07led by Saudi Arabia I think they're
01:01:13know could not be more committed to a
01:01:17kind of better more technologically
01:01:19advanced future and um and basically
01:01:23rather than government um being a kind
01:01:27of regulator and protector against what
01:01:30that future might be uh being more of an
01:01:33enabler and a funer of what that future
01:01:35is and so that's where you get things
01:01:37like the line I was just in Riad and
01:01:39there was um a uh there was a big fight
01:01:44uh Tyson Fury was fighting and they had
01:01:47built a whole soccer arena um for that
01:01:51you know that that fight was held in and
01:01:53it was to build the entire thing was
01:01:56three months and you know like that
01:01:58obviously could never happen anywhere in
01:02:00the United States and uh and so I think
01:02:05very you know in my view it's an
01:02:07incredibly positive development because
01:02:09if you've kind of studi the Middle East
01:02:11you know one of the really tough things
01:02:14which you know led to 911 and you know
01:02:16was certainly a big factor um kind of
01:02:19recently you know is a big factor you
01:02:21know in the recent conflict is that you
01:02:25um kind of modern society and I think
01:02:27you know people in in the 911 context
01:02:29referred to it as like the Jetson versus
01:02:31The Flintstones and and this kind of
01:02:33thing um and you know if you've got a
01:02:37big part of the world going into the
01:02:39future and then you know particularly in
01:02:41the Middle East there's a big part of
01:02:43the world that is you know in the pretty
01:02:45far past kind of culturally religiously
01:02:49women's rights wise Etc um you know that
01:02:53that is a natural conflict and so it's
01:02:56incredibly positive to see really the
01:02:59leading Middle Eastern countries like
01:03:01Saudi going you know we're not only
01:03:04joining the future we're leading the
01:03:05future and so I think that that that may
01:03:07be one of the best things that's
01:03:10happening in the entire world right now
01:03:12yeah and then you know look I think you
01:03:14can kind of see this you can kind of
01:03:15take a historical lens on this too which
01:03:17is basically like look we we've been you
01:03:19know we we in the we in the west and
01:03:20around the world have been trying to
01:03:21kind of make kind of modern you know
01:03:22kind of democratic you know kind of
01:03:24industrial capitalism work for the last
01:03:26you know kind of few hundred years and
01:03:27and and one of the sort of patterns that
01:03:29I think is very clear and has been the
01:03:30case in in a lot of places around the
01:03:31world in especially in the last like 30
01:03:33years um is you know basically like
01:03:36there's a lot of latent Talent out there
01:03:38um you know there are a lot of people
01:03:40who like are actually quite smart and
01:03:42have like ideas and and and um you know
01:03:44have the potential to build businesses
01:03:45and the potential to you know play a
01:03:47bigger role in their in their
01:03:48communities and in Industries you know
01:03:50that they've had a chance to in the past
01:03:51um and you don't need necessarily like
01:03:55you don't need to like I don't know
01:03:56become Singapore like or something
01:03:57overnight like you you don't need to you
01:03:59don't need to have the like most
01:04:01amazingly Advanced economy in the world
01:04:03to have people be able to really uh be
01:04:05be much better off like in one step um
01:04:07what what you can do quite literally is
01:04:09just like you know give people and I'm
01:04:10think in particular here like a bunch of
01:04:12Asian countries in the last in the last
01:04:13you 30 40 years even for you know former
01:04:15communist countries um uh you know where
01:04:18basically it's like okay if all of a
01:04:19sudden people can actually like function
01:04:21in a mark you know have a true market
01:04:22economy can actually like bid labor out
01:04:24to you know multiple employers as
01:04:26opposed to everything being owned by the
01:04:27state um if you can if it's actually
01:04:29possible to start new businesses that
01:04:30introduce a competitive Dynamic um if
01:04:33it's possible um by the way policy if
01:04:35it's possible for uh uh uh people to
01:04:37actually be able to change jobs um you
01:04:39know there's even a discrepancy in the
01:04:40US between the states in which you can
01:04:42actually restrict employee employ
01:04:44employees on non-compete agreements and
01:04:45the ones where you can't um and so so so
01:04:49there's and then if people can like use
01:04:50new technology and they're they're
01:04:51actually allowed to use new technology
01:04:52and it's not blocked by the government
01:04:54um right um and then or um uh you know
01:04:58or that they um you know our
01:04:59Technologies introduced earlier into the
01:05:01educational process right or um you know
01:05:03or all of a sudden you can have new
