00:03
Jenell Pizzaro I'm Frederick Philip on Vice and
00:06
thank you for consuming the Thunder
00:07
nerds a conversation with the people
00:09
behind the technology I love what they
00:11
do and do tech good and we are with Josh
00:16
Clark Josh oh so excited have you want
00:18
to show I'm really happy to be here
00:19
between you two Ryle is just maybe
00:26
pumpernickel he knows
00:29
don't you remember bread Josh let's talk
00:33
but it's colorful and delicious right
00:36
it's a good call yeah yeah it's very
00:38
good with tuna salad Josh don't let our
00:43
audience a little bit about you if you
00:44
don't mind
00:45
besides my love of pumpernickel I'm not
00:47
an interaction designer and founder of a
00:50
design studio in Brooklyn called big
00:51
media we focus on design for what's next
00:54
so trying to help organizations to
00:57
figure out how to adapt the emerging
00:59
technology that's here right now to make
01:01
meaningful change and hopefully make the
01:03
world a better place along the way to
01:05
all that so how are you guys doing that
01:07
how are you making the world a better
01:08
place but some of the aspirations yeah
01:11
well you know I mean a lot of it is how
01:12
can we make technology that focuses
01:15
attention instead of distracts it that
01:17
means to our lives in or out of the
01:19
reverse I think often we find ourselves
01:21
bending to technology some part of this
01:24
stuff is how can we create experiences
01:26
that you know sort of fit into a human
01:31
worldview instead of cramming humans
01:33
into a technological one so for a lot of
01:35
the last 10 years a lot of that has been
01:37
a focus on how do we sort of push the
01:40
frontiers of mobile and use all the
01:42
interesting sensor packed stuff to make
01:45
appealing mobile experiences but I would
01:48
say in the last couple of years there's
01:51
been a real shift into how do we use
01:55
machine learning and artificial
01:56
intelligence do we do things that that
02:00
the machines can do interesting or
02:03
useful things either to take sort of
02:05
menial tasks off of our plate or perhaps
02:08
to give us insight and sort of amplify
02:10
our judgment and our
02:12
ability with the kinds of patterns that
02:15
they're able to to find so a lot of that
02:19
stuff can be in the healthcare realm
02:21
it's some of the stuff that we've been
02:23
working on and she defined patterns
02:25
around that so it's a few different
02:28
aspects I'm a passionate about the new
02:31
kinds of interactions that we're able to
02:32
have but also we look at the problems
02:34
that we didn't solve without hopefully
02:37
making new problems which is often the
02:40
way the technology goes it's Werner
02:42
worse yeah and and do you think that was
02:44
a lot of your concept with like writing
02:46
designing for touch so you're you know
02:51
essentially like trying to solve maybe a
02:53
problem like or even posing like maybe
02:57
solutions for machine learning and and
03:00
growing with humans
03:02
you know like almost like a graduation
03:04
from touch devices and connected devices
03:06
yeah well if you think about it it's
03:09
like machine learning is powering all of
03:11
the emerging technologies that are
03:13
really interesting right now it's gonna
03:15
the oxygen of the interaction world
03:17
right now so all the speech recognition
03:20
natural gesture predictive interfaces
03:22
augmented reality all that stuff relies
03:25
on this kind of pattern finding that
03:28
machine learning is enabling so it's
03:29
sort of it's underneath everything yeah
03:32
and so as with designing for touch touch
03:35
was interesting because it suddenly
03:36
created this illusion that information
03:39
was physical that you can drag it around
03:40
squeeze it pinch it you know it's sort
03:43
of this useful illusion or piece of
03:45
misdirection and how we're at the next
03:47
phase where we can sort of talk to
03:49
information as if it is a person hey
03:51
Jerry yeah I'd like to have this
03:54
information or if people were to barking
03:56
Alexa Alexa it's an interesting way that
04:00
all the different platforms talk to each
04:01
other - hey Google hey Siri and I think
04:05
this goes to your your talk the new
04:07
design material right yeah that's right
04:10
I'm giving a talk it's a very brand new
04:13
talks I talk about brand new things to
04:16
think about how do designers work with
