How to Use AI and Music to Boost Creativity with Percy Liang, Computer Scientist and Classical Pianist

1
00:00:00,100 --> 00:00:05,600
Finally, I get to think about
how a I especially with the

2
00:00:05,600 --> 00:00:10,800
recent advances and generative
Ai and generative models, how

3
00:00:10,800 --> 00:00:14,800
you can actually have a program
that learns to generate music,

4
00:00:15,800 --> 00:00:19,700
what fuels, a multi passionate
life, I'm Jessica wine.

5
00:00:19,900 --> 00:00:22,800
And in this podcast, I interview
people who straddle two,

6
00:00:22,800 --> 00:00:27,900
completely different worlds, I
call them ampersands and we are

7
00:00:27,900 --> 00:00:31,500
collectively designing.
The Purse and Manifesto.

8
00:00:37,000 --> 00:00:40,500
I'm delighted to be talking with
Percy of the um computer

9
00:00:40,500 --> 00:00:42,600
scientists and classical
pianists.

10
00:00:43,000 --> 00:00:46,400
Percy is an associate professor
of computer science at Stanford

11
00:00:46,400 --> 00:00:49,400
University.
He got his bachelor's from MIT

12
00:00:49,500 --> 00:00:53,300
and PhD from UC Berkeley.
His research is on artificial,

13
00:00:53,300 --> 00:00:56,700
intelligence, more, specifically
Foundation models that can

14
00:00:56,700 --> 00:01:01,600
generate text images and music.
Percy has been playing piano

15
00:01:01,600 --> 00:01:05,600
since he was eight years old.
He's won several contests

16
00:01:05,600 --> 00:01:09,900
including the Phoenix Young
Musicians competition, the MIT

17
00:01:09,900 --> 00:01:14,500
concerto competition and kdf
sees classical Star Search.

18
00:01:14,900 --> 00:01:18,400
He enjoys giving both solo piano
recitals and playing chamber

19
00:01:18,400 --> 00:01:21,300
music, Percy, welcome to the
show.

20
00:01:21,400 --> 00:01:24,200
Thanks Jessica for having me.
I'm really excited to be here.

21
00:01:24,600 --> 00:01:28,500
When you think back to
Childhood, what were some early

22
00:01:28,500 --> 00:01:32,500
inklings that pointed toward a
future life in computer science

23
00:01:32,500 --> 00:01:35,200
and music.
Isaac I started playing piano

24
00:01:35,200 --> 00:01:42,700
when I was eight and I at the
time I liked science and math

25
00:01:42,700 --> 00:01:45,900
and eventually started getting
into programming.

26
00:01:45,900 --> 00:01:51,200
So the to piano and computer
science started quickly becoming

27
00:01:51,600 --> 00:01:53,800
integral part of my life.
I would say that in the

28
00:01:53,800 --> 00:01:58,300
beginning piano was I got
excited when I first started to

29
00:01:58,300 --> 00:02:01,000
learn because I was getting
better and then there was a

30
00:02:01,000 --> 00:02:03,000
slump I change teachers moved
around.

31
00:02:03,200 --> 00:02:06,900
That and then starting in eighth
grade, I started getting more

32
00:02:06,900 --> 00:02:11,700
momentum and since then I've
been really trying to keep up

33
00:02:11,700 --> 00:02:16,000
with piano up until just a maybe
a few years ago I feel like I

34
00:02:16,000 --> 00:02:19,200
was improving improving of
course things get a little bit

35
00:02:19,500 --> 00:02:25,200
harder when you have a full-time
job and kids and so on but I

36
00:02:25,200 --> 00:02:29,600
majored in computer science in
undergrad and at the time I was

37
00:02:29,600 --> 00:02:35,000
also actively involved in music.
I Would say that obviously

38
00:02:35,000 --> 00:02:36,800
professionally, I'm a computer
scientist.

39
00:02:36,800 --> 00:02:40,100
So in terms of actual number of
hours, I spent a lot more time

40
00:02:40,500 --> 00:02:43,700
working doing computer science
but piano has always been to

41
00:02:43,700 --> 00:02:48,300
integral part of my life.
How do you see piano and

42
00:02:48,300 --> 00:02:53,400
computer science connecting with
each other or even helping you

43
00:02:53,400 --> 00:02:57,700
learn about?
The other thing superficially it

44
00:02:57,700 --> 00:03:00,900
looks like the two are very
different parts of my life.

