Latest / Core Memory / The Child Prodigy Teaching Other People How To Learn

Transcript

The Child Prodigy Teaching Other People How To Learn · Core Memory

• 0:00
  i don't know i don't do like a crazy intro um i'll keep it
• 0:06
  pretty quick but i'll
• 0:07
  babble here just for a second and um i don't know you feel okay
• 0:12
  anything else no
• 0:14
  SPEAKER_01: good let's do it
• 0:28
  SPEAKER_01: Thank you, Sue.
• 0:29
  SPEAKER_01: Thank you for coming.
• 0:30
  SPEAKER_01: Of course.
• 0:31
  SPEAKER_01: Thanks for having me.
• 0:32
  SPEAKER_01: It's a pleasure.
• 0:33
  SPEAKER_01: Absolute pleasure.
• 0:34
  I know your brother,
• 0:35
  Patrick,
• 0:36
  who runs the Ark Institute,
• 0:38
  this temple of science.
• 0:41
  SPEAKER_00: Not too far from here.
• 0:42
  SPEAKER_01: Yeah, down the street in Palo Alto.
• 0:45
  SPEAKER_01: And this is a really cool place.
• 0:47
  SPEAKER_01: I'm sure hopefully we'll have Patrick on at some point.
• 0:50
  SPEAKER_01: He...
• 0:52
  SPEAKER_01: his life story,
• 0:53
  SPEAKER_01: which I'm going to write about and I'm behind on writing about,
• 0:59
  SPEAKER_01: just got me so fascinated with your family and you guys' background.
• 1:06
  SPEAKER_01: And then I got interested in you, you know?
• 1:09
  SPEAKER_01: And just after speaking to him a little bit.
• 1:11
  SPEAKER_01: Of course, yeah.
• 1:13
  SPEAKER_01: Okay, so set the table for people who might not know.
• 1:17
  SPEAKER_01: You are the co-founder and chief technology officer of Speak.
• 1:20
  SPEAKER_01: Yes.
• 1:21
  SPEAKER_01: It helps people learn new languages.
• 1:24
  SPEAKER_01: You guys were ahead of the AI curve, I feel like.
• 1:27
  SPEAKER_01: You started, it's almost eight years old?
• 1:30
  SPEAKER_00: Yeah, so we were YC winter 2017.
• 1:34
  I think we technically incorporated at the end of 2016,
• 1:37
  but yeah,
• 1:37
  it's been about eight years.
• 1:39
  SPEAKER_01: And it was like AI from the beginning, right?
• 1:41
  SPEAKER_00: Yeah, it was always about AI and about this idea that
• 1:47
  SPEAKER_00: at some point in five to 10 years,
• 1:49
  SPEAKER_00: this is back in 2017,
• 1:50
  SPEAKER_00: language models,
• 1:51
  speech models would become superhuman and that would let us
• 1:54
  build all sorts of
• 1:55
  SPEAKER_00: really cool AI products that would be very transformative.
• 1:57
  SPEAKER_00: And that was a good prediction.
• 1:59
  SPEAKER_00: Yeah.
• 2:00
  SPEAKER_01: Like really good.
• 2:01
  SPEAKER_01: I feel like you guys were,
• 2:03
  ahead of the curve transformers hadn't come out yet i mean
• 2:06
  people were doing ai
• 2:07
  stuff but not that was a good good bet um so we are going to
• 2:11
  get into speak another
• 2:12
  reason and i hope this is i don't think this is embarrassing
• 2:15
  maybe i mean i don't
• 2:17
  know i mean you've been described as a child prodigy yes whole
• 2:21
  life and and and
• 2:22
  we're a child prodigy i mean you and your brother i i only know
• 2:26
  you there's three
• 2:27
  SPEAKER_01: brothers yes but i only know you and patrick's story um
• 2:31
  SPEAKER_01: as well.
• 2:31
  SPEAKER_01: And so,
• 2:33
  SPEAKER_01: but both flew through school,
• 2:37
  SPEAKER_01: ended up being homeschooled,
• 2:39
  SPEAKER_01: went to university super early.
• 2:44
  You dropped out of your PhD program and we can get in all that,
• 2:49
  but basically we're just like,
• 2:50
  SPEAKER_01: Flying through, winning science contests, writing books.
• 2:53
  SPEAKER_01: You even were like a really good athlete, I think, as well.
• 2:57
  SPEAKER_00: When I was young.
• 2:58
  SPEAKER_00: That was a long time ago.
• 2:59
  SPEAKER_01: Yeah.
• 2:59
  SPEAKER_01: So just like incredible accomplishments.
• 3:04
  And I did,
• 3:04
  yeah,
• 3:05
  I just,
• 3:05
  I think it's like this fascinating window into a unique life.
• 3:10
  SPEAKER_01: And so I wanted to get into that a bit.
• 3:13
  SPEAKER_01: So maybe if we...
• 3:15
  SPEAKER_01: If you're okay,
• 3:16
  we'll kind of start biographically and then get into Speak and
• 3:20
  the rest of the
• 3:20
  SPEAKER_01: world.
• 3:21
  SPEAKER_01: Yeah, totally.
• 3:22
  SPEAKER_01: Okay.
• 3:22
  As always,
• 3:23
  this podcast is made possible by the fine people at E1
• 3:28
  Ventures.
• 3:29
  SPEAKER_01: Thank you, U1 Ventures, for doing your thing.
• 3:31
  SPEAKER_01: Okay, here we go.
• 3:32
  SPEAKER_01: So you're okay talking about this.
• 3:35
  SPEAKER_00: This is not- Oh, I'm very used to talking about it.
• 3:37
  SPEAKER_01: I know.
• 3:37
  SPEAKER_00: I know.
• 3:37
  SPEAKER_00: Let's do it.
• 3:38
  SPEAKER_01: I'm sorry.
• 3:38
  SPEAKER_01: Okay.
• 3:39
  SPEAKER_00: Don't be sorry.
• 3:41
  SPEAKER_01: So Taiwanese parents, I believe.
• 3:44
  SPEAKER_01: Yes.
• 3:44
  SPEAKER_01: But grew up around Seattle.
• 3:46
  SPEAKER_01: Yes.
• 3:46
  SPEAKER_01: Yes.
• 3:48
  SPEAKER_01: And it's like these stories, you know, I was doing my research.
• 3:51
  SPEAKER_01: I mean,
• 3:51
  SPEAKER_01: I don't know why they necessarily highlighted this,
• 3:54
  SPEAKER_01: but at age of two,
• 3:56
  building towering Lego robot structures and then six doing
• 4:02
  multiplication,
• 4:03
  SPEAKER_01: reading classical literature,
• 4:05
  SPEAKER_01: and then started homeschooling around eight years.
• 4:08
  SPEAKER_01: Yeah.
• 4:08
  SPEAKER_01: I mean,
• 4:08
  SPEAKER_01: when,
• 4:09
  SPEAKER_01: when did you,
• 4:11
  SPEAKER_01: I've read enough to see that,
• 4:12
  SPEAKER_01: that you were,
• 4:13
  SPEAKER_01: it sounded like you were frustrated at the pace of school and,
• 4:18
  SPEAKER_01: and,
• 4:18
  but you must have started to have some sense that you're a bit
• 4:21
  different than the
• 4:22
  SPEAKER_01: other kids.
• 4:22
  SPEAKER_01: And you're the oldest brother.
• 4:23
  SPEAKER_01: Yeah.
• 4:24
  SPEAKER_00: I'm the oldest and Patrick is one year younger.
• 4:28
  SPEAKER_00: And then we have a third brother, the youngest who is nine years younger.
• 4:31
  SPEAKER_01: Okay.
• 4:32
  SPEAKER_00: Yeah.
• 4:32
  SPEAKER_00: Benjamin?
• 4:33
  SPEAKER_00: Benjamin.
• 4:33
  SPEAKER_00: That's right.
• 4:34
  SPEAKER_01: Okay.
• 4:34
  SPEAKER_01: Okay.
• 4:34
  SPEAKER_01: And so, so you're the one who's like going through all of this first.
• 4:38
  SPEAKER_00: Yes.
• 4:38
  SPEAKER_01: Yeah.
• 4:39
  SPEAKER_01: So what was that?
• 4:39
  SPEAKER_01: I mean, what was that like?
• 4:41
  SPEAKER_00: Yeah.
• 4:41
  SPEAKER_00: So we were in the Seattle area,
• 4:44
  SPEAKER_00: um,
• 4:44
  SPEAKER_00: grew up in Issaquah and lived in a few places around Seattle.
• 4:49
  SPEAKER_00: But in general, um, uh, I was in public school in
• 4:56
  SPEAKER_00: public elementary school track.
• 4:59
  SPEAKER_00: They put me in the gifted track.
• 5:02
  SPEAKER_00: And I think it was third grade,
• 5:05
  SPEAKER_00: fourth grade,
• 5:06
  SPEAKER_00: where I was just...
• 5:08
  SPEAKER_00: I remember being really bored in class.
• 5:11
  And I would sort of complete all of the quizzes and the tests
• 5:15
  really fast and then
• 5:17
  SPEAKER_00: start bothering people,
• 5:18
  SPEAKER_00: start bothering other kids.
• 5:21
  SPEAKER_00: And then eventually...
• 5:24
  It was so distracting that they tried to put me in the adjacent
• 5:26
  empty classroom
• 5:27
  SPEAKER_00: just to read so that I would stay quiet and not be annoying.
• 5:31
  SPEAKER_00: And then, you know, in hindsight, this is a pretty crazy move.
• 5:38
  But my parents decided that,
• 5:40
  OK,
• 5:40
  this is clearly not working and we should homeschool him.
• 5:46
  So they took me out of school and then decided to basically
• 5:50
  devote their lives to
• 5:52
  SPEAKER_00: homeschooling their kids.
• 5:54
  And I think from my perspective,
• 5:57
  obviously I've only had one experience,
• 6:00
  right?
• 6:01
  SPEAKER_00: It felt normal at the time, but it's objectively pretty weird, right?
• 6:05
  SPEAKER_00: Very accelerated.
• 6:07
  SPEAKER_00: And I started being homeschooled.
• 6:12
  SPEAKER_00: I had a bunch of private tutors that my parents hired.
• 6:16
  SPEAKER_00: They did everything they could,
• 6:19
  SPEAKER_00: I think,
• 6:19
  to provide us with opportunities and resources to just learn as
• 6:24
  fast as possible.
• 6:24
  SPEAKER_00: And I think that naturally,
• 6:26
  SPEAKER_00: I was kind of like a sponge and I,
• 6:29
  SPEAKER_00: you know,
• 6:29
  SPEAKER_00: I just love learning and I still do today.
• 6:32
  SPEAKER_00: And we'll talk more about that.
• 6:33
  SPEAKER_00: But in general...
• 6:37
  The path progressed where eventually I was 11 and I had gone
• 6:44
  through like a sort of
• 6:45
  SPEAKER_00: virtual high school curriculum.
• 6:47
  SPEAKER_00: And it was just kind of obvious that, hey, there's nothing else.
• 6:49
  SPEAKER_00: The next step is college.
• 6:52
  SPEAKER_00: So I took the SATs, ACTs, and then started college when I was 12.
• 6:58
  SPEAKER_01: Pretty unusual.
• 7:00
  SPEAKER_01: Three or four undergraduate degrees, right?
• 7:04
  SPEAKER_00: Yeah.
• 7:05
  SPEAKER_01: By the time you were 16.
• 7:06
  SPEAKER_00: I did three degrees in four years.
• 7:07
  SPEAKER_01: Neurobiology, biochemistry, and chemistry.
• 7:10
  SPEAKER_00: Yeah.
• 7:10
  SPEAKER_01: And a minor in math.
• 7:11
  SPEAKER_01: Yeah, that's right.
• 7:12
  SPEAKER_01: Yeah.
• 7:13
  SPEAKER_01: Unbelievable.
• 7:14
  SPEAKER_01: And I saw your mom...
• 7:17
  made but seems like a sensible decision you'd apply to other
• 7:20
  schools but she she
• 7:21
  didn't want you to live in a dorm and so yeah i was i was
• 7:24
  living at home and i
• 7:25
  would take the bus to college every day and um yeah it was it
• 7:30
  was weird for sure
• 7:31
  yeah yeah and i think look i mean people read about these
• 7:35
  stories they always catch
• 7:36
  SPEAKER_01: people's attention um
• 7:39
  SPEAKER_01: Was it... Yeah, does this... I don't... Like, what is it like?
• 7:44
  SPEAKER_01: People are writing stories about you from such a young age,
• 7:47
  SPEAKER_01: and I couldn't tell if one feels like a freak,
• 7:51
  SPEAKER_01: if one's proud,
• 7:52
  SPEAKER_01: if it's just your life,
• 7:54
  SPEAKER_01: you don't know anything else,
• 7:55
  SPEAKER_01: and so it just is what it is.
• 7:57
  You kind of mentioned this,
• 7:58
  but I was also a competitive swimmer before college.
• 8:02
  SPEAKER_00: So, you know, I mean, this is much younger, like ages 9 through 12.
• 8:06
  SPEAKER_00: And I was pretty good for...
• 8:09
  SPEAKER_00: And I think through that track,
• 8:11
  SPEAKER_00: that was kind of my sort of socialization with kids of a similar age.
• 8:17
  But I think I just rapidly normalized everything and became
• 8:23
  very comfortable and
• 8:23
  SPEAKER_00: normal that,
• 8:24
  SPEAKER_00: oh yeah,
• 8:25
  SPEAKER_00: all my friends are 18 and 19 and 20,
• 8:28
  SPEAKER_00: right?
• 8:28
  SPEAKER_00: And I'm just very comfortable spending time and hanging out and having
• 8:32
  SPEAKER_00: conversations with adults that are 10,
• 8:35
  SPEAKER_00: 20 years older than me.
• 8:37
  SPEAKER_00: So that was pretty normalized.
• 8:38
  SPEAKER_01: And what did your parents do?
• 8:42
  SPEAKER_01: What did they do for their jobs?
• 8:44
  So they've also had kind of unusual lives where they were
• 8:48
  immigrants and they
• 8:51
  SPEAKER_00: basically worked on a variety of initiatives.
• 8:55
  Basically,
• 8:55
  my mom is an economist by training and then she became a
• 8:58
  stay-at-home mom.
• 8:59
  And my dad was a computer scientist by training and then did
• 9:03
  that for a while.
• 9:04
  SPEAKER_00: And then...
• 9:07
  SPEAKER_00: Really,
• 9:07
  SPEAKER_00: I think when they made the decision to homeschool me,
• 9:10
  SPEAKER_00: that really reconfigured their entire lives.
• 9:14
  And that was kind of the main focus from then on,
• 9:16
  you know,
• 9:17
  like an enormous amount of energy.
• 9:19
  SPEAKER_00: But before that, they had, I think, done just many random things.
• 9:25
  Like we built our childhood house and they were general
• 9:29
  contractors for that.
• 9:30
  SPEAKER_00: They ran an art gallery.
• 9:31
  SPEAKER_00: Yeah.
• 9:33
  SPEAKER_00: a bunch of other small businesses as well.
• 9:35
  And I think maybe there was part of me through just hearing
• 9:39
  about that and seeing that
• 9:42
  um it also normalized this idea of agency that you can do you
• 9:47
  don't have to follow
• 9:47
  the normal path you can do many things that other people might
• 9:50
  consider unusual or
• 9:51
  like non-traditional yeah and that became very normal yeah and
• 9:54
  they were they were
• 9:55
  working while they were i mean you talked about your mom but um
• 9:58
  was your dad
• 9:59
  working while you were they were mostly not i mean it was
• 10:01
  because patrick was home
• 10:03
  SPEAKER_00: school this well after a few years and i think
• 10:06
  SPEAKER_00: setting up all the structure.