04:19
machine learning as a design material
04:21
and I think that's the way to think
04:23
about any new technology is howdy what's
04:26
the new stuff that this material does
04:29
and how do you work with it in a way
04:31
that the material wants you to work with
04:33
it what is the machine morning want ya
04:36
and how does that marry with what people
04:39
want yeah well how would you design you
04:42
know maybe something small for Alexa how
04:45
what is your first step for making an
04:48
Alexa what are they called done tasks
04:50
your skills yeah right you know I think
04:53
with anything it's the first thing in it
04:56
but let me back up I think that it's
04:57
okay to build some silly toys first
05:00
especially my yeah most human at play
05:06
and we tend to sort of put aside the
05:08
demands of what are the the metrics and
05:12
okay ours that we need ahead of all this
05:14
stuff and more sort of like really
05:16
playing with a material so I think that
05:19
it's okay to first think it's like well
05:21
what would be a fun silly thing to play
05:23
in part because it pulls us out of our
05:25
rut how do we usually think about design
05:28
problems because I think that when you
05:29
when you look at who had the best
05:31
interaction designers in the industry
05:33
frankly video game designers yeah they
05:37
bring a lot because their challenges
05:38
they have to make something that's novel
05:40
and fun and that you don't have to read
05:42
the instructions for that sort of the
05:44
thing explains itself as you use it and
05:47
so I think that making games and toys
05:49
can be a great way to as a side project
05:53
get to know how to work with with a new
05:57
technology as a design material so I
05:59
think that if you're sort of new to
06:01
making Alexa scale or working with any
06:03
kind of new emerging technology making
06:06
something a little playful is a good way
06:08
to start well it's not what you want to
06:10
release as a product again we have a lot
06:13
of frivolous things that are being
06:15
pushed out as products and so I want to
06:17
make a distinction in our practice let's
06:19
play so that we can learn how to make
06:21
meaningful useful products that don't
06:24
just sort of add to the absurdity of the
06:27
world which is something I think useful
06:28
that Kate O'Neill talked about it or not
06:31
care about tech humanism yeah
06:33
you're saying create Alexis skills that
06:35
you know repeat what you say but in
06:37
piglets
06:43
you know I think one of the things that
06:45
I think about a lot and working with a
06:47
new technology is what is the what is
06:50
the grain of this new design material
06:53
what is it enable and what is it
06:55
encouraged and what's the new thing of
06:58
it it takes a while to figure that out
07:00
but when you think about speech
07:01
interfaces they're sort of like well
07:03
what are the new physical contacts that
07:06
it enables that you wouldn't be able to
07:08
do with the screen what are the
07:09
constraints of it that's the emotional
07:11
context of it what does it mean that
07:13
you're talking to this disembodied the
07:16
personality floating in the room
07:18
and there's something yeah what's it do
07:21
to the people around you so we have an
07:23
echo at home and it understands me and
07:28
our 18 year old daughter Veronica ninety
07:30
percent of the time pretty good yeah it
07:33
understands my wife Liza never yeah Oh
07:37
interesting and so you know we hooked it
07:39
up to some stuff in the house to turn on
07:42
the lights or whatever or you know it's
07:43
yeah Liza will come out in the morning
07:45
and be like Alexa turn the lamps on and
07:48
it's like playing children's Christian
07:49
music you know it's a I'll try to help
07:52
you know Alexa turn the lamps on it'll
07:54
work and she's like oh your girlfriend
07:59
awkward for us to talk to Alexa so
08:05
there's this there's this sort of
08:06
there's this real thing that happens
08:08
which is just it is just a computer
08:10
system oh yeah but it has taken on this
08:13
personality of this unwelcome person in
08:15
our house something you have to consider
08:18
in the design of these things I mean
08:20
what goes wrong how does it change
08:21
relationships how does it change us yeah
08:24
yeah how does it change us as we grow up
08:27
with it one of my buddies his son is
08:30
three years old and he's always been
08:32