45
00:03:01,200 --> 00:03:05,500
I work hard candy.
Our science and research, and

46
00:03:05,500 --> 00:03:09,900
then I go play piano.
But there are some interesting

47
00:03:09,900 --> 00:03:12,800
connections that I found
throughout the years because it

48
00:03:12,800 --> 00:03:17,100
is one brain up here at the
house to manage both of them.

49
00:03:17,300 --> 00:03:22,200
In one of the aspects is when I
had a teacher, Mel, 10 years

50
00:03:22,200 --> 00:03:27,200
ago, who really resonated with
me, he was actually a scientist,

51
00:03:27,200 --> 00:03:30,100
but then he became a
professional pianist and then a

52
00:03:30,100 --> 00:03:34,800
teacher and I think there was
this a Aspect of really

53
00:03:34,800 --> 00:03:38,900
architecting performance and
thinking about the big picture

54
00:03:39,000 --> 00:03:42,600
because often piano pieces,
especially if they're

55
00:03:42,600 --> 00:03:44,600
technically challenging to have
just a lot of notes.

56
00:03:44,600 --> 00:03:47,600
And it's very easy to get
wrapped up in how to play the

57
00:03:47,600 --> 00:03:52,200
individual notes but thinking
about the Arc of the entire Tire

58
00:03:52,200 --> 00:03:56,900
piece was something that I
really learned from him.

59
00:03:57,700 --> 00:04:04,100
And this idea is actually very
useful for for Research when I

60
00:04:04,100 --> 00:04:08,100
write papers when I write
programs to think about the big

61
00:04:08,100 --> 00:04:11,600
picture and the essence of what
you're trying to do.

62
00:04:12,100 --> 00:04:14,800
And so that's sort of an
abstract connection.

63
00:04:15,300 --> 00:04:18,700
The more recently, there's been
a tighter connection.

64
00:04:18,700 --> 00:04:22,600
So, for a while, but I've always
wanted to combine the two in

65
00:04:22,600 --> 00:04:25,500
some capacity, but never really
got the chance to do to apply

66
00:04:25,500 --> 00:04:29,900
for another career things.
But more recently, I've actually

67
00:04:29,900 --> 00:04:34,700
had two postdocs in my lab who
were Were applying machine

68
00:04:34,700 --> 00:04:38,100
learning to, to music.
And this has been actually

69
00:04:38,100 --> 00:04:44,200
really exciting because finally,
I get to think about how a I

70
00:04:44,200 --> 00:04:49,000
especially with the recent
advances and generative Ai and

71
00:04:49,000 --> 00:04:52,900
generative models, how you could
actually have a program that

72
00:04:52,900 --> 00:04:56,100
learns to generate music.
There's actually, a sort of a

73
00:04:56,100 --> 00:04:59,900
personal reason for this, which
I'll rewind a little bit.

74
00:04:59,900 --> 00:05:03,000
So, for in grad school, I was
really interested.

75
00:05:03,100 --> 00:05:05,500
Then doing this counter
competitions.

76
00:05:05,500 --> 00:05:09,600
I spend a lot of time away from
research, trying to prepare

77
00:05:09,600 --> 00:05:13,100
myself.
And, you know, I realized that I

78
00:05:13,108 --> 00:05:17,300
could never really compete with
the people who are doing this

79
00:05:17,300 --> 00:05:19,300
full-time.
If you practice eight hours a

80
00:05:19,308 --> 00:05:22,300
day, I just can't afford that
kind of time.

81
00:05:22,700 --> 00:05:27,800
But on the other hand, I felt
like I had these ideas of how I

82
00:05:27,800 --> 00:05:31,300
wanted to express things, but it
doesn't have the time to

83
00:05:31,800 --> 00:05:36,800
actually execute cute them.
So I was looking to machine

84
00:05:36,800 --> 00:05:40,900
learning and AI as a way to
extend, or augment my

85
00:05:40,900 --> 00:05:45,200
capabilities in some way, and I
haven't got there yet.

86
00:05:45,200 --> 00:05:49,700
So maybe one day, I would love
to be able to have some sort of

87
00:05:49,900 --> 00:05:54,600
way of figuring out how to
express my musical ideas.