• 10:07
  SPEAKER_00: And, you know, they were also personally teaching us for a bit.
• 10:10
  SPEAKER_00: I think that that was basically the full-time job.
• 10:12
  SPEAKER_00: So, yeah.
• 10:13
  SPEAKER_01: And if I could ask,
• 10:14
  SPEAKER_01: I mean,
• 10:14
  SPEAKER_01: how did they pay the bills and like things tight or they had done well?
• 10:21
  I think they did a good job sort of shielding us from those
• 10:25
  details when we were younger.
• 10:27
  And I think they had basically saved through their early career
• 10:30
  and essentially...
• 10:33
  decided that there's actually nothing more important than
• 10:37
  essentially giving all
• 10:38
  SPEAKER_00: the opportunities that we humanly can to our children.
• 10:42
  SPEAKER_00: That's like their fundamental life philosophy.
• 10:45
  SPEAKER_00: So,
• 10:45
  SPEAKER_00: you know,
• 10:46
  SPEAKER_00: I think they,
• 10:47
  SPEAKER_00: it wasn't easy,
• 10:48
  but I think they made it happen and they viewed as like by far
• 10:51
  the most important
• 10:51
  SPEAKER_00: thing they could be doing with their lives.
• 10:53
  I'm sure some people who hear your story or hear us talking now
• 10:58
  will think tiger parents.
• 11:01
  And every time I've seen you and Patrick talk about your
• 11:05
  parents,
• 11:06
  it seems like
• 11:08
  SPEAKER_01: quite loving and,
• 11:09
  SPEAKER_01: and,
• 11:10
  SPEAKER_01: and the,
• 11:10
  SPEAKER_01: it's not like you guys,
• 11:12
  my impression is that it wasn't like you were super stressed
• 11:15
  out and being pushed
• 11:16
  SPEAKER_01: to,
• 11:16
  it was just that you were super bright and into all kinds of
• 11:19
  interesting things.
• 11:21
  SPEAKER_00: I think there's a lot of nuance there, right?
• 11:22
  SPEAKER_00: It's not as simple as just tiger parent or not.
• 11:26
  SPEAKER_00: Right.
• 11:26
  SPEAKER_00: I think it is just this fundamental attitude of,
• 11:30
  SPEAKER_00: um,
• 11:31
  SPEAKER_00: having a bar of excellence,
• 11:32
  SPEAKER_00: caring a lot about
• 11:35
  SPEAKER_00: sort of learning and education and opportunity.
• 11:38
  SPEAKER_00: Um,
• 11:39
  there were so many sort of what I now know as like excessive
• 11:44
  things that they did
• 11:45
  SPEAKER_00: just to kind of provide us with a whole many of options,
• 11:48
  SPEAKER_00: right?
• 11:48
  SPEAKER_00: Like,
• 11:49
  SPEAKER_00: I mean,
• 11:49
  SPEAKER_00: this is kind of absurd,
• 11:50
  SPEAKER_00: but I,
• 11:51
  SPEAKER_00: I remember at like some point taking ballet lessons and we did,
• 11:55
  SPEAKER_00: um,
• 11:56
  clay pottery lessons like all sorts of private tutoring and
• 12:00
  like just things to try
• 12:02
  i think um and i think certainly there was kind of the um idea
• 12:11
  that patrick and i
• 12:13
  were different right and this was a different path of
• 12:16
  excellence um and uh yeah
• 12:23
  SPEAKER_00: that was like very much normalized
• 12:24
  SPEAKER_01: Reminds me,
• 12:25
  SPEAKER_01: I remember,
• 12:26
  SPEAKER_01: I'm sure there's quite a few people like this through history,
• 12:29
  SPEAKER_01: but reading about John Stuart Mill and what his parents did with him.
• 12:33
  I mean,
• 12:33
  by the time he was five,
• 12:34
  I think he was reading Greek and Latin and all kinds of texts.
• 12:38
  SPEAKER_01: Sort of this idea of his father wanted to...
• 12:42
  could tell that he was precocious and wanted to build a type of
• 12:47
  renaissance man,
• 12:48
  SPEAKER_01: somebody who was well-rounded.
• 12:49
  SPEAKER_01: And I mean,
• 12:50
  SPEAKER_01: it sounds like it was similar,
• 12:53
  giving you all these avenues to see what clicked and you were
• 12:57
  really drawn to.
• 12:58
  SPEAKER_01: And both you and Patrick quite drawn to science from a pretty young age.
• 13:07
  SPEAKER_00: Yeah,
• 13:07
  SPEAKER_00: I think it was always pretty natural curiosity,
• 13:11
  SPEAKER_00: this idea of understanding the world around us,
• 13:14
  SPEAKER_00: understanding biology,
• 13:15
  SPEAKER_00: understanding the human body.
• 13:17
  SPEAKER_00: That was just like always just extremely fascinating.
• 13:22
  SPEAKER_00: We also got into computers and the internet and making websites.
• 13:27
  And that was kind of the start of my own software journey,
• 13:29
  I suppose,
• 13:30
  pretty young.
• 13:31
  SPEAKER_00: Yeah.
• 13:31
  SPEAKER_01: Yeah.
• 13:32
  SPEAKER_01: And did you guys,
• 13:32
  did the two of you feel competitive or was it nice to have
• 13:35
  somebody there for this
• 13:38
  SPEAKER_00: journey?
• 13:38
  Well,
• 13:39
  I'm sure Patrick would have a pretty different perspective on
• 13:42
  it than me.
• 13:44
  SPEAKER_00: Um,
• 13:44
  SPEAKER_00: I think we were,
• 13:46
  SPEAKER_00: I mean,
• 13:47
  we were both young and I think we also felt under different
• 13:51
  pressures and,
• 13:53
  SPEAKER_00: you know,
• 13:53
  SPEAKER_00: growing up as young adolescents,
• 13:55
  SPEAKER_00: um,
• 13:57
  SPEAKER_00: We have a pretty close relationship today,
• 14:00
  SPEAKER_00: and he lives in the Bay Area as well,
• 14:03
  SPEAKER_00: not too far from me,
• 14:03
  SPEAKER_00: San Francisco,
• 14:04
  SPEAKER_00: which is awesome.
• 14:05
  SPEAKER_00: Yeah.
• 14:05
  SPEAKER_00: Yeah.
• 14:05
  SPEAKER_01: Yeah.
• 14:07
  SPEAKER_01: Okay.
• 14:07
  SPEAKER_01: And at one point,
• 14:11
  SPEAKER_01: I was reading,
• 14:12
  it felt like people were asking you about all this so much that
• 14:15
  even as a teenager,
• 14:16
  SPEAKER_01: I think you wrote like a book and I tried to find it last night,
• 14:19
  SPEAKER_01: but I couldn't find it.
• 14:21
  SPEAKER_00: I think it's probably out of print now,
• 14:22
  SPEAKER_00: but,
• 14:23
  SPEAKER_00: but yeah,
• 14:24
  SPEAKER_00: I,
• 14:24
  SPEAKER_00: um,
• 14:25
  SPEAKER_00: I was kind of like a,
• 14:27
  SPEAKER_00: like a minor celebrity in some circles,
• 14:28
  SPEAKER_00: I suppose.
• 14:29
  SPEAKER_00: Right.
• 14:29
  SPEAKER_00: Due to these unusual events.
• 14:31
  Like in the Pacific Northwest and Taiwan,
• 14:34
  it's like,
• 14:34
  yeah,
• 14:35
  yeah,
• 14:35
  that's right.
• 14:36
  SPEAKER_00: Um, and I, um,
• 14:40
  we did like a little tour of Taiwan and we visited a bunch of
• 14:44
  schools and there was
• 14:45
  SPEAKER_00: a bunch of press.
• 14:46
  SPEAKER_00: So I think like I've,
• 14:48
  I've been just very used to that and being in the spotlight for
• 14:53
  a long time.
• 14:54
  SPEAKER_01: Yeah.
• 14:54
  SPEAKER_00: Yeah.
• 14:55
  SPEAKER_01: Yeah.
• 14:55
  SPEAKER_01: And it seems like,
• 14:58
  SPEAKER_01: anecdotally it seems like 60 minutes every like six years runs.
• 15:03
  SPEAKER_01: Yeah.
• 15:03
  SPEAKER_00: It was like a human interest story.
• 15:04
  SPEAKER_00: Yeah.
• 15:04
  SPEAKER_01: There's this kid,
• 15:05
  SPEAKER_01: but I,
• 15:05
  SPEAKER_01: and then sometimes I Google these people and then they,
• 15:09
  SPEAKER_01: they can,
• 15:10
  SPEAKER_01: it can get weird.
• 15:11
  SPEAKER_01: They do,
• 15:11
  SPEAKER_01: they can take some weird,
• 15:12
  SPEAKER_01: they often don't end up kind of like where you think I feel like you,
• 15:17
  and your brother at least i know your brother a bit better just
• 15:21
  feels like quite
• 15:22
  grounded and and like a nice part you know i don't know like
• 15:26
  like normal and who
• 15:28
  gives a shit what normal is but but like you know functional
• 15:32
  and high functioning
• 15:33
  SPEAKER_01: yeah i think we turn out okay yeah yeah and did you um
• 15:40
  Yeah,
• 15:41
  I mean,
• 15:41
  were there cautionary tales you or your parents were trying?
• 15:45
  SPEAKER_01: Because I feel like,
• 15:46
  isn't this kind of a thing that this doesn't actually usually
• 15:48
  end that well?
• 15:50
  SPEAKER_00: I think the sort of, you know, main...
• 15:55
  the generic typical cautionary tale is that child has been sort
• 16:00
  of told their
• 16:01
  SPEAKER_00: entire life that they're smarter than others.
• 16:03
  SPEAKER_00: And then they grow up and find out that's not actually true or that,
• 16:07
  SPEAKER_00: you know,
• 16:07
  they've never had to deal with adversity and failure and can't
• 16:11
  handle it and have a
• 16:12
  SPEAKER_00: huge crash out and then like a breakdown or whatever.
• 16:16
  SPEAKER_00: Yeah.
• 16:18
  SPEAKER_00: I'm sure that happens a lot.
• 16:19
  SPEAKER_00: Maybe that's even the default case.
• 16:21
  SPEAKER_00: I think that definitely part of it was luck.
• 16:26
  SPEAKER_00: If you look at kind of the journey after
• 16:30
  college I came down to Stanford for grad school and I spent a
• 16:34
  few years in grad
• 16:34
  school doing a PhD in neuroscience and then dropped out for the
• 16:39
  first class of the
• 16:40
  SPEAKER_00: Teal Fellowship and that's kind of how my whole sort of startup software
• 16:44
  integration into Silicon Valley journey started yeah so me you
• 16:47
  know like maybe in
• 16:50
  SPEAKER_00: some sense I got lucky
• 16:53
  SPEAKER_00: just through that path.
• 16:54
  SPEAKER_00: And through that path,
• 16:56
  SPEAKER_00: meeting all sorts of,
• 16:59
  SPEAKER_00: you know,
• 17:00
  SPEAKER_00: also very precocious,
• 17:01
  SPEAKER_00: just incredibly intelligent people that went on to do amazing things.
• 17:04
  SPEAKER_00: So maybe, you know, fell in with the right crowd.
• 17:07
  SPEAKER_00: Right, right.
• 17:08
  SPEAKER_01: Okay, and I did want to get into that.
• 17:10
  SPEAKER_01: I mean,
• 17:11
  SPEAKER_01: so you do these undergrad degrees,
• 17:14
  SPEAKER_01: you go to Stanford,
• 17:15
  SPEAKER_01: you're heading towards a PhD in,
• 17:18
  SPEAKER_01: was it biochemistry?
• 17:19
  SPEAKER_00: Neuroscience.
• 17:19
  SPEAKER_00: Neuroscience.
• 17:20
  SPEAKER_01: And so-
• 17:25
  SPEAKER_01: obviously still quite young doing this.
• 17:27
  SPEAKER_00: Yeah.
• 17:27
  SPEAKER_00: I started grad school when I was 16.
• 17:29
  SPEAKER_00: 16.
• 17:29
  SPEAKER_00: Yeah.
• 17:30
  SPEAKER_01: And you're, you're in this.
• 17:31
  SPEAKER_00: Had my first drink at a happy hour in grad school when I was 17.
• 17:37
  SPEAKER_01: Is that, they let you do that?
• 17:41
  SPEAKER_00: Actually, a few people were horrified.
• 17:42
  SPEAKER_00: I remember.
• 17:43
  SPEAKER_01: Yeah.
• 17:44
  Did you,
• 17:45
  before we get there,
• 17:45
  did you ever meet Nathan Meervold when you were growing up?
• 17:48
  SPEAKER_00: No, I didn't.
• 17:49
  SPEAKER_00: Okay.
• 17:49
  SPEAKER_00: But I've read his books, obviously.
• 17:50
  SPEAKER_01: Yeah, but I mean, he experienced... That's true, yeah.
• 17:53
  SPEAKER_01: He gave a similar story, right?
• 17:55
  SPEAKER_01: And lived right near you guys.
• 17:56
  SPEAKER_01: That's right.
• 17:57
  SPEAKER_01: So I was curious if you had run into each other.
• 18:00
  SPEAKER_01: Okay,
• 18:00
  SPEAKER_01: and then it seemed like you were quite close to finishing your PhD,
• 18:04
  SPEAKER_01: or you must have been...
• 18:06
  SPEAKER_00: I was three years in.
• 18:07
  And,
• 18:07
  you know,
• 18:08
  so I don't know,
• 18:08
  maybe like a little bit more than halfway.
• 18:11
  SPEAKER_00: And it was Stanford.
• 18:13
  SPEAKER_00: So, you know, the heart of Silicon Valley.
• 18:16
  And a lot of my classmates were also becoming disillusioned
• 18:20
  with academia,
• 18:22
  SPEAKER_00: dropping out to college.
• 18:24
  SPEAKER_00: join startups.
• 18:24
  That was like the first place where I learned about startups
• 18:27
  and just like the whole concept.
• 18:29
  SPEAKER_00: It was really fascinating to me.
• 18:32
  And I felt like I think this could be a possible path that
• 18:39
  seems interesting,
• 18:40
  seems like a way to get more leverage and also pursue my
• 18:44
  interest in technology.
• 18:46
  SPEAKER_00: help more people eventually.
• 18:48
  And I started taking more classes at the GSB,
• 18:50
  eventually learned about the Teal Fellowship.
• 18:53
  SPEAKER_00: A friend told me about it.
• 18:55
  SPEAKER_00: And the timing just lined up really well, frankly.
• 18:57
  SPEAKER_01: And you mentioned it, and it was the very first class of Teal Fellows.
• 19:00
  Yeah,
• 19:01
  so I was 19 at the time,
• 19:02
  and it was 2011,
• 19:03
  which was the first class.
• 19:04
  And I mean,
• 19:04
  it was like a con,
• 19:06
  it was a very Peter's always controversial,
• 19:09
  I guess.
• 19:09
  Um,
• 19:10
  but,
• 19:10
  but this was super controversial thing when he announced it or
• 19:13
  20.
• 19:13
  SPEAKER_00: Yeah.