able to tell time by going Alexa what
08:35
time is it and he was talking about he
08:37
was in the store and his son just said
08:40
Alexa what time is it just expecting
08:43
that to be there like
08:44
when does that actually happen yeah yeah
08:47
well I think with kids it's really
08:48
interesting because you know I think
08:51
early on there was some concern with
08:52
Alexa of are we teaching kids like not
08:55
to say please and thank you you know
08:56
it's just like giving orders Alexa what
08:58
time is it you know it sort of this and
09:01
so Amazon has come out with Alexa for
09:03
kids which is a little nervous making
09:05
right sure but it will when it's sort of
09:08
in kid mode you'd sort of say you know
09:10
you could say please or you know you
09:12
could say thank you so you don't have to
09:14
but it's sort of these prompts around it
09:16
it is cute I think it's useful but it's
09:18
also sort of a thing it sort of also
09:20
assumes that there's a one-size-fits-all
09:22
parenting mode yeah and I think look
09:25
we're at the beginning of all of this
09:26
but it's the kind of thing of like do I
09:30
want Amazon to do my parenting and even
09:32
if so is Amazon are there sort of
09:36
parenting values there designers the
09:38
same as mine right would they would do I
09:40
want Alexa correcting my kid and if so
09:43
how it's like it's a whole source of
09:46
different interests Wow
09:47
that's a separate relationship it's my
09:49
kid having with Alexa over there when
09:51
I'm not around and cultural values to
09:54
you know like you know you know Asian
09:58
culture right that would be weird as
10:01
well too like maybe put an Alexa and a
10:03
Hispanic you know household and it's
10:05
completely different than you know what
10:07
it expected it to be right bring up a
10:09
whole huge important area sort of you
10:13
know of the machines know what we teach
10:16
them know it's like what is the data
10:18
that we're providing to them and who is
10:20
doing that data and frankly you know you
10:22
look at a lot of caman computer of
10:25
vision face recognition stuff and it
10:27
works great on pale males you know pale
10:30
males actually I would say everything
10:31
looks great on Pam yes Alexa listens to
10:33
me more than right yeah it's like things
10:36
has been so many things cut my way
10:38
because I'm a straight white middle
10:41
class educated American man oh man
10:45
things go my way all the time and in a
10:47
way we're sort of embedding that bias
10:49
into these systems because that's what
10:51
the data is and that's the data that's
10:53
been gathered and so is this something
10:55
that as we do this are these systems
10:57
just gonna rien
10:58
worse my power and privilege at the
11:01
expense of others or is it gonna
11:02
steamroll culture as as Google suggests
11:06
phrases to use to respond to email does
11:10
that mean that there's going to be a
11:11
very specific Silicon Valley vernacular
11:14
that gets amused and spread and is that
11:17
going to flatten you no other dialects
11:22
yeah would you also say to to a certain
11:24
point where you look back at computer
11:28
language and you look at maybe the last
11:30
30 or 40 years how most things that you
11:34
would write you would write in English
11:35
good right HTML on Englishmen no matter
11:38
what country you are from you wrote HTML
11:40
in English yeah you know you still
11:42
spelled color if you're in the UK with
11:45
without a you because that's that's the
11:47
HTML way to do it so do you think like
11:50
there's some kind of parallels there
11:51
what so right I mean in that case that's
11:53
sort of like it became the lingua franca
11:56
sorry franca sorry French you know you
11:59
know and it's English is the technology
12:01
language and so at some point you need
12:03
to sort of choose something especially
12:05
if you're gonna give instructions to
12:07
computers this is how it is but I think
12:08
you're right is that it does create a
12:11
bias of access to who can code that you
12:14
have to have that baseline understanding
12:16
of this other language in addition to
12:18
the logical language yeah Wow but the
12:21
risk is of course that we're also
12:22
creating those kinds of decisions
12:25
inadvertently and probably with the best
12:27
of intentions but about who can use
12:29
these systems or do I recognize myself
12:31
in these systems where's the system