88
00:05:55,100 --> 00:05:59,600
But not have to, I mean, I know
this sounds like, oh, I don't

89
00:05:59,600 --> 00:06:03,700
have to practice but, but given
the limited time, Actually be

90
00:06:03,700 --> 00:06:09,600
able to create something with AI
that I'm happy with this is

91
00:06:09,600 --> 00:06:13,900
absolutely fascinating.
I think that we hear so much

92
00:06:13,900 --> 00:06:20,600
about AI being competitive with
human work, the human creative

93
00:06:20,600 --> 00:06:26,900
aspect, but what you're really
talking about is how a I could

94
00:06:27,100 --> 00:06:30,700
help you create more and create
better.

95
00:06:30,800 --> 00:06:32,600
That's what I'm hearing.
Yeah.

96
00:06:33,000 --> 00:06:36,200
So there's a little long
tradition of AI, which is

97
00:06:36,200 --> 00:06:40,200
artificial intelligence versus
IA, which is intelligence

98
00:06:40,200 --> 00:06:44,900
augmentation, which was
something that the human

99
00:06:44,900 --> 00:06:47,800
computer interaction Community,
really embraced.

100
00:06:48,200 --> 00:06:54,300
And so these two communities
really worked in parallel, in

101
00:06:54,300 --> 00:06:58,600
some ways where a, I was focused
on building these independent

102
00:06:58,600 --> 00:07:01,600
agents that could mimic human
intelligence and operate

103
00:07:01,600 --> 00:07:05,300
independently.
Whereas I tend to think of a lot

104
00:07:05,300 --> 00:07:11,600
more about AI, as IA as tools
that augment and help us as

105
00:07:11,600 --> 00:07:17,000
humans because after all, we
should be helping ourselves in

106
00:07:17,200 --> 00:07:19,600
humanity and Society in some
way.

107
00:07:20,100 --> 00:07:26,100
I think that for the piano
aspect, this is really in some

108
00:07:26,100 --> 00:07:29,900
ways personal because music is
so, you know, it's part of my

109
00:07:29,900 --> 00:07:34,000
expression.
And anyway, I can use To augment

110
00:07:34,000 --> 00:07:37,400
and improve that.
I would be very excited about

111
00:07:37,800 --> 00:07:42,300
what would a brilliant outcome?
Look like, what would the

112
00:07:42,700 --> 00:07:46,000
augmented Percy be able to
achieve?

113
00:07:46,400 --> 00:07:53,400
So I have no immediate plans to
do this but I would say that I

114
00:07:53,400 --> 00:07:57,500
would still like to be able to
play the piano.

115
00:07:57,500 --> 00:08:02,000
I mean, I could imagine people
wanting to just have an

116
00:08:02,000 --> 00:08:05,900
alternative Her face, for
example, describe it in words or

117
00:08:05,900 --> 00:08:09,100
play something very simplified
but I do find joy and actually

118
00:08:09,100 --> 00:08:12,900
trying to execute and there's a
trade-off.

119
00:08:12,900 --> 00:08:17,700
I feel like when I'm performing
can be play with more abandoned

120
00:08:17,800 --> 00:08:23,000
and there's more energy and
creativity and risk-taking but

121
00:08:23,000 --> 00:08:26,200
you play wrong notes and
sometimes you mess things up.

122
00:08:26,700 --> 00:08:31,500
But so often you have to pull
back a little bit and keep your

123
00:08:31,500 --> 00:08:35,100
head on your shoulders and And
make sure that it's sort of

124
00:08:35,100 --> 00:08:40,000
coming out correctly, but maybe
there's a way to have your cake

125
00:08:40,000 --> 00:08:43,299
and eat it too.
If you have something that could

126
00:08:43,500 --> 00:08:47,700
fix the mistakes, we see these
things in, you know, auto-tune

127
00:08:47,700 --> 00:08:49,900
for example, for her, for
singing.

128
00:08:49,900 --> 00:08:54,000
But I think it goes beyond just
fixing mistakes.

129
00:08:54,700 --> 00:09:01,600
I think these guys could create
new suggestions of things that

130
00:09:01,600 --> 00:09:02,900
you might not have thought
about.

131
00:09:03,300 --> 00:09:07,300
I think this could teach us
about how we might want to

132
00:09:07,700 --> 00:09:12,300
interpret things or express
ideas because the way that these

133
00:09:12,300 --> 00:09:17,600
as are built they are exposed to
such a vast array of different

134
00:09:17,600 --> 00:09:22,400
data and experiences that they
could capture a lot more than

135
00:09:22,400 --> 00:09:25,600
any individual human.
Being of course, doing its own

136
00:09:25,800 --> 00:09:31,300
limited alien way in some ways.
But nonetheless, it could bring

137
00:09:31,300 --> 00:09:33,300
new ideas.
In.