• 19:14
  SPEAKER_01: I mean,
• 19:14
  SPEAKER_01: you know,
• 19:14
  SPEAKER_01: and like,
• 19:15
  SPEAKER_01: as time has gone on,
• 19:16
  SPEAKER_01: um,
• 19:18
  SPEAKER_01: you feel like it would be an easier to see,
• 19:20
  SPEAKER_01: to show easier decision to make,
• 19:22
  SPEAKER_01: um,
• 19:23
  SPEAKER_01: just because other people have gone down this path,
• 19:26
  SPEAKER_01: but I mean,
• 19:26
  SPEAKER_01: that must've been,
• 19:27
  SPEAKER_01: um,
• 19:29
  SPEAKER_01: I don't know if scary is the right word or so,
• 19:30
  SPEAKER_01: but you know,
• 19:31
  to be the very first to make this jump and take this hundred
• 19:34
  thousand dollars and
• 19:36
  SPEAKER_01: not only that,
• 19:36
  SPEAKER_01: but be pushing against society's wants and desires.
• 19:41
  SPEAKER_01: I will say, I think I was...
• 19:46
  SPEAKER_00: a bit more of an unusual case because I did finish college, right?
• 19:49
  SPEAKER_00: Everyone else was like a college dropout.
• 19:51
  SPEAKER_00: So I was a grad school dropout.
• 19:53
  SPEAKER_00: Still a dropout though.
• 19:54
  SPEAKER_01: Right, right, right.
• 19:55
  SPEAKER_00: So yeah.
• 19:56
  SPEAKER_01: And how did your parents react to that?
• 19:59
  SPEAKER_00: No, they were supportive.
• 20:00
  SPEAKER_00: I think that they wanted me to blaze my own path and be happy.
• 20:05
  SPEAKER_00: And I think that they could see that this made sense as well.
• 20:09
  SPEAKER_01: Okay.
• 20:10
  SPEAKER_01: Yeah.
• 20:10
  SPEAKER_01: Okay.
• 20:10
  And you talked about wanting to help people and you've done a
• 20:16
  couple startups.
• 20:17
  They all would fall into like education,
• 20:23
  I think it's fair to say,
• 20:24
  and helping people that way.
• 20:27
  And even when you and Patrick,
• 20:30
  I think when you were teenagers,
• 20:31
  you also did...
• 20:34
  SPEAKER_01: weren't you kind of fundraising or,
• 20:35
  SPEAKER_01: or you'd created a charity or like,
• 20:37
  SPEAKER_01: this seems like it's always been a thing.
• 20:38
  SPEAKER_00: Yeah.
• 20:39
  SPEAKER_00: I think,
• 20:40
  SPEAKER_00: um,
• 20:40
  SPEAKER_00: sort of going back to the earlier childhood years,
• 20:44
  SPEAKER_00: something that my parents,
• 20:46
  SPEAKER_00: our parents always emphasized was sort of virtues and,
• 20:52
  SPEAKER_00: you know,
• 20:53
  SPEAKER_00: having a strong moral compass and a sort of purpose in your life.
• 21:00
  so you know i i definitely credit them with starting that
• 21:04
  philosophy within us that
• 21:09
  part of the meaning of life is the opportunity to help other
• 21:12
  people and to raise
• 21:13
  SPEAKER_00: them up as well.
• 21:14
  And I think that that has always been something I've really
• 21:17
  loved about the ethos
• 21:19
  SPEAKER_00: of Silicon Valley and tech where it's the opposite of a zero sum game,
• 21:23
  SPEAKER_00: right?
• 21:23
  Like literally the act of building a startup is the act of
• 21:27
  going from zero to one
• 21:29
  SPEAKER_00: of creation in some sense.
• 21:31
  SPEAKER_00: So I think that's always been a part of
• 21:35
  SPEAKER_00: my core life philosophy.
• 21:38
  SPEAKER_00: And,
• 21:38
  SPEAKER_00: you know,
• 21:38
  SPEAKER_00: obviously like through my own experiences,
• 21:40
  SPEAKER_00: just being able to learn stuff really fast and really enjoying that.
• 21:44
  I was,
• 21:45
  I've just always been really obsessed with the idea of how do I
• 21:47
  scale this,
• 21:47
  right?
• 21:48
  SPEAKER_00: How do I build ways for other people to experience the same thing?
• 21:52
  SPEAKER_00: Okay.
• 21:52
  SPEAKER_00: Because I totally think it's possible.
• 21:54
  SPEAKER_01: Well,
• 21:54
  SPEAKER_01: and this is,
• 21:55
  SPEAKER_01: this is like why I really,
• 21:56
  SPEAKER_01: really wanted you to come onto the show because when,
• 22:00
  SPEAKER_01: um,
• 22:01
  SPEAKER_01: When I look at your brother, it makes more sense to me in some ways.
• 22:08
  SPEAKER_01: Super, super bright.
• 22:11
  SPEAKER_01: could do anything you want.
• 22:14
  Um,
• 22:14
  he,
• 22:15
  you know,
• 22:15
  he's pushing the edges of biology for people who don't know.
• 22:19
  SPEAKER_01: I mean, Ark Institute does a variety of science.
• 22:22
  SPEAKER_01: Um, Patrick is, is doing a lot around gene editing.
• 22:26
  SPEAKER_01: I mean,
• 22:27
  SPEAKER_01: really like helped was designed CRISPR and,
• 22:31
  SPEAKER_01: and,
• 22:31
  SPEAKER_00: you know,
• 22:31
  SPEAKER_01: I mean like,
• 22:32
  SPEAKER_01: yeah,
• 22:32
  SPEAKER_01: like cutting edge stuff.
• 22:34
  SPEAKER_01: And so that's sort of where my brain might be.
• 22:38
  If I had all these talents at my disposal,
• 22:40
  it's like,
• 22:40
  okay,
• 22:40
  I'm going to go cure cancer.
• 22:41
  SPEAKER_01: I guess that's the stereotypical kind of thing.
• 22:45
  SPEAKER_01: Sometimes I think of Patrick and John Collison,
• 22:48
  SPEAKER_01: who I know well,
• 22:50
  SPEAKER_01: and I do not want this at all to sound trite.
• 22:57
  SPEAKER_01: Stripe is incredible.
• 22:58
  SPEAKER_01: They've built this multibillion-dollar company.
• 23:00
  They've not only made wealth for themselves,
• 23:03
  they've made it possible for...
• 23:05
  thousands upon thousands of companies to start and make money
• 23:08
  all around the world.
• 23:09
  SPEAKER_01: I get that.
• 23:09
  SPEAKER_01: They're also interested in so much stuff.
• 23:12
  SPEAKER_01: They take classes on physics, on law and everything.
• 23:16
  SPEAKER_01: They read everything.
• 23:17
  SPEAKER_01: I am sometimes- True renaissance men.
• 23:19
  SPEAKER_00: Right.
• 23:19
  SPEAKER_01: They are.
• 23:20
  And so sometimes I'm surprised that they have stayed in finance
• 23:24
  as long as stayed
• 23:25
  SPEAKER_01: on one thing,
• 23:26
  SPEAKER_01: I suppose,
• 23:27
  SPEAKER_01: as long as they have.
• 23:28
  SPEAKER_02: Yeah.
• 23:28
  You're kind of heading on this path where,
• 23:31
  you know,
• 23:31
  you look like you might be
• 23:33
  SPEAKER_01: Unlocking the secrets of the mind, maybe.
• 23:36
  SPEAKER_01: And then when you take the startup route, now you've done two.
• 23:40
  SPEAKER_01: So the first one is Aerie Labs?
• 23:42
  SPEAKER_00: Yeah, that was the startup that I started.
• 23:44
  SPEAKER_00: I mean,
• 23:45
  it was the first company that I started right after dropping
• 23:47
  out of grad school
• 23:48
  SPEAKER_00: when I started the Tool Fellowship.
• 23:49
  SPEAKER_01: And around education?
• 23:51
  SPEAKER_00: It was kind of- It was educational games.
• 23:53
  SPEAKER_00: Educational games.
• 23:54
  SPEAKER_00: So that was my first failed startup, I would say.
• 23:57
  Okay.
• 23:58
  SPEAKER_01: And then, okay, so this is my, you know, it's a long-winded question.
• 24:02
  SPEAKER_01: But yeah, I mean, you could have done anything.
• 24:04
  I know you talked about sort of how you wanted to help people
• 24:08
  and how you saw it.
• 24:09
  But when I was hearing about your story,
• 24:12
  I was kind of like,
• 24:13
  man,
• 24:14
  I don't know if I had...
• 24:17
  SPEAKER_01: all these choices,
• 24:18
  I think I would probably be inclined to try to cure cancer or
• 24:22
  get into unlocking
• 24:24
  SPEAKER_01: the mysteries of the mind or something like that.
• 24:26
  SPEAKER_00: Yeah.
• 24:26
  SPEAKER_00: Well,
• 24:26
  SPEAKER_00: I mean,
• 24:27
  I would argue that the promise of software and of technology is
• 24:34
  that it allows us
• 24:36
  SPEAKER_00: to build systems that a billion people can use,
• 24:39
  SPEAKER_00: right?
• 24:39
  SPEAKER_00: That's kind of unprecedented.
• 24:41
  SPEAKER_00: So you could argue that if you're a doctor,
• 24:43
  SPEAKER_00: for example,
• 24:44
  SPEAKER_00: you can help one person at a time in a single surgery.
• 24:47
  SPEAKER_00: I'm not saying I'm a utilitarian,
• 24:50
  but I think the promise of where I want to take things is
• 24:57
  really using the power of
• 24:59
  AI and these types of models to really unlock this 100x speed
• 25:06
  for everyone on Earth
• 25:08
  SPEAKER_00: to learn things.
• 25:08
  SPEAKER_00: And I think that is...
• 25:13
  I think a really profound goal and sort of inevitable
• 25:17
  consequence of the AI revolution.
• 25:20
  SPEAKER_00: So that's kind of where I see like the ultimate, ultimate goal.
• 25:25
  SPEAKER_01: So you feel your quest is to unlock human potential?
• 25:30
  SPEAKER_01: Is that?
• 25:31
  SPEAKER_00: That's one way of putting it.
• 25:31
  SPEAKER_00: Yeah.
• 25:32
  SPEAKER_00: I mean,
• 25:33
  SPEAKER_00: you know,
• 25:33
  SPEAKER_00: like maybe in a less grandiose form,
• 25:38
  it would just be to really help everyone learn things a hundred
• 25:41
  times faster.
• 25:42
  SPEAKER_00: Okay.
• 25:43
  SPEAKER_01: Okay.
• 25:43
  SPEAKER_00: Because I think that's how you lift people up.
• 25:46
  SPEAKER_00: I think that's how people change their lives.
• 25:49
  SPEAKER_01: I want to get into this.
• 25:50
  I'm going to ask this one now because there's other questions I
• 25:53
  have from what you just said.
• 25:54
  SPEAKER_01: I was reading some description in here.
• 26:00
  SPEAKER_01: In this interview I saw you did, you had this very methodical...
• 26:06
  SPEAKER_01: way of learning,
• 26:07
  SPEAKER_01: I saw you do some quote where you said like,
• 26:09
  SPEAKER_01: yeah,
• 26:09
  SPEAKER_01: I mean,
• 26:09
  SPEAKER_01: maybe some of this is genetics,
• 26:11
  SPEAKER_01: but a lot of this was just,
• 26:12
  SPEAKER_01: was focus and rigor and,
• 26:15
  SPEAKER_01: and goal setting.
• 26:16
  SPEAKER_01: Um, it reminded me, I interviewed Lowell Wood one time.
• 26:20
  SPEAKER_01: I don't know if you know.
• 26:20
  SPEAKER_01: No.
• 26:21
  So he's,
• 26:21
  he,
• 26:22
  a couple of years ago surpassed Thomas Edison as the US leading
• 26:26
  patent holder.
• 26:27
  SPEAKER_01: Oh, yeah.
• 26:28
  SPEAKER_01: Yeah.
• 26:28
  He's,
• 26:29
  he's,
• 26:29
  he's older and,
• 26:30
  um,
• 26:32
  worked at Lawrence Livermore labs for many years,
• 26:34
  um,
• 26:34
  did
• 26:36
  SPEAKER_01: He's amazing.
• 26:37
  SPEAKER_01: He knows like chemistry, you know, did a lot of nuclear.
• 26:42
  SPEAKER_01: He knows a lot about everything.
• 26:45
  SPEAKER_01: I was sitting with him one time and he's like,
• 26:47
  SPEAKER_01: uh,
• 26:47
  SPEAKER_01: you know,
• 26:47
  SPEAKER_01: I'm not that much smarter than other people.
• 26:50
  SPEAKER_01: And,
• 26:50
  SPEAKER_01: and it's just,
• 26:51
  SPEAKER_01: when I read,
• 26:52
  SPEAKER_01: I just remember everything I read,
• 26:54
  SPEAKER_00: but he said,
• 26:54
  SPEAKER_01: he's like,
• 26:55
  SPEAKER_01: anyone can do that.
• 26:57
  SPEAKER_01: You know,
• 26:57
  SPEAKER_01: you just have to,
• 26:58
  you have to be disciplined and you have to train yourself for
• 27:01
  this retention.
• 27:01
  SPEAKER_01: And I, I
• 27:03
  SPEAKER_01: I'm just, I'm like not convinced that's true.
• 27:05
  SPEAKER_01: And did I get that quote?
• 27:07
  Like you're thinking on,
• 27:09
  um,
• 27:10
  so you see some genetics in this,
• 27:13
  but you also see,
• 27:15
  I don't know.
• 27:16
  I understand that might sound ridiculous coming from somebody
• 27:19
  who went to college
• 27:20
  SPEAKER_00: at 12 for sure.
• 27:21
  SPEAKER_00: But I do, um, I am pretty convinced that, um,
• 27:28
  everyone can learn things 100x faster right there there are
• 27:32
  ways of explaining
• 27:35
  concepts to people and you know like i think something i'm good
• 27:38
  at personally is
• 27:40
  explaining a concept or you know some some sort of idea in a
• 27:44
  way that is easy to
• 27:47
  understand and digest and to do that is not easy right that is
• 27:52
  a skill like you
• 27:52
  SPEAKER_00: need to
• 27:54
  SPEAKER_00: Absolutely.
• 27:54
  Sort of have like a full mental model of the person that you're
• 27:56
  talking to and the
• 27:57
  SPEAKER_00: person that you're teaching or tutoring.
• 28:00
  SPEAKER_00: But I do think that it's something that is scalable,
• 28:04
  SPEAKER_00: that everyone can do,
• 28:06
  SPEAKER_00: and that is solvable through technology.
• 28:08
  SPEAKER_01: And how did Aerie Labs try to do that?
• 28:13
  SPEAKER_00: So the approach originally there was it was focused on kids,
• 28:17
  SPEAKER_00: which turned out to be just a really hard business.
• 28:20
  SPEAKER_00: And it was essentially more focused on early education,
• 28:26
  SPEAKER_00: developing basic sort of math,
• 28:28
  SPEAKER_00: reading,
• 28:29
  SPEAKER_00: writing,
• 28:29
  SPEAKER_00: grammar skills.
• 28:31
  SPEAKER_00: You know, it was my first time in Silicon Valley out of grad school.
• 28:37
  SPEAKER_00: Young founder, didn't really know what I was doing.
• 28:38
  SPEAKER_00: So that never really went anywhere.
• 28:40
  And there were a few years afterwards where I worked on sort of
• 28:46
  side projects that
• 28:48
  SPEAKER_00: may or may not have turned into a startup.
• 28:51
  SPEAKER_00: They were all sort of in the same vein,
• 28:54
  SPEAKER_00: roughly though,
• 28:55
  of let's use the power and leverage of software to build
• 28:59
  systems that teach people
• 29:01
  SPEAKER_00: things faster.