12:33
recognize me you know there was a study
12:36
of Google speech translation from 2016
12:40
so only a couple years ago they found
12:42
that there's a bias against
12:44
understanding women so 70% of the time
12:47
the system would understand the speech
12:50
of a random woman better than of the man
12:53
despite the fact that women's tend to
12:55
speak more slowly or clearly longer
12:58
vowels grateful recognition for these
13:00
systems did not this is clearly a case
13:03
of just bad data skewed toward
13:06
understanding dudes because that soon
13:09
presumably don't this system
13:11
so a lot of this as we think about how
13:14
do we create these machine driven
13:16
systems as how do we give them the data
13:18
how do we make sure that we are aware of
13:22
our own blind spots and actively trying
13:24
to fill them and a lot of that goes to
13:26
the teams who do we work with mm-hmm
13:28
the teams cannot look like me you know
13:31
it's like and I don't mean just in terms
13:33
of demographics race ethnicity gender
13:35
age that kind of stuff but also you know
13:39
almost worldview yeah how do we get
13:41
outside of the tech bubble and bring in
13:43
philosophers and civil servants and
13:47
artists and every walk of life because
13:50
this stuff is affecting our entire civic
13:53
life now these systems determine how we
13:56
speak to one another
13:57
yeah it's which really shaping our
14:00
global culture yeah right and and let's
14:03
at least do that with intention and not
14:05
like machines decide to do I like to say
14:08
that the future should not be
14:10
self-driving you know we need to sort of
14:12
decide where we want it to go or the
14:14
machines will decide for us mm-hmm I
14:17
mean we can see it too many addicts like
14:20
in Ed's you can see but you know towards
14:22
the end or the beginning to add they're
14:24
talking to Google they're talking to
14:25
alexa in the ad right so that like Alexa
14:29
can hear you like the commercial can
14:31
hear the commercial itself right and
14:32
then so Alexa is gonna be like okay cool
14:34
I heard that gun at like I got it down
14:36
right and so you're being marketed to
14:38
multiple ways just through like your TV
14:41
now because there's this device that is
14:44
is part of your family really like
14:47
that's kind of can easily be punked we
14:58
should all Google parents talking to
15:00
alexa it's like there are all these
15:02
cases Elect it's a parrot's adding stuff
15:07
to shopping lists turning lights on and
15:09
on really because they hear you I'm
15:13
talking to your Alexa and so they just
15:15
pick it up right sure they're just like
15:17
yeah
15:21
like it's so looks like order pizza for
15:23
her dachshund and Alexa can't tell the
15:31
difference I mean that's they're getting
15:33
better at it but this is a an
15:35
interesting nuance of designing for this
15:37
stuff that in terms of sort of design
15:39
and for the grain of the technology the
15:42
grain here is that the machines are
15:44
weird and so the design that we have to
15:46
do is not designing fixed paths through
15:50
information that's under our control
15:51
anymore we used to say design for the
15:53
happy path now how can you design it for
15:55
error and failure and how do you recover
16:00
from that gracefully and help people
16:01
understand what's going on when they get
16:03
bad or suspect right you don't want to
16:06
be technically wrong and that sums I
16:08
mean it's a book but it's a it's about
16:11
like how machine learning can be you
16:14
know biased or just completely behind a
16:17
lot of electrics
16:25
yeah um how do you go about you know
16:28
designing a skill or you know anything
16:33
like a machine learning kind of
16:35
situation for lookup big enterprise like
16:38
how can they use Alexa's and and you
16:42
know their Google assistants etc to make
16:44
their either their jobs easier or you
16:48
know their product yeah I mean I think
16:50
that there's always two questions that
16:52
you ask when you're about to build one
16:53
of these things is how do I build the
16:55
right thing and then how do I build the
16:57
thing right in this often we don't spend
17:01
enough time on how do we build the right
17:03
thing it's often sort of something that
17:05
somebody has an idea and then it's like
17:07
we're gonna make this thing and I think
17:09
that the the work of design especially
17:11