138
00:09:33,400 --> 00:09:34,800
So ideally I would like to
learn.

139
00:09:34,800 --> 00:09:40,200
I would like to better myself.
In addition to creating more

140
00:09:40,400 --> 00:09:43,800
polished output, we're going to
share an excerpt from your

141
00:09:43,800 --> 00:09:46,700
recording of Franz Liszt's
Piano, Sonata in B.

142
00:09:46,700 --> 00:09:50,000
Minor, this is one of the most
challenging pieces of piano

143
00:09:50,000 --> 00:09:54,600
music ever.
It is 30 minutes straight and

144
00:09:54,600 --> 00:09:59,000
there are so many theories about
this Sonata ranging from being a

145
00:09:59,000 --> 00:10:02,900
portrait of the Faust Legend to
certain biblical Theory.

146
00:10:03,000 --> 00:10:07,500
Is to just being artistic
expression, having worked on

147
00:10:07,500 --> 00:10:10,900
this piece extensively.
How do you interpret it?

148
00:10:11,200 --> 00:10:16,000
It's one of these pieces that
you learn and then you come back

149
00:10:16,000 --> 00:10:20,500
to and you play and each time
you come back you see it in a

150
00:10:20,500 --> 00:10:24,900
slightly different light and
What attracted me to this piece

151
00:10:24,900 --> 00:10:30,800
is it has a few motifs in it and
these motifs get transformed

152
00:10:31,000 --> 00:10:37,800
over and over again and you See
them in different ways and it's

153
00:10:37,800 --> 00:10:42,200
just a kind of a very coherent
cohesive piece of work.

154
00:10:42,700 --> 00:10:48,200
So it is interesting to see how
list was able to take a very

155
00:10:48,200 --> 00:10:54,400
simple motive rum pum pum pum
pum pum pum and transform it in

156
00:10:54,400 --> 00:10:56,600
all these different variations
lyrical.

157
00:10:56,700 --> 00:11:02,800
Staccato, old gentle and that
sort of creativity with.

158
00:11:03,000 --> 00:11:08,300
When is very inspirational.
And, of course, it's a very

159
00:11:08,300 --> 00:11:11,300
Monumental work and it's
extremely challenging,

160
00:11:11,300 --> 00:11:15,000
especially being able to play
for 30 minutes without stopping.

161
00:11:15,000 --> 00:11:19,700
So, everytime I play through it.
It feels like, you know, there's

162
00:11:19,700 --> 00:11:23,000
a sort of a physical and
emotional reaction after having

163
00:11:23,000 --> 00:11:25,000
gone through.
It's a journey.

164
00:12:01,800 --> 00:12:08,100
Now, being a father of young
children and having more life

165
00:12:08,100 --> 00:12:12,000
experience, how does this Sonata
resonate with you in this

166
00:12:12,000 --> 00:12:15,000
particular season of life?
I suppose.

167
00:12:15,000 --> 00:12:21,000
One thing about raising children
is that everything is constantly

168
00:12:21,000 --> 00:12:25,900
moving and changing just when
you think that you've landed and

169
00:12:26,900 --> 00:12:33,400
Certain routine things change
and it's very capricious

170
00:12:33,700 --> 00:12:39,600
unexpected, and a lot of this is
reminiscent of the list.

171
00:12:39,600 --> 00:12:44,600
Sonata where everything keeps on
moving, it involves over time.

172
00:12:45,200 --> 00:12:48,700
There are some things refrains
where you come back to the same

173
00:12:48,700 --> 00:12:51,200
ideas but sometimes I'll be in a
different key and it will be

174
00:12:51,200 --> 00:12:56,100
slightly different and it's it
suggests that life.

175
00:12:56,600 --> 00:13:01,100
Raising children is, is always
moving to a new place.

176
00:13:01,100 --> 00:13:08,900
It's never exactly in stasis.
Let's take a short break and

177
00:13:08,900 --> 00:13:11,500
pause for a moment of
reflection.