01:05:04kinds of financial services you can have
01:05:05new kinds of you know Capital available
01:05:07you can have lending new kinds of debt
01:05:09instruments available or you can have a
01:05:10stock market in a country that did
01:05:11previously have one like all all all of
01:05:13these things are basically like Steps
01:05:14along the road uh to what you might call
01:05:16sort of modern modern modern capitalism
01:05:19modern market economy and it basically
01:05:20turns out in the data I think basically
01:05:22is every single one of those steps pays
01:05:24right like basically every step of the
01:05:26way there's Improvement on the margin um
01:05:28and then if you take like 30 or 40 or 50
01:05:29of those steps you basically end up like
01:05:31Singapore um you end up just like with
01:05:33this just like amazing what what looks
01:05:34like a miracle uh of like you know of
01:05:37sort of modern you know sort of
01:05:38market-based industrial civilization um
01:05:40and so like you know cont all the bad
01:05:42news in the world it is striking how
01:05:43many countries in the last 30 years have
01:05:45kind of been walking down this path um
01:05:47and I think doing doing very well with
01:05:48it um and then of course you know look
01:05:50if you can move even faster so much the
01:05:52better um and so if you you know if you
01:05:54actually have like fully enlightened
01:05:55leadership that really wants to wants to
01:05:56do this and really you know cares about
01:05:58their people and really wants to their
01:05:59economies to advance like you can do
01:06:01these things much more quickly um it is
01:06:03you know amazing how you knowbody
01:06:05inequality it's amazing how still
01:06:07unequal the world is on a lot of these
01:06:08things um like I said or even even in
01:06:10the US there's a lot of you know
01:06:13different different regions have very
01:06:14different Dynamics in terms of whether
01:06:15people actually have the ability to
01:06:16reach their full potential um but you
01:06:19know you don't need Perfection you just
01:06:21need to like take the buot you just need
01:06:24boot off the throat like if yeah I'm
01:06:26just thinking like the former communist
01:06:28countries like like there you know this
01:06:29is a big part of the story of China over
01:06:31the last 30 years is you just need to
01:06:32take like Mouse boot off their throat
01:06:33and all of a sudden it turns out that
01:06:35like China's full of entrepreneurs and
01:06:36then all of a sudden it turns out
01:06:37there's like explosive economic growth
01:06:39um and a dramatic rise in income like it
01:06:41didn't they didn't need like they just
01:06:42need to let people do their thing they
01:06:44didn't need to do anything right right
01:06:45right right yeah right exactly now you
01:06:47know now now they're getting into the
01:06:49more complicated part of it and so
01:06:50there's a bunch of other other issues
01:06:51but like a lot of it was to just like
01:06:53let people actually trade and transact
01:06:55and work um and use technology and bring
01:06:58in ideas from the outside and so forth
01:06:59and so on so there's just I just have
01:07:01this like very fundamentally long run
01:07:03optimistic view about the world which is
01:07:04we we we are still as a as a you know
01:07:06country and as a society and as a
01:07:09civilization and as a planet we're we're
01:07:11still under punching our weight relative
01:07:12to what the people uh on on planet Earth
01:07:14are capable of and I think we we we
01:07:15actually have a path where a lot more
01:07:17people can contribute and and everybody
01:07:18can become much better
01:07:20off yeah yeah yeah no question all right
01:07:24right the next question is a crypto
01:07:28Carl do you see recent crypto scandals
01:07:32parentheses like FTX leading to more
01:07:35government regulations in the
01:07:36cryptocurrency space yeah uh I will
01:07:39start with that um so it turns out that
01:07:43in uh crypto um probably the best thing
01:07:48that can happen to the space is uh
01:07:51actual regulation um as opposed to
01:07:54what's been going on which is sort of
01:07:56Rogue uh government agency enforcement
01:08:00um and non-enforcement of various things
01:08:03so crypto is you know I'd say one of the
01:08:07most important if not the most kind of
01:08:09important technological
01:08:11breakthrough um that will lead us to you
01:08:14know hopefully a a fair better more
01:08:19productive more entrepreneurial more
01:08:20techno optimistic Society in that it
01:08:24decentralizes um these very uh powerful
01:08:29monopolies um in a way that kind of
01:08:33allows the people who are contributing
01:08:35value to kind of uh be rewarded
01:08:38economically for their contribution and
01:08:39this is in all kinds of areas such as
01:08:42you know in the creative field right now
01:08:45like your creative work um that you do
01:08:49uh you know the take rate from the the