• 29:02
  SPEAKER_00: that provide leverage in that way.
• 29:04
  Um,
• 29:04
  and I think like,
• 29:05
  you know,
• 29:06
  arguably you can categorize speak like that as well.
• 29:10
  SPEAKER_01: Yeah.
• 29:10
  SPEAKER_01: And,
• 29:10
  SPEAKER_01: and just before we get to speak,
• 29:12
  SPEAKER_01: the,
• 29:14
  SPEAKER_01: this has been a obsession of Silicon Valley,
• 29:16
  SPEAKER_01: especially so Aries starts in 2011.
• 29:18
  SPEAKER_01: Yeah.
• 29:19
  SPEAKER_01: And then I'd say it was,
• 29:21
  SPEAKER_01: it was around that time or a little bit later,
• 29:23
  SPEAKER_01: we had things like,
• 29:24
  SPEAKER_01: um,
• 29:25
  SPEAKER_01: old school,
• 29:26
  SPEAKER_01: um,
• 29:26
  SPEAKER_01: just tons of education,
• 29:27
  SPEAKER_01: um,
• 29:29
  SPEAKER_01: In Silicon Valley, we get dinged for this, right?
• 29:31
  SPEAKER_01: People would be like, oh, you're trying to fix education with technology.
• 29:34
  SPEAKER_01: And like, you guys don't understand.
• 29:36
  SPEAKER_01: This is a much softer, more nuanced thing.
• 29:39
  SPEAKER_01: And all this money got spent.
• 29:40
  SPEAKER_01: I had a son who went to old school.
• 29:42
  SPEAKER_01: It was definitely a mixed experience.
• 29:46
  SPEAKER_01: It didn't end well.
• 29:47
  SPEAKER_01: And all this hope, right?
• 29:50
  SPEAKER_01: Because there's a lot of this that makes sense.
• 29:51
  SPEAKER_01: It's like, oh, of course we... Old school is a little different.
• 29:53
  SPEAKER_01: It was like, we'll apply...
• 29:56
  SPEAKER_01: technology to the infrastructure part.
• 29:58
  Schools shouldn't have to all spend individual money on all the
• 30:01
  software in the back.
• 30:02
  SPEAKER_01: We'll do this really well and efficiently,
• 30:04
  SPEAKER_01: and we'll do,
• 30:05
  SPEAKER_01: um,
• 30:07
  cameras in the classroom to see the actual interactions between
• 30:10
  the teachers and
• 30:11
  SPEAKER_01: kids.
• 30:11
  SPEAKER_01: We'll actually monitor their grades and see if you're,
• 30:14
  SPEAKER_01: there were things that made a lot of sense.
• 30:15
  SPEAKER_01: It didn't sort of work out.
• 30:16
  SPEAKER_01: I saw when I was reading, it looked like you guys
• 30:19
  SPEAKER_01: at the end got caught up in the whole,
• 30:22
  SPEAKER_01: you know,
• 30:22
  because there was a backlash and then when these companies
• 30:25
  didn't do well,
• 30:26
  SPEAKER_01: everyone like shat on them and was like,
• 30:28
  SPEAKER_01: see,
• 30:28
  SPEAKER_01: we told you.
• 30:29
  SPEAKER_01: And it seemed like you guys got caught up in that.
• 30:33
  SPEAKER_01: Totally, painfully obvious question.
• 30:36
  SPEAKER_01: I mean,
• 30:36
  SPEAKER_01: yeah,
• 30:36
  SPEAKER_01: as somebody who had had this string of life success and everything,
• 30:42
  SPEAKER_01: yeah,
• 30:42
  SPEAKER_01: how did you deal with that?
• 30:43
  SPEAKER_00: Yeah, it was painful for sure.
• 30:45
  SPEAKER_00: It sucked.
• 30:46
  I think just like the public spotlight and the expectations as
• 30:51
  a Teal Fellow.
• 30:51
  But I think also I felt in my heart that it was just like a
• 31:02
  blip on the journey in that
• 31:05
  you know,
• 31:05
  I knew what I wanted to do with my life and I knew what I stood
• 31:08
  for.
• 31:09
  SPEAKER_00: And, you know, I was like onto the next thing.
• 31:12
  SPEAKER_00: So yeah, definitely, definitely sucked.
• 31:17
  SPEAKER_00: But I think that
• 31:19
  SPEAKER_00: You know,
• 31:19
  it's the sort of thing where you just kind of like stand up
• 31:21
  after and then keep
• 31:22
  SPEAKER_00: walking.
• 31:23
  SPEAKER_00: Yeah.
• 31:23
  SPEAKER_00: Yeah.
• 31:23
  SPEAKER_01: I mean, what else can you do?
• 31:26
  I think I've always only followed my own internal clock and
• 31:32
  internal standards for myself.
• 31:35
  So anything external is frankly just not like if somebody
• 31:39
  doesn't like what I'm doing or is...
• 31:42
  SPEAKER_00: Okay, that's a bad example.
• 31:43
  But like external standards are kind of irrelevant to the way
• 31:47
  that I think about
• 31:48
  SPEAKER_00: the world,
• 31:48
  SPEAKER_00: if that makes sense.
• 31:49
  SPEAKER_00: So it's like I go by my own internal measuring tape.
• 31:55
  SPEAKER_00: Yeah.
• 31:55
  SPEAKER_01: It does make sense.
• 31:56
  SPEAKER_01: Again, this is why I wanted to talk to you.
• 31:57
  SPEAKER_01: When I looked at your biography,
• 31:58
  SPEAKER_01: I'm like,
• 31:59
  SPEAKER_01: oh,
• 31:59
  SPEAKER_01: this dude,
• 31:59
  SPEAKER_01: this dude is,
• 32:01
  SPEAKER_01: I think people could be like,
• 32:02
  SPEAKER_01: oh,
• 32:03
  SPEAKER_01: yeah.
• 32:05
  you're tiger parents and you've been programmed to go after
• 32:08
  life some particular way.
• 32:09
  SPEAKER_01: And then when I looked at your biography,
• 32:10
  SPEAKER_01: it's just like,
• 32:11
  SPEAKER_01: oh,
• 32:12
  SPEAKER_01: you made a bunch of like courageous,
• 32:13
  SPEAKER_01: hard decisions,
• 32:14
  SPEAKER_01: I think.
• 32:15
  SPEAKER_01: And it seems clear that you're chasing whatever you're into.
• 32:18
  SPEAKER_01: So that's, I found this like really, I think it's hard to do.
• 32:22
  SPEAKER_01: I find this super fascinating.
• 32:24
  SPEAKER_00: Yeah, it definitely took some big risks.
• 32:25
  I think,
• 32:26
  you know,
• 32:26
  like maybe you can argue they were impulsive,
• 32:29
  but I've always found that,
• 32:31
  um,
• 32:34
  SPEAKER_00: I pretty strongly believe in just following my intuition and my gut,
• 32:39
  SPEAKER_00: and it's always worked out so far.
• 32:43
  SPEAKER_01: Okay.
• 32:44
  SPEAKER_01: So you said after the startup was struggling,
• 32:48
  SPEAKER_01: you were trying,
• 32:49
  SPEAKER_01: you had other ideas bouncing around in your head.
• 32:51
  And so tell me when Speak emerges up to the top and how it
• 32:57
  really got you.
• 32:57
  SPEAKER_00: Yeah.
• 32:58
  So the sort of short story and we're kind of breezing over
• 33:03
  several years here,
• 33:04
  but
• 33:05
  SPEAKER_00: Through the Teal Fellowship,
• 33:06
  SPEAKER_00: actually,
• 33:07
  SPEAKER_00: the first few years of the fellowship,
• 33:10
  SPEAKER_00: I met many of my closest friends today.
• 33:12
  And it was just an awesome social circle and sort of
• 33:18
  introduction to the world of
• 33:20
  SPEAKER_00: technology,
• 33:20
  SPEAKER_00: I suppose.
• 33:22
  SPEAKER_00: So we'll always be indebted
• 33:25
  SPEAKER_00: to the fellowship for that.
• 33:27
  SPEAKER_00: And lived in Mountain View and Palo Alto for a while.
• 33:32
  And then I moved up with a bunch of friends to a group house in
• 33:35
  San Francisco on
• 33:36
  SPEAKER_00: Fell Street on the Panhandle.
• 33:37
  SPEAKER_00: It's always the group house.
• 33:38
  SPEAKER_00: Yes.
• 33:39
  SPEAKER_00: And so this...
• 33:41
  SPEAKER_00: It was kind of like...
• 33:44
  I hesitate to call it a hacker house because I feel like it was
• 33:46
  a little bit sort
• 33:47
  SPEAKER_00: of cleaner and more mature than that.
• 33:51
  SPEAKER_00: But it was a bunch of tool follows.
• 33:53
  SPEAKER_00: It was a bunch of our friends.
• 33:55
  SPEAKER_00: It was a great...
• 33:56
  built-in social circle um we would just leave the front door
• 34:00
  open and we had just
• 34:01
  like 50 friends or so that would just often come and hang out
• 34:06
  in in the living room
• 34:08
  um are there folks we know in that cohort um well so my
• 34:12
  co-founder connor uh our
• 34:15
  coo colton um you may know delian from founders fund oh he was
• 34:19
  yeah yeah he was in
• 34:20
  SPEAKER_01: that okay
• 34:20
  SPEAKER_00: Yeah, yeah, yeah.
• 34:21
  SPEAKER_00: They're, you know, like several other people as well.
• 34:24
  SPEAKER_00: So all, some of my closest friends today.
• 34:27
  SPEAKER_00: It was a lot of fun.
• 34:28
  SPEAKER_00: I think it was almost like, like it felt like almost...
• 34:32
  SPEAKER_00: my true college experience, if you will.
• 34:35
  SPEAKER_00: Um,
• 34:35
  SPEAKER_00: and yeah,
• 34:36
  like we all kind of came up together in our early twenties and
• 34:39
  we're working on
• 34:40
  SPEAKER_00: startups and,
• 34:40
  SPEAKER_00: um,
• 34:41
  we started speak out of the third floor of that house and our
• 34:45
  wifi password,
• 34:46
  SPEAKER_00: uh,
• 34:47
  SPEAKER_00: is still 1864 file,
• 34:48
  SPEAKER_00: which was the address of that house.
• 34:51
  SPEAKER_01: And you said I skipped over a few years.
• 34:53
  SPEAKER_01: Um, yeah.
• 34:55
  SPEAKER_01: What was going on in that time?
• 34:56
  SPEAKER_00: We were,
• 34:56
  SPEAKER_00: uh,
• 34:57
  SPEAKER_00: well,
• 34:57
  SPEAKER_00: we,
• 34:57
  SPEAKER_00: we were living in the house and,
• 35:00
  SPEAKER_00: um,
• 35:00
  SPEAKER_00: I was working on a few other startup efforts,
• 35:03
  SPEAKER_00: built some products,
• 35:04
  SPEAKER_00: nothing really clicked in the market,
• 35:06
  but I think like all this time I was sort of gaining experience
• 35:10
  as an entrepreneur
• 35:13
  SPEAKER_00: and in thinking about the world and about markets and,
• 35:17
  SPEAKER_00: and,
• 35:17
  SPEAKER_00: you know,
• 35:17
  SPEAKER_00: what I,
• 35:17
  SPEAKER_00: what,
• 35:18
  SPEAKER_00: what I wanted to build.
• 35:19
  SPEAKER_00: And, um,
• 35:21
  SPEAKER_00: Essentially,
• 35:21
  SPEAKER_00: the Speak story is that in 2016,
• 35:23
  SPEAKER_00: I was talking with my now co-founder,
• 35:28
  SPEAKER_00: Connor,
• 35:30
  SPEAKER_00: who was one of my best friends,
• 35:31
  SPEAKER_00: and we were just
• 35:35
  SPEAKER_00: looking at some recent papers that came out, deep speech, right?
• 35:38
  It's the new deep learning speech recognition model from Baidu
• 35:42
  that seems like it's
• 35:43
  SPEAKER_00: starting to work.
• 35:44
  SPEAKER_00: And this is when deep learning winter was starting to thaw.
• 35:48
  SPEAKER_00: And it was just very obvious to us initially that
• 35:52
  SPEAKER_00: if this stuff keeps working this way,
• 35:54
  SPEAKER_00: then it clearly will be the dominant technology of the future.
• 35:57
  And we were talking about,
• 35:59
  okay,
• 35:59
  what do we want to be doing with our lives in five years?
• 36:02
  We were kind of in between things,
• 36:04
  a little bit lost,
• 36:05
  just,
• 36:05
  you know,
• 36:06
  trying to figure it out.
• 36:07
  SPEAKER_00: And, um,
• 36:09
  We both immediately had the same concept,
• 36:13
  which was we want to work with AI somehow.
• 36:17
  And we think that there is huge potential in building products
• 36:21
  that AI uniquely
• 36:24
  SPEAKER_00: unlocks or makes possible or makes 10x better.
• 36:28
  And it would be cool to be running a product lab together or
• 36:32
  some sort of AI product company.
• 36:36
  SPEAKER_00: And we decided to take a sabbatical and basically
• 36:39
  SPEAKER_00: drop all the companies we were doing and just study deep learning.
• 36:43
  SPEAKER_00: And we went pretty deep,
• 36:45
  SPEAKER_00: read a lot of papers,
• 36:46
  SPEAKER_00: wrote a bunch of code for a bunch of models,
• 36:49
  SPEAKER_00: snuck into a bunch of classes at Stanford and Berkeley.
• 36:54
  SPEAKER_00: And through that,
• 36:57
  SPEAKER_00: Speak started.
• 36:58
  SPEAKER_00: That was kind of the,
• 36:59
  SPEAKER_00: we considered a bunch of ideas,
• 37:01
  but I think it was clear to us that language learning was super
• 37:06
  universal.
• 37:07
  Everyone struggles with fluency and most people you know want
• 37:10
  to learn another language.
• 37:13
  SPEAKER_00: It's very commercially valuable.
• 37:14
  SPEAKER_00: If you're serious about fluency,
• 37:16
  you're hiring a human tutor and the whole market is like 50 to
• 37:21
  $100 billion of
• 37:22
  SPEAKER_00: spend a year around the world.
• 37:24
  SPEAKER_00: And uniquely it's very human based, but AI is going to change that.
• 37:29
  SPEAKER_00: And we,
• 37:31
  SPEAKER_00: like I mentioned earlier,
• 37:32
  SPEAKER_00: had this prediction that,
• 37:33
  SPEAKER_00: okay,
• 37:33
  SPEAKER_00: speech models are going like this language models are going like this.
• 37:36
  SPEAKER_00: And remember this is 2017, right?
• 37:39
  SPEAKER_00: So we felt like,
• 37:41
  SPEAKER_00: okay,
• 37:41
  SPEAKER_00: in five or 10 years,
• 37:43
  SPEAKER_00: they're going to take over the world,
• 37:44
  SPEAKER_00: obviously,
• 37:45
  SPEAKER_00: right?
• 37:45
  SPEAKER_00: They're going to become superhuman.
• 37:47
  SPEAKER_00: It will let us build products that can act like humans.
• 37:52
  SPEAKER_00: And, um,
• 37:55
  SPEAKER_00: Specifically,
• 37:56
  SPEAKER_00: basically like in our sabbatical,
• 37:57
  we had built an accent discrimination model for fun as like a
• 38:01
  toy problem.
• 38:02
  SPEAKER_00: But that was kind of the genesis of the idea that,
• 38:05
  SPEAKER_00: hey,
• 38:05
  we can build some sort of model that can understand non-native
• 38:09
  speakers and give
• 38:09
  SPEAKER_00: them feedback.