178
00:13:11,800 --> 00:13:20,400
What in your life is moving to a
new place outside of Hosting

179
00:13:20,400 --> 00:13:23,000
this podcast.
I partner with leaders in the

180
00:13:23,000 --> 00:13:26,100
workplace as they rise up,
perhaps you've gotten a

181
00:13:26,100 --> 00:13:28,300
promotion.
Should've landed a new role or

182
00:13:28,300 --> 00:13:31,700
taken on a lot more
responsibility and you need a

183
00:13:31,700 --> 00:13:35,100
trusted sounding board to
support you as you support your

184
00:13:35,100 --> 00:13:37,700
team.
I coach individuals and

185
00:13:37,700 --> 00:13:42,600
Leadership teams to rise to the
challenge with my a and blend of

186
00:13:42,600 --> 00:13:46,600
analytical and creative.
If any of this intrigues, you

187
00:13:46,800 --> 00:13:48,700
reach out to me at Jessica
one.com.

188
00:13:50,000 --> 00:13:56,800
Now, back to the show, you have
mentioned that you Were really,

189
00:13:56,800 --> 00:14:00,700
really into competitions.
You are currently working on

190
00:14:00,700 --> 00:14:06,300
learning all 24 Chopin, etudes.
It seems like this yearning for

191
00:14:06,300 --> 00:14:09,900
Learning and achievement has
kind of been with you throughout

192
00:14:09,900 --> 00:14:14,700
your life.
How do you decide what that next

193
00:14:14,900 --> 00:14:20,100
thing is that you want to work
towards for the last 20 years.

194
00:14:20,100 --> 00:14:25,600
I've consistently tried to set a
goal of giving a recital and I

195
00:14:25,608 --> 00:14:31,700
find that This is good because
it allows me to share music with

196
00:14:31,800 --> 00:14:35,700
other people, but it also gives
me a focus point.

197
00:14:36,000 --> 00:14:39,200
I learned in you to learn a
certain amount of music to, to

198
00:14:39,200 --> 00:14:41,900
be able to play and not
embarrass myself.

199
00:14:42,400 --> 00:14:46,600
And so this is led to steady
progress and I think there is

200
00:14:46,600 --> 00:14:51,400
something really exciting and
about growing when you're

201
00:14:51,400 --> 00:14:55,300
exposed to new music.
Obviously, many of these these

202
00:14:55,300 --> 00:14:59,200
pieces, like, let's not Anna you
could play every day for the

203
00:14:59,208 --> 00:15:02,700
rest of life and you'll see
something new but there's so

204
00:15:02,700 --> 00:15:06,500
much interesting and mesmerizing
music out there.

205
00:15:07,100 --> 00:15:11,400
I'm way more than any individual
human kid Explorer so just

206
00:15:11,400 --> 00:15:15,700
trying to do as much of it.
I think is one of the driving

207
00:15:15,700 --> 00:15:21,100
factors and for this, this
latest project of learning that

208
00:15:21,400 --> 00:15:26,100
24, etudes, I've played my first
Etude in.

209
00:15:26,500 --> 00:15:29,700
High School.
Okay, the number number 3, Opus

210
00:15:29,700 --> 00:15:35,900
10 and over time I've learned
more and more and about six

211
00:15:35,900 --> 00:15:39,800
years ago I decided Well I'll
set a goal of Performing the

212
00:15:39,800 --> 00:15:44,500
first 12 of them.
And I did that which was in

213
00:15:44,500 --> 00:15:49,900
itself, a quite a, quite a heavy
lift and then so I decided that

214
00:15:49,900 --> 00:15:53,400
I would try to, you know, I'm
halfway there, I might as well

215
00:15:53,400 --> 00:15:56,000
continue and so I'm working on
it.

216
00:15:56,400 --> 00:15:59,800
It's through the pandemic and
having kids, it's definitely a

217
00:15:59,800 --> 00:16:05,700
lot, a lot slower and I'm not
sure I would perform all 24.

218
00:16:05,700 --> 00:16:08,100
I don't know if the audience
would want to sit through an

219
00:16:08,100 --> 00:16:12,400
hour Chopin etudes or whether my
fingers would fall off at the

220
00:16:12,400 --> 00:16:14,500
end.
But but nonetheless, I will

221
00:16:15,300 --> 00:16:18,100
learn them and maybe I'll record
them.

222
00:16:18,700 --> 00:16:24,000
Both piano and computer science
are heavily finger Reliant and

223
00:16:24,000 --> 00:16:26,300
you've talked a little bit about
how you keep your fingers.

224
00:16:26,400 --> 00:16:30,900
In shape how do you keep your
fingers healthy you know over a

225
00:16:30,900 --> 00:16:35,700
long period of years decades of
playing piano and also as a

226
00:16:35,700 --> 00:16:40,500
computer scientist for piano
obviously it's much more

227
00:16:40,500 --> 00:16:42,900
physically demanding.
Like you're pressing the keys,

228
00:16:42,900 --> 00:16:47,500
you're going around and I've
been fortunate enough.