about UX researchers is trying to figure
17:13
out where can we actually make the most
17:15
impact not just the impact for the
17:18
business but in the impact of the
17:20
person's life who's using it and I think
17:22
that whether it is sort of some of the
17:24
efficiency of talking to alexa in a
17:26
moment anywhere hands-free eyes-free and
17:29
what are those opportunities and we're
17:31
in sort of the customer journey can you
17:33
have the biggest impact to of answering
17:36
a question quickly and casually in that
17:39
way but I think there's also sort of an
17:42
interesting thing as much of machine
17:43
learning generally which is you know
17:46
what are the moments where we find
17:48
ourselves doing error-prone joyless
17:51
repetitive tasks but the machines can
17:56
sort of do that part and bring some new
17:58
insights so that we can do the part
18:00
that's not horrible that's the story of
18:04
technology and it it sometimes has
18:09
unintended consequences to that we can't
18:11
always predict ah who knows what's gonna
18:14
happen in the future yeah yeah yeah but
18:18
the other part of building and riot is
18:20
the great news is that all of these
18:23
technologies certainly including lots of
18:25
skills really low attends a really sort
18:28
of low barrier to entry it was like if
18:31
anybody who has heard some basics or
18:34
front-end development skills can quickly
18:37
sort of get
18:38
mr. berries are a simple hello world
18:40
Lexus skill go in and so there's the
18:44
hard work and I work with clients a lot
18:47
on sort of product invention exercises
18:50
to sort of figure out how do we make
18:52
sure that we're building the right thing
18:53
where do we build this in there but once
18:55
you do it you can also get really sort
18:57
of get prototypes up really quickly on
18:59
these new speech platforms but also for
19:03
other stuff it turns out Amazon and
19:06
Microsoft and Google and IBM are given
19:09
away their machine learning API is at
19:12
least sort of common uses of so it's
19:14
like you don't know how to use do
19:15
computer vision camera vision no problem
19:18
it's like okay eventually for free it's
19:20
like the guts of what make LexA go you
19:23
can get at that essentially for free or
19:25
at low cost from Amazon they're giving
19:27
this stuff away because they want your
19:30
data they want to get better at it
19:31
yeah so at least for prototyping stuff
19:34
that's a great environment yeah it's
19:36
really cool to UM I know there's Asher
19:39
and there's like this thing where it can
19:42
like tell who is who and what is what
19:45
like could literally find celebrities in
19:47
a crowd like oh that's Bernie Sanders
19:48
right like that's the coolest freakin
19:50
thing not only can I find Bernie Sanders
19:53
an identified Bernie Sanders in a crowd
19:54
it will also tell you what his emotional
19:57
state is an estimated age and so there's
20:00
like all that stuff of analyzing video
20:03
or photo faces it's you know it's all
20:06
very accessible yeah so I think that the
20:08
trickier part at the moment is there is
20:12
a whole sort of set of stuff that you
20:13
can do when when you want to build your
20:16
own models in your own sort of stuff but
20:18
for a whole vast number of particularly
20:20
consumer facing technologies stuff is
20:24
available for you to play with today and
20:25
with no sort of prior knowledge of you
20:28
can save money you really can I think
20:30
you should get on some of those silly
20:31
and finally I remember they were talking
20:34
about there's a lot of silly fun things
20:35
to be committed yeah I have a friend who
20:38
had made a Twitter bot because he you
20:42
know wants to be a little bit more of an
20:43
hi Duane a little bit more visible on
20:46
Twitter and so he uses Instagram and so
20:51
he takes pictures of things
20:52
you know that he's doing right like
20:54
coding or going to meet ups etc and take
20:57
a picture of it and runs it through
20:58
auger and it'll post it on Twitter and
21:01
they'll say hey this is the thing that
21:03
I've made right like it's like oh it's
21:05
twelve people in front of a bus getting
21:07
a taco right and it'll be like is that
21:09
accurate or not right and I think that
21:12
is the coolest thing it's just a he just
21:14
wanted to be more visible on Twitter and
21:16
he created this like like recognition
21:20
thing just just for funsies so I think