• 38:11
  SPEAKER_00: And previously, you needed to hire a human to do that.
• 38:14
  SPEAKER_00: So, yeah, that's how it started.
• 38:16
  SPEAKER_01: Well,
• 38:16
  SPEAKER_01: and I do remember image recognition started to kind of work,
• 38:20
  and then translation was the first big thing that I really
• 38:25
  remember coming out of
• 38:26
  SPEAKER_01: Google and Microsoft as like,
• 38:28
  SPEAKER_01: oh,
• 38:28
  SPEAKER_01: okay,
• 38:28
  SPEAKER_01: this machine learning stuff is kicking off.
• 38:30
  SPEAKER_01: But definitely no LLM things.
• 38:33
  SPEAKER_00: Yeah, that wasn't until 2022.
• 38:34
  SPEAKER_00: Right.
• 38:34
  SPEAKER_00: Yeah.
• 38:35
  SPEAKER_01: And when you are doing this, you're reading all these papers.
• 38:39
  SPEAKER_01: I know clearly you've identified this need.
• 38:43
  SPEAKER_01: You see the basis for a company, but OpenAI is just forming then.
• 38:47
  SPEAKER_01: DeepMind has been around.
• 38:48
  Yeah,
• 38:50
  you didn't think,
• 38:52
  I want to go work on the really foundational stuff of this.
• 38:57
  Yeah,
• 38:57
  we wanted to build products that people could use,
• 39:00
  not do fundamental research.
• 39:02
  SPEAKER_00: And I think...
• 39:04
  SPEAKER_00: We were,
• 39:06
  SPEAKER_00: you know,
• 39:06
  SPEAKER_00: this is kind of a repeated pattern for ML startups,
• 39:09
  SPEAKER_00: but we were excited to work on ML problems.
• 39:13
  And the reality was that for the first few years,
• 39:15
  we were struggling to find product market fit.
• 39:18
  SPEAKER_00: And we were just building different versions of the Speak app.
• 39:22
  SPEAKER_00: And it felt like until we basically solved this very existential issue,
• 39:28
  SPEAKER_00: We couldn't do what we really wanted to do, which was more ML research.
• 39:33
  SPEAKER_00: But yeah, it took a while.
• 39:35
  SPEAKER_00: I would say 2017 through 2019, we launched three times.
• 39:43
  SPEAKER_00: Nothing was really working,
• 39:45
  but we just had such a sort of strong view of this future that
• 39:50
  we knew would come
• 39:51
  SPEAKER_00: to pass.
• 39:52
  SPEAKER_00: that we never pivoted.
• 39:54
  We just kept going and we made the decision to essentially like
• 39:59
  relaunch the modern
• 40:00
  SPEAKER_00: version of speak,
• 40:01
  SPEAKER_00: but really focus it.
• 40:02
  SPEAKER_00: So we chose South Korea as our first market.
• 40:05
  SPEAKER_00: That was like a critical decision.
• 40:07
  SPEAKER_00: Yeah.
• 40:07
  SPEAKER_01: So there's a lot of people in South Korea that want to learn English.
• 40:10
  SPEAKER_00: Yeah, all over Asia, but South Korea was particularly crazy.
• 40:13
  SPEAKER_00: And we basically felt like,
• 40:16
  SPEAKER_00: okay,
• 40:17
  SPEAKER_00: people here are really serious about English and it's all human based.
• 40:22
  SPEAKER_00: So they're paying for private tutors.
• 40:24
  There are skyscrapers full of English classrooms that people
• 40:27
  schedule and then they go to.
• 40:28
  If we can win here in some sense and build something that
• 40:32
  starts to become popular
• 40:33
  SPEAKER_00: and is meaningfully different,
• 40:35
  SPEAKER_00: then that's a strong signal.
• 40:37
  SPEAKER_00: Like we chose that market in a sense because it was a hard market.
• 40:40
  SPEAKER_01: So this wasn't the original intuition.
• 40:42
  How did you,
• 40:45
  you came to realize that over a couple of years that particular
• 40:48
  market existed or?
• 40:52
  SPEAKER_00: So there were a bunch of things.
• 40:53
  SPEAKER_00: I would say that in terms of vision,
• 40:55
  it was the vision of using speech recognition models and
• 41:00
  language models to build a
• 41:03
  SPEAKER_00: better way to reach fluency.
• 41:05
  SPEAKER_00: We were limited by the technology and what we could do initially.
• 41:08
  And we were spending a lot of energy on finding PMF through
• 41:12
  just testing many
• 41:13
  SPEAKER_00: versions of the product,
• 41:14
  SPEAKER_00: right?
• 41:15
  SPEAKER_00: So,
• 41:15
  SPEAKER_00: you know,
• 41:17
  there were a lot of pretty specific learnings that we put into
• 41:20
  Modern Speak from
• 41:21
  SPEAKER_00: the past few years of iterations,
• 41:23
  SPEAKER_00: like how the course worked,
• 41:25
  SPEAKER_00: you know,
• 41:25
  SPEAKER_00: how the content worked,
• 41:27
  SPEAKER_00: what types of actors we should hire for,
• 41:29
  SPEAKER_00: you know,
• 41:30
  SPEAKER_00: for the teaching,
• 41:31
  SPEAKER_00: et cetera,
• 41:31
  SPEAKER_00: et cetera,
• 41:31
  SPEAKER_00: right?
• 41:32
  SPEAKER_00: Like hundreds of things.
• 41:34
  SPEAKER_01: Okay.
• 41:34
  SPEAKER_01: And then, and then it takes off like a rocket.
• 41:39
  SPEAKER_01: Doesn't it after that?
• 41:40
  SPEAKER_00: Yeah.
• 41:40
  SPEAKER_00: Yeah.
• 41:41
  SPEAKER_00: Like it felt like we really picked the lock in some sense.
• 41:45
  Um,
• 41:45
  and the first version of the app that started to work wasn't
• 41:49
  even that great,
• 41:50
  but it was.
• 41:52
  more of the supplemental speaking practice tool where we had
• 41:56
  all this content as a
• 41:57
  user it was kind of an outlet to speak and I think that people
• 42:03
  had been just really
• 42:05
  wanting an outlet to speak and actually make the sounds and get
• 42:09
  some sort of
• 42:10
  SPEAKER_00: immediate feedback even if it wasn't
• 42:12
  SPEAKER_00: like all the feedback that we had dreamed of in the beginning.
• 42:15
  SPEAKER_00: Right.
• 42:16
  SPEAKER_00: Um, so yeah, that, that started to grow quite, you know, quite quickly.
• 42:21
  SPEAKER_00: And then, um, in 2022, that was the LM moment.
• 42:25
  And that was kind of the like first five or 10 year thing that,
• 42:29
  that we had predicted.
• 42:30
  SPEAKER_00: Okay.
• 42:30
  SPEAKER_00: And that was when we were like, oh shit, it's here.
• 42:33
  And I have the stats here somewhere,
• 42:35
  but you guys just raised some money.
• 42:37
  SPEAKER_01: Last November.
• 42:38
  SPEAKER_00: Last November.
• 42:39
  SPEAKER_01: And you're a unicorn.
• 42:41
  SPEAKER_01: You've unicorned.
• 42:43
  SPEAKER_01: And how many people use Speak?
• 42:45
  SPEAKER_00: Yeah, so we have over 10 million downloads.
• 42:52
  SPEAKER_00: We're the biggest English app in South Korea at this point.
• 42:55
  SPEAKER_00: Things have grown from...
• 42:56
  SPEAKER_00: It was really cool to see it go from obviously nothing to,
• 43:02
  SPEAKER_00: oh,
• 43:03
  SPEAKER_00: when we visit Seoul,
• 43:04
  SPEAKER_00: some people would be using it on the street,
• 43:06
  SPEAKER_00: and we'd see that,
• 43:07
  SPEAKER_00: to now our teachers in the app are minor celebrities,
• 43:12
  SPEAKER_00: and people will go up to them on the street and say hi.
• 43:15
  SPEAKER_00: And we do giant influencer commercials with the most famous people,
• 43:19
  SPEAKER_00: and billboards,
• 43:20
  SPEAKER_00: and TV commercials,
• 43:21
  SPEAKER_00: and really everything.
• 43:23
  SPEAKER_00: So yeah,
• 43:24
  SPEAKER_00: now we have a decent amount of scale and are breaking out,
• 43:30
  SPEAKER_00: teaching English in many more countries in the US as well.
• 43:33
  SPEAKER_01: What other countries have really taken off?
• 43:37
  So there's a bunch of other Asian countries that were the first
• 43:40
  ones that we went to.
• 43:41
  SPEAKER_00: So we're growing really fast.
• 43:43
  SPEAKER_00: Japan, Taiwan, a bunch of others.
• 43:46
  SPEAKER_00: US is growing really fast as well.
• 43:49
  SPEAKER_00: There's people that are interested in learning English in the US.
• 43:53
  SPEAKER_00: ESL.
• 43:54
  And then there's native English speakers that want to learn
• 43:58
  Spanish and French,
• 43:59
  SPEAKER_00: Chinese,
• 43:59
  SPEAKER_00: Japanese,
• 44:00
  SPEAKER_00: Korean,
• 44:00
  SPEAKER_00: et cetera.
• 44:01
  SPEAKER_00: And we're launching all of that this year, like very shortly.
• 44:03
  SPEAKER_01: And you guys have always emphasized speaking more.
• 44:06
  SPEAKER_01: I mean, I see people on Duolingo all the time and then
• 44:11
  SPEAKER_01: They always want to, the gamification thing works really well for them.
• 44:14
  I know my mom stresses out if she's not hitting her streak and
• 44:18
  God bless her.
• 44:19
  SPEAKER_01: I don't want to undermine her quest to learn Spanish,
• 44:22
  SPEAKER_01: but I feel like she's been doing dueling rope for,
• 44:25
  SPEAKER_01: I want to say seven years or something and has made progress,
• 44:29
  SPEAKER_01: but not,
• 44:31
  SPEAKER_01: she could do with some help.
• 44:33
  SPEAKER_00: Yeah.
• 44:33
  SPEAKER_00: Yeah.
• 44:34
  SPEAKER_00: You know,
• 44:35
  the way that we have always benchmarked our product and
• 44:40
  conceptualized what the
• 44:41
  SPEAKER_00: product should be and where we're going is against humans,
• 44:44
  SPEAKER_00: right?
• 44:44
  So the thought experiment is really think about what the single
• 44:48
  best human English
• 44:50
  shooter or Spanish shooter or Japanese shooter or whatever on
• 44:53
  earth could do.
• 44:55
  Let's build that,
• 44:55
  put it on your phone,
• 44:56
  scale it to a billion people,
• 44:58
  right?
• 44:58
  SPEAKER_00: So it is really a sort of experience that was not possible before AI.
• 45:04
  SPEAKER_00: And therefore,
• 45:06
  SPEAKER_00: we kind of see it as the next generation of language learning where,
• 45:10
  SPEAKER_00: you know,
• 45:11
  SPEAKER_00: one framing is,
• 45:12
  SPEAKER_00: do you remember Rosetta Stone?
• 45:14
  SPEAKER_00: So maybe call that Gen 1 of language learning, right?
• 45:17
  SPEAKER_00: CDs, airport kiosks.
• 45:19
  SPEAKER_01: Hemsler.
• 45:19
  SPEAKER_00: Yeah.
• 45:20
  SPEAKER_00: Yeah.
• 45:22
  SPEAKER_00: Gen 2 was associated with the mobile era.
• 45:25
  SPEAKER_00: So apps like Duolingo,
• 45:26
  just massively popular but also very casual language learning
• 45:30
  apps that are
• 45:33
  SPEAKER_00: extremely well gamified,
• 45:34
  SPEAKER_00: super engaging.
• 45:36
  SPEAKER_00: And now Gen 3 in the AI era
• 45:42
  SPEAKER_00: is products like Speak,
• 45:43
  SPEAKER_00: and we're the category leader here,
• 45:45
  SPEAKER_00: where it's focused on functional fluency,
• 45:48
  SPEAKER_00: on speaking.
• 45:50
  It's almost like a private gym where you can use the muscles in
• 45:55
  your mouth,
• 45:55
  SPEAKER_00: like actually make the sounds.
• 45:57
  SPEAKER_00: It's not embarrassing.
• 46:00
  SPEAKER_00: You're free to make mistakes.
• 46:02
  SPEAKER_00: You get over the confidence hump.
• 46:04
  We have a whole pedagogical method we've developed that is
• 46:07
  highly focused on just
• 46:11
  SPEAKER_00: becoming automatic and on spoken fluency.
• 46:14
  SPEAKER_00: So that's the whole philosophy.
• 46:15
  SPEAKER_00: Yeah.
• 46:16
  SPEAKER_00: So we see it as something that is pretty new.
• 46:19
  SPEAKER_01: I'm sure what I say next will not be novel to you,
• 46:22
  SPEAKER_01: but these are the obvious questions that were tumbling through my head.
• 46:25
  SPEAKER_01: The second ChatGPT started,
• 46:28
  the voice stuff started working and I opened it the first day
• 46:33
  and I was like,
• 46:34
  SPEAKER_01: pretend you're my Spanish tutor.
• 46:37
  SPEAKER_01: I go to Mexico a lot and I was like,
• 46:39
  SPEAKER_01: don't speak to me in a Spain dialect,
• 46:41
  SPEAKER_01: speak to me in a Mexican dialect,
• 46:44
  SPEAKER_01: it killed it.
• 46:44
  SPEAKER_01: And now I'm having a conversation with this app and it's correcting,
• 46:48
  SPEAKER_01: it's telling me how I could do better.
• 46:51
  SPEAKER_01: And I just immediately thought of Duolingo in particular,
• 46:54
  SPEAKER_01: just because I see my mom using it all the time.
• 46:56
  SPEAKER_01: I was like, oh, they're toast, you know, this is over.
• 46:59
  SPEAKER_01: I know you guys have a partnership with OpenAI, but, and again,
• 47:06
  I know you've thought about this,
• 47:07
  but yeah,
• 47:07
  I mean,
• 47:08
  it seems like a curse and a blessing.
• 47:09
  It seems like the underlying technology,
• 47:11
  good,
• 47:12
  helps push your things forward.
• 47:13
  SPEAKER_01: On the other hand,
• 47:14
  these apps themselves seem quite functional in that they're all
• 47:18
  heading to where
• 47:18
  SPEAKER_01: they could be our tutor.
• 47:20
  And I know clearly focus is a good thing and OpenAI is not
• 47:25
  zeroed in on this.
• 47:26
  SPEAKER_01: But yeah,
• 47:27
  SPEAKER_01: I mean,
• 47:27
  it must be a little bit of a shock to the system if we have a
• 47:31
  general purpose AIs
• 47:32
  SPEAKER_01: that can do a lot of this.
• 47:34
  Well,
• 47:35
  so I think the framing here is there's kind of two types of AI
• 47:40
  companies,
• 47:41
  right?
• 47:41
  There's the ones that are really scared when the models get
• 47:45
  better and are worried
• 47:47
  SPEAKER_00: that ChatGPT will eat them.
• 47:49
  And then there's the ones that are excited because the models
• 47:52
  are better.
• 47:53
  And now we can do more things and build this roadmap that we
• 47:57
  had been dreaming of.
• 47:57
  SPEAKER_00: So I would put us into the latter category.
• 48:00
  SPEAKER_00: The way that I view it is the more people that know that AI,
• 48:06
  SPEAKER_00: whether it's a text model or a voice model,
• 48:08
  SPEAKER_00: is good for learning things or for having a conversation in Spanish,
• 48:13
  SPEAKER_00: that's almost like lead gen for us, right?