229
00:16:47,500 --> 00:16:51,000
I think in my career not to have
serious risk problems.

230
00:16:51,100 --> 00:16:53,700
Knock on wood there have been a
few moments.

231
00:16:53,700 --> 00:16:58,600
Why I feel like I was at the
cusp like I was typing in a lot

232
00:16:58,600 --> 00:17:03,500
or playing a lot where it would
it would hurt a little bit that

233
00:17:03,500 --> 00:17:05,400
was a sign.
I should probably hold back as

234
00:17:05,400 --> 00:17:10,700
and try to give it some rest.
They're also been times where

235
00:17:10,800 --> 00:17:15,300
it's not so much the wrist but
the actual fingers touching the

236
00:17:15,300 --> 00:17:19,200
keyboard over and over again
that they get fairly bruised up.

237
00:17:19,500 --> 00:17:24,099
And I remember once in a kind of
competition going and pulling my

238
00:17:24,099 --> 00:17:28,300
thing and then I realized that
Blood on the piano.

239
00:17:28,700 --> 00:17:33,900
So I apologize profusely to the
next contestant and try to wipe

240
00:17:33,900 --> 00:17:37,200
it up as much as I could.
Yeah, it could get really

241
00:17:37,200 --> 00:17:39,600
intense.
I think actually more important

242
00:17:39,600 --> 00:17:42,600
is like the back and the
shoulders, if you're into

243
00:17:42,600 --> 00:17:46,700
something whether it be at a
computer or in front of a piano,

244
00:17:47,300 --> 00:17:52,800
I sometimes forget about my
physical body until I'm reminded

245
00:17:52,800 --> 00:17:55,400
by it.
When my shoulder start hurting

246
00:17:55,800 --> 00:18:01,000
and That's something that
requires obviously taking breaks

247
00:18:01,000 --> 00:18:04,800
sometimes, even going between
different activities sitting in

248
00:18:04,800 --> 00:18:06,500
front of a computer.
And I'm going to a piano and

249
00:18:06,500 --> 00:18:10,000
going back there.
Both can be demanding but there

250
00:18:10,300 --> 00:18:14,500
may be demanding, a slightly
different ways and that can give

251
00:18:14,800 --> 00:18:18,000
some muscles or part of the
brain.

252
00:18:18,000 --> 00:18:21,800
A bit of a break.
While the other part as active,

253
00:18:22,200 --> 00:18:25,600
what has fueled you to keep
going as both a computer

254
00:18:25,600 --> 00:18:30,600
scientist And as a musician.
So for computer science and AI

255
00:18:30,600 --> 00:18:36,700
researcher the idea of having
computers that can think and

256
00:18:36,700 --> 00:18:43,600
reason and do various tasks that
humans are we were usually in

257
00:18:43,600 --> 00:18:47,400
the human scope has always been
kind of a dream.

258
00:18:47,400 --> 00:18:52,700
It's shocking how much progress
there is in the field and it's

259
00:18:52,700 --> 00:18:58,600
just very exciting part of it is
the Research and the ability to

260
00:18:58,800 --> 00:19:02,400
find new things and discover on
the music side.

261
00:19:02,700 --> 00:19:08,300
I feel like it's something that
my body just needs in a in a

262
00:19:08,600 --> 00:19:11,800
way.
So why do you why do we sleep or

263
00:19:11,900 --> 00:19:18,400
eat or do exercise?
It's rejuvenating and it's keeps

264
00:19:18,400 --> 00:19:22,300
me balanced, so there's that
aspect of it.

265
00:19:22,600 --> 00:19:27,600
And for me there's another
aspect which is I am curious

266
00:19:27,600 --> 00:19:31,700
about how what you can do.
So I play a lot of classical

267
00:19:31,700 --> 00:19:35,800
piano where the score is
written, but there's so much

268
00:19:35,800 --> 00:19:38,500
interpretation.
And if you look listen to

269
00:19:39,400 --> 00:19:44,500
Beethoven Sonata by Richter or
five Horowitz or Daniel

270
00:19:44,500 --> 00:19:49,300
barenboim, they're all very
different and I think developing

271
00:19:49,700 --> 00:19:54,300
my own voice and figuring out
what I want to say in this

272
00:19:54,300 --> 00:19:58,600
music.
I find that Very appealing and I

273
00:19:58,608 --> 00:20:02,100
guess part of it is also related
to that aspect of research.