21:22
being hilarious it's like it's really
21:35
sort of charming sometimes and that's
21:37
what I mean by the fact that the
21:38
machines are weird is they see the world
21:41
in different ways and we do and so I
21:42
think that there's a lot of fear that
21:44
the machines will somehow replace us
21:46
they're sort of the Hollywood version of
21:48
that and they understand the world so
21:51
differently and with such a sort of a
21:53
different perspective I think the better
21:55
way to think of it is they can actually
21:56
just be like really interesting and
21:59
useful sidekick companions yeah to us
22:02
that in the best form helped to amplify
22:06
our judgement and engage it in
22:08
interesting ways that we can be better
22:11
or more interesting versions of
22:13
ourselves instead of having to do in
22:15
sort of some of this long we can look to
22:17
the sound a little back yeah well said
22:19
you know I wanted to ask you about one
22:21
of you are a very interesting creation
22:24
from early on look at the couch to 5k
22:37
held to be the one to organize everybody
22:40
to run every time you get all a comprar
22:41
you know I only started doing it
22:44
recently and it's something that I found
22:46
it's can go anywhere sort of and it
22:49
turns out to be a nice way for people
22:52
who were like me gonna go out for
22:54
anybody to do it together and we have a
22:56
really casual with talks it's like so
22:59
speakers and attendees about 10 people
23:02
today which is a record it's like a lot
23:04
of people will come out at 6:00
23:05
more than yeah after the conference
23:08
party the night before yeah but yeah and
23:12
and so yeah but thanks for mentioning
23:13
couch to 5k that was something that I
23:15
wrote this schedule a little over 20
23:17
years ago oh my gosh really hasn't been
23:19
that long and I wrote it for my mom to
23:21
sort of help my mother was the original
23:24
couch to 5k she's still doing it like it
23:31
she's in her seventies she walks more
23:33
now but she's like walking half
23:35
marathons oh my goodness yeah yeah it's
23:40
great but since then couch to 5k insult
23:42
like millions of people it's been
23:43
adopted by the UK's National Health
23:46
Service as a National MS plan it's grown
23:50
so far beyond me and I really can't take
23:53
credit for it it's for all the people
23:55
who've found some inspiration image to
23:58
make some changes or try some new things
24:01
in their lives so it's it's been a cool
24:03
thing oh that's so neat yeah I love that
24:05
that's something that you were able to
24:06
take you make something and now it's its
24:09
own kind of entity out there helping
24:11
people it really shows that sometimes
24:13
you just sort of drop these seeds and it
24:15
grows up into something much bigger than
24:17
you would imagine it's a it's been
24:20
genuinely a humbling thing to see this
24:23
idea that I put out there almost like my
24:25
original open-source project that just
24:26
yeah I haven't made any of the apps or
24:29
podcasts and yet this whole ecosystem of
24:31
apps and podcasts and websites and
24:33
communities growing up around it I would
24:36
have a sort of the core schedule itself
24:38
I'm so amazing yeah you'll be able to
24:40
enjoy the felicity of that for time to
24:43
come it is amazing just hearing people's
24:45
stories about because it's not just sort
24:47
of oh I ran a 5k it turns out that for a
24:50
lot of people with Fitness they run into
24:54
it it's a defeat you know it's like I
24:55
tried to do this I couldn't do it I am
24:58
NOT this is not something that I can do
25:01
it's not something to have within me and
25:03
it's sort of because you run into these
25:05
defeats at shapes just your perception
25:08
of yourself yeah and one of the things
25:11
that I've heard from people who found
25:12
it's like wow I am now able to jog for
25:15
30 minutes I
25:18
got a new job I changed my relationship
25:20
like I've heard these stories of people
25:22
who this experience changes their life
25:25
so it's an interesting thing of that I
25:29
think that you can think about with
25:31
design in general is this thing that you
25:33
are creating may have ripple effects
25:36
beyond what you're doing this could be
25:38
terrible effects it could be very
25:40
positive effects you don't necessarily
25:42
know yeah you can sort of do your best
25:45
and sort of try to corral it in that