01:08:50big social media companies or from
01:08:52Google search is you know they take all
01:08:54the money the take rate is everything so
01:08:57you create you know the most uh amazing
01:09:00content and it lives on their platform
01:09:02and they get all the money or you know
01:09:04whatever 99% of the money whatever it is
01:09:06it's a giant amount um and so crypto is
01:09:08a kind of a way to
01:09:10really change that for the much better
01:09:14uh you know where most of the take rate
01:09:15goes to the crater and not to the not to
01:09:18the distribution kind of arm um so very
01:09:22very important now the problem with
01:09:24crypto or a challenge with crypto is
01:09:26that uh because of the nature of it you
01:09:29know you can use it to build these
01:09:31amazing services that provide like you
01:09:35know True Rewards to all of the you know
01:09:39people who are customers and who are
01:09:41stakeholders and who are like everybody
01:09:44participates in the network gets
01:09:46rewarded um but you can also use it to
01:09:49basically just create a casino um and
01:09:52you know a uh a place to gamble which is
01:09:54what Sam bankman free did he created a
01:09:56giant casino and uh you know and then on
01:09:59top of that committed fraud uh so you
01:10:03know but and the way he was able to do
01:10:05it and get away with it is that the
01:10:08regulation of this brand new technology
01:10:10which needs you know specific regulation
01:10:13to Outlaw this casino behavior that he
01:10:16had in creating this you know this coin
01:10:20this FDX coin um that ended up that
01:10:23valued it like dramatically higher than
01:10:25it was really should have been valued in
01:10:27all these kind of behaviors that he did
01:10:30um there's a very easy regulatory answer
01:10:33to that um that we don't have and so I
01:10:36think that we need high quality
01:10:39regulation at the space and then we need
01:10:42The Regulators to stop you know behaving
01:10:45like in this weird random way where you
01:10:49know they're treating uh kind of things
01:10:52they're treating everything very very
01:10:53genetically generically you know based
01:10:55on kind of regulatory uh ideas from the
01:11:021930s which is you know not really
01:11:05relevant for this kind of brand new
01:11:07technology and so you know I think we're
01:11:10hopeful that kind of FDX kind of alerts
01:11:14the you know the the the really good
01:11:16policy makers which it has and you know
01:11:18people on both sides of the aisle uh to
01:11:21kind of put in place sensible uh regul
01:11:23for crypto which will kind of unlock
01:11:26this uh you get really a much better
01:11:28world for all of us um so anyway um
01:11:33hopefully this you know turns out to be
01:11:35the the bad thing that turns into the
01:11:36good thing all right uh next question um
01:11:41uh how have your views on crypto changed
01:11:43over the past five years and I would say
01:11:46uh so I'll start with that one too um I
01:11:49don't think that they've changed much
01:11:51other than uh kind of
01:11:54both the importance of you know kind of
01:11:58proper regulation uh I think at least in
01:12:02my view is dramatically more clear and
01:12:05then the kind of danger of of of not
01:12:08having it and having Regulatory Agencies
01:12:12you know treat like a token that could
01:12:14be everything from a Pokemon card to a
01:12:18stock certificate um you know treating
01:12:21every token as a stock CER certificate
01:12:24uh you know it' be it's kind of like
01:12:26weird and there was a really funny um
01:12:30exchange between uh Congressman Richie
01:12:32Torres uh and Gary Gensler where
01:12:35Congressman Torres said well like is a
01:12:38Pokemon card a security and Gary Gensler
01:12:41says of course not he says well it's a
01:12:43tokenized Pokemon card of security and
01:12:45Gary Gensler said well I don't know you
01:12:48know and so like that's how like goofy
01:12:51the regulatory state is and as a result
01:12:53what what happens is you know in the
01:12:54lack of clarity you have these scamsters
01:12:58casinos criminals um taking advantage of
01:13:01the kind of fuzzy nature of it and then
01:13:04the real entrepreneurs who are trying to
01:13:06improve the world are kind of hamstrung
01:13:09uh by the lack of clarity so um that
01:13:12that's probably the biggest thing that's
01:13:14kind of changed for me in the last five
01:13:16years yeah I would just say also look
01:13:17the other thing that happens and we've
01:13:18seen this across you know kind of
01:13:20repeated so-called crypto Winters is the
01:13:21other thing that happens and this
01:13:22happens in every other area Tech also by
01:13:24the way as sort of things you know sort
01:13:25of the Waves e and flow um is um you
01:13:28know there's often a sort of you know
01:13:29when there's sort of a reduction of kind
01:13:30of enthusiasm or press coverage or
01:13:31something um in reduction in height