• 48:15
  SPEAKER_00: That increases awareness.
• 48:17
  SPEAKER_00: And I think that there...
• 48:22
  until superintelligence takes over and the economy's over
• 48:25
  anyway there will always
• 48:26
  be the need for more specialized workflows and tools and you
• 48:32
  know if you think
• 48:33
  about the solution that we're building with speak right the
• 48:35
  voice part where you're
• 48:36
  talking with a you know mexican spanish tutor is one part of it
• 48:41
  right it's a really
• 48:42
  SPEAKER_00: cool part but as soon as you want to go deeper
• 48:45
  SPEAKER_00: You know,
• 48:45
  SPEAKER_00: I'm guessing you had a few conversations,
• 48:47
  SPEAKER_00: but then after that,
• 48:49
  SPEAKER_00: if you want to keep learning,
• 48:50
  SPEAKER_00: you kind of hit a wall.
• 48:51
  SPEAKER_00: You need more systems.
• 48:52
  SPEAKER_00: You need a whole sort of learning operating system.
• 48:54
  SPEAKER_00: That's a lot of what we're building.
• 48:55
  SPEAKER_01: It's sort of frustrating and mentally exhausting,
• 48:59
  SPEAKER_01: really,
• 48:59
  SPEAKER_01: because it is not designed for what I'm trying to do.
• 49:02
  Yeah,
• 49:02
  and it needs to really understand the full state of your
• 49:07
  Spanish knowledge.
• 49:07
  SPEAKER_00: What are the words you know and you don't know?
• 49:09
  SPEAKER_00: What are your goals?
• 49:11
  It should explicitly design drills and practice that are based
• 49:16
  on the types of
• 49:16
  SPEAKER_00: mistakes you've been making over the past week.
• 49:18
  There's all these very sort of hyper-specific,
• 49:22
  vertical,
• 49:23
  domain-specific things.
• 49:24
  SPEAKER_00: products and flows that we're building out.
• 49:27
  SPEAKER_00: And I think that, of course, ChatGPT will continue getting better.
• 49:32
  SPEAKER_00: I think it's just going to take over and win, frankly.
• 49:35
  SPEAKER_00: But, you know, it's a horizontal personal assistant use case.
• 49:38
  SPEAKER_00: I do think they will win that use case.
• 49:40
  SPEAKER_00: But I think the horizontal personal assistant that can do everything.
• 49:45
  But I think that is very different from the best language
• 49:50
  shooter on Earth.
• 49:51
  SPEAKER_01: Okay.
• 49:52
  SPEAKER_01: And Founder Fund and Sam Owen are investors in Speak?
• 49:57
  SPEAKER_01: They are.
• 49:58
  SPEAKER_00: Okay.
• 49:59
  SPEAKER_00: We have a long list of awesome investors.
• 50:01
  SPEAKER_01: I saw that.
• 50:04
  SPEAKER_01: And then OpenAI is an investor or just a partner?
• 50:06
  SPEAKER_00: OpenAI led our series B. Okay.
• 50:08
  SPEAKER_01: So that was cool because I think we both saw...
• 50:15
  SPEAKER_00: in 2022,
• 50:16
  SPEAKER_00: quite immediately,
• 50:17
  that language learning was just a very perfect sort of
• 50:22
  application layer vertical
• 50:23
  SPEAKER_00: for other technology,
• 50:24
  SPEAKER_00: right?
• 50:25
  SPEAKER_00: For the LMs, now the voice models.
• 50:28
  SPEAKER_00: Even early on,
• 50:29
  SPEAKER_00: everyone was worried about,
• 50:31
  SPEAKER_00: oh,
• 50:31
  SPEAKER_00: these LMs hallucinate and it's a huge problem,
• 50:34
  SPEAKER_00: maybe unsolvable,
• 50:35
  SPEAKER_00: right?
• 50:35
  SPEAKER_00: That was like the main thing.
• 50:36
  SPEAKER_00: And we were just like,
• 50:39
  it doesn't matter at all because we're building these role
• 50:42
  plays where you're
• 50:43
  talking with your Uber driver or you're just trying to set up a
• 50:47
  functional scenario
• 50:48
  SPEAKER_00: driven by AI.
• 50:50
  SPEAKER_00: And it doesn't matter if it's not 100% factual.
• 50:53
  SPEAKER_00: That's not really the point, right?
• 50:54
  SPEAKER_00: The point is now,
• 50:56
  SPEAKER_00: you can talk with somebody and get feedback.
• 50:59
  So,
• 50:59
  you know,
• 51:00
  it's just like a weirdly good fit and has been this whole time.
• 51:03
  SPEAKER_01: And when the AI seemed so good at learning languages,
• 51:07
  SPEAKER_01: it seems like they know all languages already.
• 51:10
  SPEAKER_01: And why is it hard for you guys to move country to country?
• 51:15
  Is it because you have to set up all these systems that are
• 51:19
  unique country by
• 51:20
  SPEAKER_01: country and what people are trying to do?
• 51:21
  SPEAKER_00: Yeah, so a big part of the product is the content.
• 51:26
  And we've always done everything in-house and verticalized and
• 51:30
  really seen it as
• 51:31
  SPEAKER_00: super core to the whole experience.
• 51:35
  SPEAKER_00: And it is localized for the native language.
• 51:38
  SPEAKER_00: We take a lot of care in building it.
• 51:41
  So going from teaching English in Korea to teaching English in
• 51:48
  Japan for the first
• 51:49
  SPEAKER_00: time,
• 51:50
  SPEAKER_00: there was like...
• 51:52
  SPEAKER_00: some amount of change, but we're good at that now.
• 51:55
  SPEAKER_00: Now what we're doing is going from teaching English to teaching French,
• 51:58
  SPEAKER_00: teaching Chinese,
• 51:59
  SPEAKER_00: Japanese,
• 51:59
  SPEAKER_00: right?
• 52:00
  SPEAKER_00: And there are sort of new challenges.
• 52:03
  But a lot of the underlying platforms around proficiency
• 52:05
  tracking and so on are the same.
• 52:09
  SPEAKER_01: Again, not crazy, insightful question.
• 52:12
  SPEAKER_01: Couldn't stop it from bouncing around my head, though.
• 52:14
  SPEAKER_01: But you guys are early to AI.
• 52:16
  SPEAKER_01: You make the right bet.
• 52:18
  SPEAKER_01: Comes along.
• 52:19
  SPEAKER_01: It's amazing.
• 52:21
  SPEAKER_01: People talk all the time.
• 52:22
  SPEAKER_01: Maybe we don't even need...
• 52:25
  to learn languages now because we're all going to have our
• 52:27
  we're pretty close to
• 52:29
  having your airpods that just are translating in real time um
• 52:35
  victim of your own
• 52:36
  success type thing so a lot of people ask that um my take here
• 52:41
  is that that's
• 52:42
  SPEAKER_00: actually a completely different thing and what will happen is that
• 52:46
  SPEAKER_00: we will build an awesome real-time translation experience into Speak.
• 52:52
  SPEAKER_00: That'll be integrated into the rest of the experience.
• 52:54
  And if you're an American traveling daily for a quick trip,
• 52:58
  you might use that.
• 53:01
  SPEAKER_00: And then later we'll integrate it into your Italian learning experience.
• 53:07
  SPEAKER_00: That will never eat the core use case of Speak,
• 53:09
  SPEAKER_00: which is that people want to become fluent for connection.
• 53:15
  Our users in Korea want to be able to look me in the eye and
• 53:18
  speak English to me.
• 53:19
  And also there's the sort of like more prosaic logical reason
• 53:26
  of you can't actually
• 53:28
  SPEAKER_00: do real time translation fast enough where it feels good.
• 53:32
  Like for example,
• 53:33
  in German,
• 53:34
  the verb is at the end of the sentence,
• 53:35
  right?
• 53:36
  So logically you can't translate a German sentence into English
• 53:39
  until you've heard
• 53:40
  SPEAKER_00: the whole sentence.
• 53:41
  SPEAKER_01: I speak a little German, I hadn't fully thought about that.
• 53:43
  SPEAKER_00: So you have a minimum one sentence lag.
• 53:46
  SPEAKER_00: which actually makes all the difference.
• 53:49
  SPEAKER_00: Totally.
• 53:49
  SPEAKER_00: It's like very binary.
• 53:50
  SPEAKER_01: Yeah.
• 53:50
  SPEAKER_01: And then,
• 53:51
  SPEAKER_01: I mean,
• 53:51
  SPEAKER_01: I think there's this,
• 53:53
  SPEAKER_01: I do,
• 53:53
  SPEAKER_01: I do wonder if we'll hang onto this forever,
• 53:55
  SPEAKER_01: but there's this obvious cultural,
• 53:57
  SPEAKER_01: um,
• 53:57
  SPEAKER_01: you win a lot of respect and love when you can speak somebody's language,
• 54:02
  SPEAKER_00: even if you're stuck through it.
• 54:03
  SPEAKER_00: That's another huge part of it.
• 54:04
  SPEAKER_00: Yeah.
• 54:04
  SPEAKER_00: Yeah.
• 54:04
  SPEAKER_01: And if it feels like you're cheating.
• 54:05
  SPEAKER_01: It's about human connection.
• 54:06
  SPEAKER_01: Right.
• 54:06
  SPEAKER_00: Yeah.
• 54:07
  SPEAKER_01: Right.
• 54:07
  And you think we hang onto that for,
• 54:10
  that sticks around for 30,
• 54:12
  40,
• 54:12
  50 years?
• 54:12
  SPEAKER_01: I think so.
• 54:12
  SPEAKER_01: Yeah.
• 54:12
  SPEAKER_01: Yeah.
• 54:12
  SPEAKER_01: Yeah.
• 54:12
  SPEAKER_01: Okay.
• 54:16
  um i'd be willing to make that bet yeah sometimes i think yes
• 54:20
  sometimes i think no
• 54:21
  um uh yeah that one's tricky i think we you had i'm changing
• 54:28
  gears for a second
• 54:29
  just yeah real quick the bet i'm even more willing to make
• 54:32
  though is beyond
• 54:34
  language one of the most profound consequences of ai will be it
• 54:39
  will completely
• 54:39
  SPEAKER_00: change how we learn things right for me you know i've
• 54:43
  SPEAKER_00: talked a bit about how much I love learning things.
• 54:46
  It's been really indescribably crazy just how much ChatGPT has
• 54:52
  unlocked for me.
• 54:53
  It's 100x the range of even questions that I allow myself to
• 55:00
  ask because now I know
• 55:00
  SPEAKER_00: I can get an instant high-quality answer.
• 55:03
  SPEAKER_00: It's so cool.
• 55:04
  SPEAKER_00: So I think that...
• 55:06
  SPEAKER_00: you know, our sort of long-term vision is beyond language.
• 55:11
  And a lot of the platforms that we're building are
• 55:15
  theoretically applicable to
• 55:17
  SPEAKER_00: other types of knowledge as well.
• 55:19
  SPEAKER_01: Okay, this is exactly what I wanted to talk about in a lot of ways.
• 55:22
  SPEAKER_01: I was...
• 55:24
  reading is it goes back a little bit to a thing we were talking
• 55:27
  about earlier but
• 55:29
  you were talking about your goal setting in this in an
• 55:32
  interview and that you're
• 55:35
  you felt like people could be they can learn more like we were
• 55:40
  talking about but
• 55:41
  you have to be quite precise and and you were talking about
• 55:45
  setting goals every
• 55:47
  SPEAKER_01: single day what you're trying to accomplish by the end of the day
• 55:50
  checking your metrics have you it was intense and and i don't
• 55:55
  think most people
• 55:56
  live like this can if you still do this system do you still do
• 56:00
  a system yeah so i
• 56:03
  think of it as sort of my personal os also sort of work os i
• 56:09
  spend most of my time
• 56:09
  working right so you know at this point it's really the same
• 56:13
  thing but i think that
• 56:15
  SPEAKER_00: um being pretty ruthless about
• 56:18
  as a founder prioritizing what you need to get done and what is
• 56:21
  actually p0 um one
• 56:24
  of my friends um eric simons who i don't know if you know him
• 56:28
  but he runs this
• 56:29
  company called bolt it's like ai vibe coding one of the big
• 56:33
  ones now but he has
• 56:34
  SPEAKER_00: this analogy that i've heard him say publicly around how
• 56:38
  it's kind of like a lone fire truck in a town that's on fire
• 56:41
  and you have to let
• 56:43
  SPEAKER_00: most of the fires burn,
• 56:44
  SPEAKER_00: right.
• 56:45
  SPEAKER_00: And prioritize the few fires that will save the whole town.
• 56:49
  SPEAKER_00: Right.
• 56:49
  So don't worry about saving individual buildings,
• 56:51
  worry about saving the town.
• 56:53
  Um,
• 56:53
  and I think you do have to be fairly clear and ruthless about
• 56:56
  that,
• 56:57
  um,
• 56:58
  in execution.
• 56:59
  And that that's the thing that as we've grown at speak,
• 57:03
  we're now 110 people,
• 57:03
  um,
• 57:07
  SPEAKER_00: Not that long ago, we were 20 people.
• 57:10
  SPEAKER_00: So things have changed a lot.
• 57:11
  SPEAKER_00: We're still, obviously, in the grand scheme, pretty small.
• 57:14
  SPEAKER_00: But I think as we've grown,
• 57:19
  the importance of being clear about what you're going to get
• 57:24
  done on a daily basis
• 57:25
  and on a weekly basis and monthly basis has only sort of risen
• 57:29
  in importance.
• 57:30
  SPEAKER_01: What are your methods?
• 57:32
  SPEAKER_01: Do you set goals for the next day each night?
• 57:35
  SPEAKER_01: Do you do it in the morning?
• 57:37
  SPEAKER_01: None of the above?
• 57:38
  SPEAKER_00: I try to do it in the morning.
• 57:41
  SPEAKER_00: I think I'm still experimenting with things, frankly.
• 57:43
  SPEAKER_00: It's like a work in progress.
• 57:44
  SPEAKER_00: And I think that as...
• 57:47
  SPEAKER_00: Sort of, you know, I'm now married.
• 57:49
  SPEAKER_00: I don't have kids yet, but, you know, I'm thinking about it soon.
• 57:53
  So I kind of expect that all of this will continue to change
• 57:56
  and adapt and be pretty fluid.
• 57:59
  But I think that there's some sort of core axioms there around
• 58:03
  intentionality and
• 58:05
  just like really setting up systems in your life so that you
• 58:09
  don't really have to
• 58:10
  SPEAKER_00: rely on motivation.
• 58:12
  SPEAKER_00: it's just kind of the assumed default, if you will.
• 58:15
  SPEAKER_01: It's like the Brian Johnson,
• 58:17
  SPEAKER_01: Brian Johnson philosophy of,
• 58:19
  SPEAKER_00: I mean,
• 58:20
  SPEAKER_00: you know,
• 58:20
  SPEAKER_00: this is not like a particularly original line of thought,
• 58:24
  SPEAKER_00: but it's,
• 58:25
  SPEAKER_00: it's worked quite well.
• 58:26
  SPEAKER_01: Granted, but do you have specifics?
• 58:28
  People are so interested in this stuff and you say,
• 58:31
  like,
• 58:31
  you've thought about this a lot.
• 58:33
  I mean,
• 58:34
  as specific as you're willing to be,
• 58:36
  what,
• 58:36
  what,
• 58:37
  tell me your daily routine.
• 58:38
  SPEAKER_01: Uh,
• 58:40
  SPEAKER_00: Sure.