274
00:20:02,100 --> 00:20:06,300
How do you say something
generally new, right?

275
00:20:06,300 --> 00:20:11,100
Because everyone has played many
of these these pieces.

276
00:20:11,400 --> 00:20:14,900
But what can what can I bring to
the table?

277
00:20:14,900 --> 00:20:17,200
That's saying something fresh
and new.

278
00:20:17,200 --> 00:20:21,100
And so in that way, that's a bit
similar to the way we think

279
00:20:21,100 --> 00:20:23,600
about research.
What advice do you have for

280
00:20:23,600 --> 00:20:27,500
people who are pursuing or
Thinking of pursuing, an

281
00:20:27,500 --> 00:20:31,800
ampersand lifestyle, it's
challenging, to be honest to

282
00:20:31,800 --> 00:20:37,500
have two activities that you
both love and want to keep up

283
00:20:38,400 --> 00:20:41,600
and you often have to make
trade-offs.

284
00:20:42,100 --> 00:20:46,700
But I think it's extremely
rewarding and I think from a

285
00:20:46,700 --> 00:20:51,400
practical perspective, I think
the the way to make it work is

286
00:20:51,400 --> 00:20:57,700
to figure out.
How to interleave them, in a way

287
00:20:58,100 --> 00:21:02,800
in sort of, in time and mental
capacity.

288
00:21:03,100 --> 00:21:05,700
Because they do tap into
different aspects.

289
00:21:06,300 --> 00:21:11,200
And there are complementary,
which means that it's actually

290
00:21:11,400 --> 00:21:17,200
in some ways, perhaps easier to
do the two things compared to

291
00:21:17,200 --> 00:21:20,900
doing one thing.
But twice as much, for example,

292
00:21:20,900 --> 00:21:25,100
after a long day of doing
research, I actually really feel

293
00:21:25,100 --> 00:21:30,600
rejuvenated to go and practice.
But if I were practicing all day

294
00:21:30,600 --> 00:21:34,100
long, I would probably not want
to practice or my fingers would

295
00:21:34,100 --> 00:21:39,000
tell me, not to do practice.
So there's there's kind of gaps,

296
00:21:39,100 --> 00:21:41,100
I guess it's a in computer
science.

297
00:21:41,100 --> 00:21:45,700
You can call pipelining where
there's there's overlapping in

298
00:21:45,800 --> 00:21:53,700
in some way because your body
has the ability to multitask in

299
00:21:53,700 --> 00:21:56,400
a way.
What do you think should be in

300
00:21:56,400 --> 00:21:59,600
the a saint Manifesto?
You know, there is a sense of

301
00:21:59,600 --> 00:22:05,900
self that I feel like I am me.
And the to the extent that all

302
00:22:06,200 --> 00:22:11,100
my activity stem from this
single source of Direction, I

303
00:22:11,100 --> 00:22:16,000
think makes it sort of easier in
a way, right?

304
00:22:16,000 --> 00:22:20,800
So I don't think, even though
playing classical piano and

305
00:22:20,800 --> 00:22:23,800
doing research are distinct
activities.

306
00:22:24,300 --> 00:22:30,400
For the most part, I feel like
the motivation and the way that

307
00:22:30,400 --> 00:22:35,500
they fit together in my life is
is really a synergistic.

308
00:22:35,700 --> 00:22:40,900
And I think maybe this is the
thing that makes it work for me.

309
00:22:41,100 --> 00:22:45,000
It's not that I have two lives.
I'm Jekyll and Hyde and they're

310
00:22:45,000 --> 00:22:48,400
just like two voices in my head
too kind of running around, but

311
00:22:48,400 --> 00:22:54,000
it's really one me.
And the one me sees both of

312
00:22:54,000 --> 00:22:57,700
these is contributing to my
personal growth, like

313
00:22:57,700 --> 00:23:02,900
professional growth.
And as I've mentioned, how the

314
00:23:02,900 --> 00:23:08,900
Quest for novelty, the ability
to see the big picture, there is

315
00:23:08,900 --> 00:23:14,200
a lot of ideas that get
transferred over between the

316
00:23:14,200 --> 00:23:17,500
different skills in machine
learning, we there's a notion of

317
00:23:17,500 --> 00:23:21,900
transfer learning where you
learn on one task You train a

318
00:23:21,900 --> 00:23:26,800
model on one task and it then it
can help you on other tasks.