01:13:33basically what happens is like the the
01:13:35actual Engineers keep working um and so
01:13:38and right and the actual entrepreneur
01:13:39keep building their businesses um and so
01:13:41the way that we at least always think
01:13:42about these things is you know basically
01:13:44when in doubt go look at the substance
01:13:46um and the substance here is like what
01:13:48are the what are the really smart
01:13:49engineers and entrepreneurs doing um and
01:13:51then what you basically get is you get
01:13:52this period you know this actually
01:13:54happened with the internet in the early
01:13:552000s um after theom crash where a lot
01:13:57of people you know if you just read the
01:13:59Press coverage it was like the
01:14:00internet's dead it was the whole thing
01:14:01was stupid the whole time um what
01:14:03actually happened was like a lot of
01:14:04really smart people at a lot of
01:14:05companies you know including companies
01:14:06like Google and Amazon and others right
01:14:08and new companies like Facebook like
01:14:09they just kept working um and then it
01:14:12turned out you know five years later it
01:14:13was like wow like there's all these like
01:14:15amazing new ideas and like you know and
01:14:17you know in that case it was like
01:14:18broadband and mobile broadband and
01:14:19smartphones and this and social media
01:14:21and you know kind of on and on and on
01:14:22right and all this stuff and it's like
01:14:24wow where did where did all that stuff
01:14:25come from and it's like well that's what
01:14:26the smart people were doing when when
01:14:28everybody else you know thought that
01:14:30when the internet was dead internet was
01:14:31dead right um and so I I think that that
01:14:34is just I I've just I've always viewed
01:14:35that as just like so much of how this
01:14:37industry Works uh the hype comes and
01:14:39goes the the substance is what matters
01:14:40and I just you know we we we we get to
01:14:42in our day jobs you know obviously spend
01:14:44a lot of time with you know many of the
01:14:45best crypto entrepreneurs and engineers
01:14:47and I I just the projects that I'm
01:14:48seeing are incredibly exciting so I'm U
01:14:51I feel very good about that yeah yeah
01:14:54yeah no doubt no doubt all right
01:14:57um here's a here's actually this is a
01:15:00good one for you um it was probably a
01:15:02good one for me too but uh uh Evan toin
01:15:06Feld asks what was your last strongly
01:15:11held belief that you've changed your
01:15:12mind on and why God there there are so
01:15:16many let's pick a pick a domain I don't
01:15:18know what domain should we talk about
01:15:19like well how about on how about on
01:15:24that's a good question I mean look
01:15:25there's just there's a lot of stuff we
01:15:27talked about this earlier with this kind
01:15:28of question about you know kind of the
01:15:29go bottles versus you know the the the
01:15:31Google sort of search model versus
01:15:32internet model or something which is
01:15:34just like like there's some just really
01:15:35big open questions right now like
01:15:37honestly for a lot of the AI stuff I'm
01:15:38trying to not actually draw judgments um
01:15:40just because the field is like flying so
01:15:42fast um it's moving so fast and I I you
01:15:44know there there are it is I find it
01:15:47very exciting but it is amazing how many
01:15:49of the most fundamental questions have
01:15:50really smart people right now on both
01:15:51sides um and so you know there are very
01:15:54smart people who are saying things like
01:15:56well you know I give you a bunch of
01:15:58examples there are a bunch of very smart
01:15:59people saying look LMS are way overhyped
01:16:02um it's basically they're basically just
01:16:04like fancy search engines um they're
01:16:06basically yeah they can recombine
01:16:07existing information but they can't
01:16:08actually understand anything they can't
01:16:09reason they can't plan they can't this
01:16:11they can't that and this this idea of
01:16:13like an emergent basically what they
01:16:14call the world model an emergent kind of
01:16:16uh generalized you know basically
01:16:18capability inside them is basically a
01:16:19mirage and then there are papers you can
01:16:21read where it's basically like yeah when
01:16:23these things perform well on tests is
01:16:24because the the content of the test is
01:16:25in the training data and so they're just
01:16:27like reg dating the training data back
01:16:28at you look there are other extremely
01:16:31smart people including people I talked
01:16:32to just this week who are like oh no
01:16:34like no no no like what's happening
01:16:35inside the neural networks is actually
01:16:37like the development of entirely new
01:16:38computational capabilities including
01:16:40full reasoning skills and you know and
01:16:42then you layer in you know this is the
01:16:43excitement around the qar thing and and
01:16:45other advances like this you layer in