• 58:41
  So I guess like in the morning I will,
• 58:44
  and sometimes happens the previous day as well.
• 58:47
  And some of these P0s are sort of pre-scheduled for like
• 58:52
  Wednesday on Monday,
• 58:53
  for example.
• 58:54
  SPEAKER_00: So it's kind of contextual,
• 58:56
  but I will just try to have like a list of the P0s for the day
• 59:04
  that I absolutely
• 59:05
  SPEAKER_00: must get done.
• 59:07
  SPEAKER_00: That you type out.
• 59:08
  SPEAKER_01: Or write down?
• 59:09
  SPEAKER_00: I've tried both.
• 59:10
  SPEAKER_00: I now type it out.
• 59:11
  SPEAKER_00: Okay.
• 59:12
  SPEAKER_00: I think I saw like probably 10 years ago at this point,
• 59:15
  I think maybe it was Sam Altman had a thing where he would use
• 59:19
  index cards and he
• 59:19
  SPEAKER_00: would write them on like an index card.
• 59:21
  SPEAKER_00: I've seen that.
• 59:22
  SPEAKER_00: I tried that for a while.
• 59:24
  SPEAKER_01: So Sam has a ongoing Word document that is always open,
• 59:31
  SPEAKER_01: that seems to be categorized by short-term,
• 59:34
  SPEAKER_01: long-term type things.
• 59:37
  SPEAKER_01: Are you doing this just in a document or where?
• 59:40
  SPEAKER_00: So I have a Trello board.
• 59:45
  SPEAKER_00: I went through a phase where I'm like, okay, simplify.
• 59:50
  SPEAKER_00: Let's just use Apple Notes, right?
• 59:53
  SPEAKER_00: Just a note.
• 59:54
  SPEAKER_00: We use Notion at work.
• 59:56
  SPEAKER_00: So I have like a mission control file in Notion as well.
• 59:59
  SPEAKER_00: And I try to do this on a daily basis.
• 1:00:00
  SPEAKER_00: And I also have a weekly scratch file where I do this on a weekly basis.
• 1:00:06
  I'm realizing now as I talk through this,
• 1:00:08
  that I'm still very actively experimenting.
• 1:00:11
  SPEAKER_00: And I think that as we've grown,
• 1:00:15
  the nature of my role and my responsibilities have also changed
• 1:00:17
  quite a bit.
• 1:00:18
  And,
• 1:00:19
  you know,
• 1:00:19
  so,
• 1:00:20
  so that's probably,
• 1:00:21
  you know,
• 1:00:21
  part of the driver of this.
• 1:00:23
  SPEAKER_00: Um, yeah.
• 1:00:25
  SPEAKER_00: And, uh,
• 1:00:28
  SPEAKER_00: Let's see, what else?
• 1:00:29
  SPEAKER_00: I use reminders heavily on my phone.
• 1:00:31
  SPEAKER_00: Oh yeah, another thing is that I'm heavy into the dictation.
• 1:00:36
  SPEAKER_00: I've dictated 95% of my text messages for the past five years and I will
• 1:00:41
  SPEAKER_00: shamelessly do it in public.
• 1:00:45
  SPEAKER_00: And I just try to rely on systems that are simple and reliable.
• 1:00:51
  SPEAKER_00: I use reminders a lot.
• 1:00:53
  SPEAKER_00: I'll often be talking to my phone
• 1:00:55
  SPEAKER_00: I try to always prioritize and bias towards action.
• 1:00:59
  And I like to do things immediately if I know that in the next
• 1:01:03
  five minutes I can
• 1:01:04
  SPEAKER_00: get it done and then it's like off my back.
• 1:01:06
  SPEAKER_00: So I think I have,
• 1:01:08
  SPEAKER_00: frankly at this point,
• 1:01:09
  SPEAKER_00: probably an overly strong bias towards doing things immediately.
• 1:01:11
  SPEAKER_00: So I should probably dial it back a little bit.
• 1:01:15
  But that's something that I think has worked very well for the
• 1:01:18
  past few years.
• 1:01:19
  SPEAKER_01: Okay,
• 1:01:19
  SPEAKER_01: and then in this quote that I read from you before,
• 1:01:22
  SPEAKER_01: it said,
• 1:01:23
  SPEAKER_01: you know,
• 1:01:23
  set the goals every day and then review them at night and
• 1:01:28
  measure how well you did.
• 1:01:30
  SPEAKER_01: Do you still do that?
• 1:01:31
  SPEAKER_00: Yeah,
• 1:01:31
  I'm probably not as disciplined about doing that on a daily
• 1:01:34
  basis anymore,
• 1:01:35
  SPEAKER_00: but I do like to do weekly reviews,
• 1:01:37
  SPEAKER_00: for example,
• 1:01:38
  SPEAKER_00: and then think about,
• 1:01:39
  SPEAKER_00: okay,
• 1:01:39
  SPEAKER_00: like,
• 1:01:39
  SPEAKER_00: how did I do each day,
• 1:01:40
  SPEAKER_00: right?
• 1:01:41
  SPEAKER_00: Did I actually finish all my P0s?
• 1:01:43
  SPEAKER_00: Are there outstanding things?
• 1:01:45
  SPEAKER_00: I mean, frankly, like,
• 1:01:47
  The state of speak right now is that there's just a sort of
• 1:01:52
  unrelenting number of
• 1:01:53
  SPEAKER_00: fires and issues.
• 1:01:55
  So something that I've been trying to be a lot more intentional
• 1:01:57
  about is not being
• 1:02:00
  SPEAKER_00: in reactive mode all the time.
• 1:02:01
  SPEAKER_00: In Q4 of last year, it got pretty bad.
• 1:02:05
  SPEAKER_00: I felt like I was always in meetings, always in reactive mode.
• 1:02:10
  I've been trying to be a lot more strategic and actually carve
• 1:02:12
  out holds on my calendar.
• 1:02:14
  SPEAKER_00: We have shared public calendars at Speak.
• 1:02:16
  So people like putting on meetings because there's a lot of
• 1:02:19
  urgent things,
• 1:02:20
  right?
• 1:02:20
  But I've been a lot more intentional about carving out holds
• 1:02:23
  for strategic thinking
• 1:02:25
  SPEAKER_00: and visioning.
• 1:02:26
  SPEAKER_01: Okay.
• 1:02:27
  SPEAKER_01: I saw you had thoughts on the education system,
• 1:02:30
  which I find fascinating because of your background and having
• 1:02:34
  worked on education.
• 1:02:36
  If you could do anything to the American education system,
• 1:02:40
  what would you do?
• 1:02:42
  SPEAKER_00: Well, so part of the...
• 1:02:48
  SPEAKER_00: inspiration for Speak was the Diamond Age,
• 1:02:51
  SPEAKER_00: like the Neil Stephenson book,
• 1:02:53
  SPEAKER_00: where every child has essentially this AI little book.
• 1:02:59
  SPEAKER_00: It's the young lady's illustrated primer that
• 1:03:03
  is this companion that's ever-present that teaches them
• 1:03:06
  everything through a
• 1:03:07
  SPEAKER_00: story-like format and it fully understands them.
• 1:03:10
  SPEAKER_00: You know, it's the perfect AI tutor, right?
• 1:03:12
  SPEAKER_00: And that was, you know, very sci-fi concept a decade ago.
• 1:03:17
  SPEAKER_00: Now it feels very much within reach.
• 1:03:19
  SPEAKER_00: And I think that is the future.
• 1:03:21
  SPEAKER_00: That's sort of what we're building towards with Speak as well.
• 1:03:24
  SPEAKER_00: So, you know, probably that's like
• 1:03:28
  I would say that's the dream for every child to have something
• 1:03:31
  like that.
• 1:03:31
  SPEAKER_01: Yeah,
• 1:03:31
  and I think the quote I saw was before LLMs and everything had
• 1:03:35
  really taken off,
• 1:03:36
  but you were talking about smaller class sizes and school
• 1:03:39
  mostly being a place
• 1:03:40
  SPEAKER_01: maybe to socialize,
• 1:03:42
  SPEAKER_01: exercise.
• 1:03:43
  SPEAKER_01: Do you think that's where we're heading?
• 1:03:45
  SPEAKER_00: Yeah, I mean, I think that...
• 1:03:47
  most people would probably agree at this point that schooling
• 1:03:51
  is mostly childcare.
• 1:03:52
  SPEAKER_00: You're not actually, you know, learning very much.
• 1:03:55
  And if you actually focus on learning,
• 1:03:57
  then you could probably learn things a hundred X faster.
• 1:04:00
  SPEAKER_00: Right.
• 1:04:01
  SPEAKER_00: And there's many,
• 1:04:03
  SPEAKER_00: uh,
• 1:04:03
  SPEAKER_00: sort of AI school companies now,
• 1:04:06
  like synthesis school that are super cool and are starting to
• 1:04:09
  prove this out.
• 1:04:10
  SPEAKER_00: Um, so I would say that like,
• 1:04:13
  SPEAKER_00: that feels like definitely the future.
• 1:04:14
  SPEAKER_00: I don't really know how to solve the childcare issue.
• 1:04:18
  SPEAKER_00: I think that's more of a structural sort of societal issue.
• 1:04:22
  SPEAKER_00: But in general,
• 1:04:25
  SPEAKER_00: it's the same principles,
• 1:04:26
  SPEAKER_00: I think,
• 1:04:27
  SPEAKER_00: that I've always been interested in and building to all of my companies,
• 1:04:31
  SPEAKER_00: which is that smaller groups are better.
• 1:04:33
  SPEAKER_00: The best in terms of learning is private tutoring, right?
• 1:04:36
  SPEAKER_00: Like the Bloom Two Sigma problem.
• 1:04:37
  SPEAKER_00: If you have a
• 1:04:38
  SPEAKER_00: Private tutor outcomes are two standard deviations better.
• 1:04:42
  SPEAKER_00: And, you know, that also intuitively makes sense.
• 1:04:46
  SPEAKER_00: I think that's where the world should move.
• 1:04:48
  SPEAKER_01: This is, again, not... Please do not take this offensively at all.
• 1:04:53
  SPEAKER_01: I mean this from a very genuine place.
• 1:04:59
  SPEAKER_01: Same thing, kind of, as the Patrick John question.
• 1:05:03
  SPEAKER_01: You could do anything.
• 1:05:04
  It is a great achievement to build something that thousands and
• 1:05:09
  perhaps millions of
• 1:05:10
  SPEAKER_01: people use that helps them learn a language,
• 1:05:12
  SPEAKER_01: totally get that.
• 1:05:14
  SPEAKER_01: Just a great achievement to make like a...
• 1:05:17
  SPEAKER_01: Thriving company, right?
• 1:05:18
  SPEAKER_01: Jobs, it's a puzzle on its own.
• 1:05:22
  SPEAKER_01: But same question.
• 1:05:23
  SPEAKER_01: I mean, you've been doing this roughly eight years?
• 1:05:28
  SPEAKER_01: Eight years, yeah.
• 1:05:28
  SPEAKER_01: Eight years, right.
• 1:05:29
  SPEAKER_01: I mean,
• 1:05:30
  SPEAKER_01: you know,
• 1:05:30
  SPEAKER_01: and I just,
• 1:05:31
  I sort of put myself in your shoes and then you don't see your
• 1:05:35
  brother off at the
• 1:05:37
  Ark Institute with all this money getting to chase whatever
• 1:05:41
  curiosity is grabbing
• 1:05:42
  SPEAKER_01: him at the moment and wishing you could do that.
• 1:05:46
  SPEAKER_00: Um, I think that I, I don't know.
• 1:05:53
  SPEAKER_00: I just don't really see it that way.
• 1:05:54
  I think that what we're working on has the potential to be
• 1:05:59
  insanely impactful and
• 1:06:00
  SPEAKER_00: transformative for hundreds of millions of people around the world.
• 1:06:04
  SPEAKER_00: So, you know,
• 1:06:07
  SPEAKER_00: arguably doing this in the lens of a startup and a for-profit company,
• 1:06:13
  SPEAKER_00: um,
• 1:06:14
  SPEAKER_00: is a more direct way of achieving impact.
• 1:06:16
  SPEAKER_00: It's a different type of impact, right?
• 1:06:18
  SPEAKER_00: You know, it's like, I think it's a big world.
• 1:06:20
  SPEAKER_00: There's many different things that you can do.
• 1:06:22
  SPEAKER_00: I think that, um,
• 1:06:25
  SPEAKER_00: what I'm working on right now feels like,
• 1:06:28
  SPEAKER_00: like,
• 1:06:29
  SPEAKER_00: yes,
• 1:06:29
  SPEAKER_00: it's taken eight years to get to this point and I wish I had been faster,
• 1:06:33
  SPEAKER_00: but now it feels like it's on an exponential takeoff sort of runway,
• 1:06:40
  SPEAKER_00: if you will.
• 1:06:41
  And,
• 1:06:43
  um,
• 1:06:44
  the opportunity here is really the opportunity of lifetime,
• 1:06:47
  right?
• 1:06:47
  SPEAKER_00: It is the opportunity to build a company that, um,
• 1:06:54
  SPEAKER_00: can sort of scale what I always dreamed about as a child, right?
• 1:06:58
  Which is that,
• 1:06:58
  like,
• 1:06:59
  I find it really fun to learn fast and to teach other people
• 1:07:03
  things fast.
• 1:07:04
  And I think that if everyone could do this,
• 1:07:06
  it would solve a ton of societal problems,
• 1:07:08
  right?
• 1:07:09
  SPEAKER_00: And it would help people improve their own lives in, like, many ways.
• 1:07:13
  SPEAKER_00: So I think, arguably, this is...
• 1:07:17
  equally valid of a lever to do that and in general i think um
• 1:07:23
  i'm a pretty strong
• 1:07:24
  believer in capitalism and in building companies um and i think
• 1:07:29
  that is probably
• 1:07:31
  the you know the strongest driver of human prosperity that
• 1:07:35
  humanity has ever
• 1:07:36
  SPEAKER_00: created um so you know that's the approach
• 1:07:40
  And the language part,
• 1:07:41
  I totally get for the reasons we discussed,
• 1:07:43
  this bond between people.
• 1:07:46
  SPEAKER_01: I spend a lot of time in Mexico.
• 1:07:48
  SPEAKER_01: It doesn't get more intimate.
• 1:07:49
  You actually get to be part of the culture and make real
• 1:07:52
  friends and changes your
• 1:07:54
  SPEAKER_01: whole experience.
• 1:07:55
  The intelligence part,
• 1:07:57
  there's that tweet from Ilya from a couple of years ago.
• 1:08:00
  SPEAKER_01: If you've been valuing
• 1:08:02
  intelligence um as a human trait you're in for a rude awakening
• 1:08:07
  i am starting to
• 1:08:08
  feel like i'm tilting more um there's a fair amount of anxiety
• 1:08:12
  absolutely right i
• 1:08:13
  mean some of your your ultimate quest doesn't that get
• 1:08:17
  undermined if if learning
• 1:08:20
  things turns out to not really be the thing that humans need to
• 1:08:23
  focus on yeah you
• 1:08:25
  SPEAKER_00: know i suppose it's possible i think that um it is
• 1:08:31
  SPEAKER_00: I think that if superintelligence is achieved,
• 1:08:34
  SPEAKER_00: there will be a completely different sort of economy,
• 1:08:38
  SPEAKER_00: a new world order.
• 1:08:39
  SPEAKER_01: Worry about that.
• 1:08:42
  SPEAKER_00: I don't really worry too much about it.