319
00:23:27,500 --> 00:23:32,300
And this is I think is
especially true of humans.

320
00:23:32,800 --> 00:23:36,000
But it's interesting from that
vantage point to think of the

321
00:23:36,000 --> 00:23:41,400
role that music plays in
computers science and vice

322
00:23:41,400 --> 00:23:44,700
versa.
I love this idea of transfer

323
00:23:44,700 --> 00:23:50,800
learning and thinking about the
many sides of a human, you know,

324
00:23:50,900 --> 00:23:57,600
Feeding into the others Percy.
It has been fascinating talking

325
00:23:57,600 --> 00:24:00,600
with you today, folks.
You can check out Percy's work

326
00:24:00,600 --> 00:24:05,400
at CSD stanford.edu backslash
tilde P.

327
00:24:05,400 --> 00:24:12,000
Liang, that's pli.
Ang if you liked this show hit,

328
00:24:12,000 --> 00:24:15,200
like, subscribe and share with
your friends and fellow

329
00:24:15,200 --> 00:24:18,600
ampersands.
I got a chance to reflect on my

330
00:24:18,600 --> 00:24:22,500
interview with Percy, there were
so many nuggets of wisdom so I'm

331
00:24:22,500 --> 00:24:26,400
going to try my best to distill
down to my top 3 1.

332
00:24:26,800 --> 00:24:30,700
This whole notion of AI vs IA
was fascinating to me as a

333
00:24:30,700 --> 00:24:33,500
creative.
There's so much talk today about

334
00:24:33,500 --> 00:24:38,500
chap, GPT replacing humans, for
tricking humans, but the way

335
00:24:38,500 --> 00:24:43,100
Brucie is approaching this work
is around how we realize our

336
00:24:43,100 --> 00:24:48,000
full potentials as humans to
create and how Can a I be a

337
00:24:48,000 --> 00:24:52,900
support system for us to tap
into our greater or most fullest

338
00:24:52,900 --> 00:24:56,700
ability.
There is an aspect in here about

339
00:24:56,700 --> 00:25:00,500
The Human Experience of creation
and how precious and

340
00:25:00,500 --> 00:25:03,100
irreplaceable that is.
So, is there a way that

341
00:25:03,100 --> 00:25:07,100
technology can actually help us
get to the joy of creating

342
00:25:07,300 --> 00:25:12,100
faster to Percy talked about
pieces that you've come back to

343
00:25:12,100 --> 00:25:14,200
over and over again?
And how they affect you in

344
00:25:14,200 --> 00:25:18,300
different life stages?
I have felt felt this so much in

345
00:25:18,300 --> 00:25:21,100
my life and not just with music,
I perform.

346
00:25:21,200 --> 00:25:25,200
But with the Arts that I consume
one of my favorite movies is the

347
00:25:25,200 --> 00:25:27,700
sound of music.
When I was a kid.

348
00:25:27,800 --> 00:25:30,700
I identified with Gretel, my
teenage years.

349
00:25:30,700 --> 00:25:34,900
I saw my struggles in the easel.
I am now fully in My Maria

350
00:25:34,900 --> 00:25:38,600
gears, and maybe even
identifying more and more with

351
00:25:38,600 --> 00:25:41,500
mother abbess.
So, yes, go Climb Every

352
00:25:41,500 --> 00:25:46,700
Mountain.
Three this theme of one brain.

353
00:25:46,700 --> 00:25:49,200
One main came up.
Several times in our

354
00:25:49,200 --> 00:25:54,000
conversation, many of us a Sands
use our Two Worlds to bolster

355
00:25:54,000 --> 00:25:57,400
the other and even if
externally, they look like

356
00:25:57,400 --> 00:26:01,700
different worlds internally,
they're driven by the same brain

357
00:26:01,700 --> 00:26:07,400
and the same person Sort of
implicit in a lot of the

358
00:26:07,400 --> 00:26:10,200
framing, as well, you know, I
have my professional life

359
00:26:10,200 --> 00:26:11,700
primarily our computer
scientist.

360
00:26:11,700 --> 00:26:17,600
And I have this this other life
that keeps me sane, but in some

361
00:26:17,600 --> 00:26:20,500
ways, maybe that's not quite
right.

362
00:26:20,600 --> 00:26:22,800
And in fact, it's a two-way
street