01:16:47some of these new advances and you're
01:16:48going to have like capabilities you know
01:16:50I mean look Elon was just on stage and
01:16:51said you know AGI within three years
01:16:53you know based on the current based on
01:16:54the current technology track um and so
01:16:57like like you know like like if if
01:16:59that's that if that's as big as a
01:17:00question as it is then like all the sub
01:17:01questions underneath that are also
01:17:03likely to be you know kind of very open
01:17:05questions um yeah you know another one
01:17:07just like the synthetic data thing like
01:17:08you know synthetic data we we actually I
01:17:10have a I have a very very important High
01:17:12Saks bet with our partner Martin cassado
01:17:15yeah yeah we have bet the princely sum
01:17:17of $1 um on on on the following question
01:17:20which is like does synthetic is
01:17:21synthetic training data actually useful
01:17:22for train tring large language models
01:17:24and um and you know the the the the con
01:17:27argument um is um sort of sort of that
01:17:30information Theory argument which is uh
01:17:32synthetic training data is sort of
01:17:34necessarily Downstream of sort of you
01:17:35know human training data um you know so
01:17:38and so it's there going to be a
01:17:39recombination of signal that's already
01:17:40present in the training data um and so
01:17:42therefore it can't add any value and
01:17:44then there's a lot of smart people now
01:17:45who are including Recent research papers
01:17:47on the other side of that saying no
01:17:48actually synthetic training data works
01:17:49as well as and maybe better than human
01:17:51training data um and they argument there
01:17:53would just be like look just look at the
01:17:54output the output of a gp4 query or the
01:17:57output of a midgy image generator is
01:17:59often actually better than what most
01:18:00people can do um and so why wouldn't it
01:18:03be better than the human training data
01:18:05um and like so that's still an open
01:18:06question so I I'm I'm trying to yeah I
01:18:08don't know this is a incredibly long and
01:18:10Winding Way to dodge the question but um
01:18:13you know it's like it's like I don't
01:18:14know like beliefs I I'm find I don't
01:18:16know as I get older I find belief seem
01:18:17to seem to be less and less important
01:18:20um yeah well I I I think that's actually
01:18:23the probably the biggest lesson right is
01:18:27um to to give up on the need for
01:18:31consistency like once you let go of that
01:18:34you become much much smarter I think um
01:18:37which is uh you know quite a big uh
01:18:42quite quite a big advance in life is to
01:18:45you know at that point when you say you
01:18:47know what I believe that yesterday and
01:18:50uh I no longer believe it today that's a
01:18:53that's a huge um amount of freedom and
01:18:57increase in Effectiveness yeah J Jordan
01:18:59Peterson once said he he was talking
01:19:00about why why do people react so
01:19:02emotionally and negatively to having
01:19:04their beliefs challenged um right people
01:19:07get really mad like if if you know if
01:19:09you believe a and I try to convince you
01:19:10not like people get like really upset um
01:19:13and he his theory on it is's a you know
01:19:14psychologist way background his theory
01:19:16on it is basically people that the
01:19:18belief occupies the same sort of mental
01:19:20circuitry as your children um ah so
01:19:24there was my baby my baby i b this idea
01:19:27yeah this idea is my baby and like it's
01:19:30my child um and like like it's CT of my
01:19:32identity and like I'm responsible for it
01:19:34um and I've committed myself to it and
01:19:36I'm emotionally entangled with it and if
01:19:38you're going to like tell me that I need
01:19:39to let go of my belief it's like telling
01:19:40me I need to kill my
01:19:41baby right and it generates that same
01:19:44level of like visceral response like for
01:19:45the same reason and you say there's
01:19:46obvious problem with it which is you
01:19:47know these aren't these aren't you know
01:19:48it's not your baby right like it's not
01:19:51yeah you know it's it's not real it
01:19:52can't love you back um and so um you
01:19:56know yeah being being yeah more and more
01:19:58I try to just get myself into the state
01:20:00where it's like okay I don't I don't I
01:20:01don't actually need beliefs I need
01:20:04knowledge and then I need the ability to
01:20:06actually like process new information
01:20:07and try to actually be correct um which
01:20:10which it just turns out is just
01:20:11incredibly hard yeah well it's hard
01:20:14psychologically but uh once you learn to
01:20:17do it it turns out to be a much better
01:20:20life you know and and you get Furious
01:20:23less yes all right well thank you
01:20:26everyone for joining us that was uh the
01:20:28latest episode of the mark and Ben show
01:20:30and we look forward to seeing you next
01:20:32time great thank you everybody thank you