• 1:08:43
  SPEAKER_00: I think that,
• 1:08:47
  SPEAKER_00: frankly,
• 1:08:47
  this is kind of how I feel as well about everyone that's on X
• 1:08:51
  talking about...
• 1:08:54
  everything AI and I'm actually on the ground trying to build a
• 1:08:59
  cutting edge AI
• 1:09:00
  SPEAKER_00: product and things are just always slower.
• 1:09:04
  SPEAKER_00: In fact, there's like way more inertia in the real world, right?
• 1:09:08
  Think about how long it's been since GPT-4 came out and how
• 1:09:11
  people were talking
• 1:09:11
  SPEAKER_00: about how the world is over because GPT-4 came out.
• 1:09:16
  Obviously we've made a lot of progress since then,
• 1:09:17
  but has the world really changed that much?
• 1:09:19
  SPEAKER_00: I would argue not that much.
• 1:09:21
  So it's gonna take an enormous amount of effort to build things
• 1:09:25
  that people find
• 1:09:25
  SPEAKER_00: valuable and actually like affect their lives.
• 1:09:30
  SPEAKER_01: And I'm kind of with you.
• 1:09:33
  SPEAKER_01: I feel like this is going to ooze through society a bit.
• 1:09:36
  SPEAKER_01: However,
• 1:09:36
  SPEAKER_01: this is not somebody from X.
• 1:09:39
  SPEAKER_01: It's like somebody,
• 1:09:42
  SPEAKER_01: people,
• 1:09:42
  SPEAKER_01: lots of people I know who work at the top labs 100% or,
• 1:09:45
  SPEAKER_01: you know,
• 1:09:47
  SPEAKER_01: two years,
• 1:09:47
  SPEAKER_01: three years,
• 1:09:48
  SPEAKER_01: whatever number you want to pick.
• 1:09:50
  SPEAKER_01: It's all the world is 180 flipped upside down.
• 1:09:52
  SPEAKER_01: And those are...
• 1:09:54
  SPEAKER_01: people working on foundational stuff who I think are pretty rational,
• 1:09:57
  SPEAKER_01: but you think that's not how this is gonna play out.
• 1:10:02
  SPEAKER_00: Well, I'm not saying it won't happen, right?
• 1:10:05
  I'm saying that I think change will take longer than most
• 1:10:09
  people think it will.
• 1:10:10
  I just think there's a lot of inertia in the world and that we
• 1:10:16
  need more people
• 1:10:16
  building applications that actually use these technologies and
• 1:10:21
  translate them into
• 1:10:22
  SPEAKER_00: economic value,
• 1:10:23
  SPEAKER_00: right?
• 1:10:23
  SPEAKER_00: If you look at most AI companies over the past,
• 1:10:27
  SPEAKER_00: I mean,
• 1:10:28
  SPEAKER_00: since 2022,
• 1:10:28
  SPEAKER_00: it's been mostly picks and shovels
• 1:10:31
  SPEAKER_00: There haven't been very many breakthrough consumer apps.
• 1:10:36
  We're one of them,
• 1:10:37
  but there just really actually like aren't that many others.
• 1:10:40
  SPEAKER_01: Yeah, I think that's right.
• 1:10:41
  SPEAKER_00: More are coming though.
• 1:10:41
  SPEAKER_01: Yeah, yeah.
• 1:10:42
  SPEAKER_00: I think it's starting.
• 1:10:43
  SPEAKER_01: Well, Andrew, thank you so much for your time.
• 1:10:47
  SPEAKER_01: I appreciate you being indulgent on that.
• 1:10:50
  SPEAKER_00: Of course.
• 1:10:51
  SPEAKER_01: Because this childhood product is going to follow you forever.
• 1:10:54
  SPEAKER_01: Yeah, no, I mean, I... Sense.
• 1:10:56
  SPEAKER_00: Yeah, I don't mind it.
• 1:10:58
  SPEAKER_00: Like, I don't...
• 1:11:00
  SPEAKER_00: I like talking about it, right?
• 1:11:02
  SPEAKER_00: Because I think it's an interesting story.
• 1:11:04
  I don't,
• 1:11:04
  because I don't think of it as like the circus freak kind of
• 1:11:07
  thing.
• 1:11:07
  SPEAKER_01: I find it just fascinating.
• 1:11:09
  SPEAKER_01: I mean, it's like, I love human potential and.
• 1:11:12
  I think it's an interesting story and that it also connects
• 1:11:15
  really well to a lot of
• 1:11:16
  SPEAKER_00: what I care about and am currently working on.
• 1:11:20
  SPEAKER_00: So I love talking about it.
• 1:11:22
  Is there anything that you would have done differently or you
• 1:11:29
  wish this had played
• 1:11:30
  SPEAKER_01: out a different way or,
• 1:11:32
  SPEAKER_01: or I guess that's such a hard question for anybody really?
• 1:11:34
  SPEAKER_00: Yeah.
• 1:11:35
  SPEAKER_00: I mean,
• 1:11:36
  SPEAKER_00: uh,
• 1:11:37
  SPEAKER_00: I think,
• 1:11:39
  I think when I think everyone has decisions they would probably
• 1:11:43
  make differently.
• 1:11:44
  SPEAKER_00: Right.
• 1:11:45
  SPEAKER_00: Um, I think that.
• 1:11:49
  SPEAKER_00: I've been extremely lucky.
• 1:11:51
  And there was probably a world where I could have shortened
• 1:11:57
  some of the pain early
• 1:11:58
  SPEAKER_00: on in a career.
• 1:12:01
  SPEAKER_00: But also, you know, there's no counterfactual who knows.
• 1:12:04
  SPEAKER_00: And I think it is definitely weird to think about how
• 1:12:10
  SPEAKER_00: you know,
• 1:12:11
  SPEAKER_00: when I dropped out of grad school and I started,
• 1:12:13
  SPEAKER_00: I was,
• 1:12:14
  SPEAKER_00: you know,
• 1:12:15
  SPEAKER_00: this sort of young prodigy.
• 1:12:16
  SPEAKER_00: And at this point now,
• 1:12:19
  SPEAKER_00: right,
• 1:12:19
  SPEAKER_00: I see all these sort of like 17-year-old YC founders,
• 1:12:22
  SPEAKER_00: and I'm like,
• 1:12:22
  SPEAKER_00: wow,
• 1:12:23
  SPEAKER_00: I've been at this for a while,
• 1:12:24
  SPEAKER_00: actually.
• 1:12:26
  SPEAKER_00: And I see some of my early self in them as well, right?
• 1:12:29
  SPEAKER_00: And it's so different and so interesting.
• 1:12:32
  SPEAKER_00: So, you know, in some sense, I think...
• 1:12:38
  I feel like I've been fortunate enough to have been sort of
• 1:12:43
  advised and lifted up
• 1:12:44
  SPEAKER_00: with no expectation of return by many people over the past 15 years.
• 1:12:48
  SPEAKER_00: And I want to start sort of giving back and doing that as well.
• 1:12:53
  SPEAKER_00: And I just feel like also...
• 1:12:58
  Everything in my life and in my career has been almost building
• 1:13:02
  up to this moment
• 1:13:04
  where I really do feel like Speak is in the perfect position to
• 1:13:10
  build this thing
• 1:13:10
  SPEAKER_00: that I'd been dreaming of and really take over the industry.
• 1:13:14
  SPEAKER_00: So I wouldn't really give that up for anything.
• 1:13:16
  SPEAKER_01: Okay.
• 1:13:16
  SPEAKER_01: No, it's cool.
• 1:13:17
  SPEAKER_01: I mean, God, I was looking at the numbers.
• 1:13:18
  I mean,
• 1:13:18
  you guys are growing,
• 1:13:19
  what is it,
• 1:13:20
  like 50% a year or something like that?
• 1:13:22
  SPEAKER_00: A lot more than that.
• 1:13:23
  SPEAKER_01: A lot more than that.
• 1:13:24
  SPEAKER_01: Yeah, yeah, yeah.
• 1:13:26
  SPEAKER_01: What do you and Patrick talk about when you guys hang out?
• 1:13:30
  SPEAKER_00: Oh man, he's super social.
• 1:13:34
  SPEAKER_01: I wouldn't say notorious, but everyone tells me he's like every party.
• 1:13:38
  SPEAKER_00: I don't know how he does it.
• 1:13:40
  And then somehow he finds the time to keep publishing nature
• 1:13:43
  and science papers.
• 1:13:44
  SPEAKER_00: That's the most mystifying part of it.
• 1:13:47
  SPEAKER_00: So I think it's been
• 1:13:49
  SPEAKER_00: It's been really fun as well to see his explosive trajectory.
• 1:13:54
  And he now also does AI investing and thinks a lot about AI and
• 1:14:00
  computational
• 1:14:00
  SPEAKER_00: biology in his work.
• 1:14:03
  So it's always really fun just to be able to talk to him about
• 1:14:09
  the future and what
• 1:14:10
  SPEAKER_00: he's building as well.
• 1:14:11
  SPEAKER_01: Did you have to do Taiwanese talk shows?
• 1:14:14
  SPEAKER_01: Yeah.
• 1:14:15
  SPEAKER_01: Yeah.
• 1:14:15
  SPEAKER_01: You did?
• 1:14:16
  SPEAKER_00: A lot of them.
• 1:14:17
  SPEAKER_01: As what?
• 1:14:17
  SPEAKER_01: Like a 12 year old?
• 1:14:18
  SPEAKER_01: Yeah.
• 1:14:18
  SPEAKER_00: Yeah.
• 1:14:19
  SPEAKER_00: As like a, yeah, 10 or 11 year old.
• 1:14:21
  SPEAKER_00: So I've, yeah.
• 1:14:23
  SPEAKER_00: All of this is very weird for sure.
• 1:14:24
  SPEAKER_00: But I'm,
• 1:14:25
  SPEAKER_00: I'm like,
• 1:14:26
  SPEAKER_00: yeah,
• 1:14:26
  SPEAKER_00: I'm like doing stuff on camera is,
• 1:14:30
  SPEAKER_00: is a thing I've,
• 1:14:31
  SPEAKER_00: I've been doing for a long time.
• 1:14:32
  SPEAKER_00: So it's like not all weird for me.
• 1:14:33
  SPEAKER_01: Yeah.
• 1:14:34
  And so,
• 1:14:35
  and I went hunting for the book,
• 1:14:36
  but I'm not gonna,
• 1:14:37
  I'm probably not gonna find it.
• 1:14:38
  SPEAKER_00: I think it's out of print.
• 1:14:39
  SPEAKER_00: Maybe I'll hunt one up.
• 1:14:41
  SPEAKER_01: We've got to put like a PDF out there or something.
• 1:14:47
  SPEAKER_01: But it won a bunch of awards or some massive award.
• 1:14:51
  SPEAKER_00: Yeah.
• 1:14:51
  Well,
• 1:14:51
  so I think people have always been interested in my own
• 1:14:57
  personal life story.
• 1:14:58
  SPEAKER_00: So there was like a sort of autobiography when I was 12.
• 1:15:02
  And then there was another one where Patrick and I interviewed
• 1:15:05
  a bunch of Nobel
• 1:15:06
  SPEAKER_00: Prize winner scientists and were super interested in understanding.
• 1:15:11
  SPEAKER_00: you know,
• 1:15:11
  SPEAKER_00: the methods behind their success and,
• 1:15:13
  SPEAKER_00: um,
• 1:15:14
  SPEAKER_00: you know,
• 1:15:15
  just like the commonalities and the ways that they would think
• 1:15:19
  about super
• 1:15:20
  SPEAKER_00: intensely,
• 1:15:21
  SPEAKER_00: you know,
• 1:15:21
  SPEAKER_00: usually,
• 1:15:21
  SPEAKER_00: um,
• 1:15:24
  SPEAKER_00: what they were devoting their life toward.
• 1:15:26
  I don't want to sound this to sound as kiss-assy as it's going
• 1:15:29
  to come off,
• 1:15:29
  SPEAKER_01: but I,
• 1:15:30
  SPEAKER_01: I,
• 1:15:32
  SPEAKER_01: having met you and your brother and Patrick and John,
• 1:15:34
  SPEAKER_01: um,
• 1:15:37
  SPEAKER_01: I've always wanted,
• 1:15:37
  SPEAKER_01: I have wanted to,
• 1:15:39
  SPEAKER_01: I think there is a genetic component to this,
• 1:15:41
  but then something your parents were all doing was right
• 1:15:44
  because the very like high
• 1:15:47
  SPEAKER_01: integrity,
• 1:15:49
  SPEAKER_01: personable,
• 1:15:50
  SPEAKER_01: both,
• 1:15:54
  SPEAKER_01: head roaring at a million miles an hour,
• 1:15:56
  SPEAKER_01: but also able to have conversations with just about anybody,
• 1:16:00
  SPEAKER_01: polite.
• 1:16:01
  SPEAKER_01: I don't know.
• 1:16:02
  SPEAKER_01: I want to do a book on something about parents that get things.
• 1:16:06
  SPEAKER_01: There's some sort of brother's book in there.
• 1:16:09
  SPEAKER_01: There's this other book that nobody's read,
• 1:16:11
  SPEAKER_01: The Bulls and the Hintons,
• 1:16:13
  SPEAKER_01: about these two exceptional families that had a lot of...
• 1:16:20
  marriages and overlapping,
• 1:16:21
  you know,
• 1:16:22
  it's like 300 years of,
• 1:16:24
  of just success going.
• 1:16:26
  SPEAKER_00: Bulls and the Hintons.
• 1:16:27
  SPEAKER_00: Yeah.
• 1:16:27
  SPEAKER_01: Going from, from bull inventing logic that would one day play roles.
• 1:16:35
  SPEAKER_00: Yes.
• 1:16:35
  SPEAKER_00: Yes.
• 1:16:35
  SPEAKER_00: Yeah.
• 1:16:35
  SPEAKER_01: The mathematician in,
• 1:16:37
  SPEAKER_01: in,
• 1:16:37
  SPEAKER_01: in AIs to,
• 1:16:38
  SPEAKER_01: to Jeff Hinton,
• 1:16:40
  SPEAKER_00: who like,
• 1:16:40
  SPEAKER_01: whatever it is,
• 1:16:41
  SPEAKER_01: 200 years later,
• 1:16:42
  SPEAKER_01: you know,
• 1:16:43
  SPEAKER_01: but they had,
• 1:16:43
  SPEAKER_01: there's
• 1:16:45
  SPEAKER_01: One of the people in the family invented like the, uh, jungle gym.
• 1:16:48
  SPEAKER_01: The other one invented the concept of like the Tesseract.
• 1:16:51
  SPEAKER_01: One went to like
• 1:16:53
  China and started schools and it was just somebody invented the
• 1:16:56
  concept of GDP you
• 1:16:59
  know it's like it's wild yeah somebody had to invent that right
• 1:17:03
  and then you know
• 1:17:04
  so obviously there's some sort of genetics at play but there's
• 1:17:06
  also this I feel
• 1:17:08
  like there's this there's these people who sort of figured
• 1:17:11
  families that figured
• 1:17:12
  life out a little better than the rest of us so I would read
• 1:17:16
  that book please write
• 1:17:17
  it alright I'll work on that thank you so much for coming in I
• 1:17:20
  really appreciate
• 1:17:21
  SPEAKER_00: the time of course thank you
• 1:17:26
  core memory is hosted by me ashley vance the show is produced
• 1:17:30
  by david nicholson
• 1:17:32
  and me our theme music is by james mercer and john sortland and
• 1:17:37
  the show is edited
• 1:17:39
  by the john sortland thank you all for listening as mentioned
• 1:17:45
  the show is sponsored
• 1:17:46
  SPEAKER_01: by you on ventures thanks