- Hey, welcome to Adobe's Think Tank. We're gonna get through this interview pretty quickly because I want to get back to watching Wild Wild
Country in the hotel room but with me is Jake Porway, founder and executive director of DataKind and it's kind of the easiest
question in the world to ask first but I kind of wanted to talk to you about data for
good and just using data for the best of intentions. - [Jake] What does that mean? - What does that mean, right? That was like kind of a question
b
ut not a full question. That was me going like hey, let's just, let's just throw some stuff. - Yeah, sure, no worries. Are you testing right now or are we going? - Oh we're totally going.
- Oh great, just making sure. I was like, should I give a test answer? Oh great. - It's so comfortable
that we can just rip off. - I know, make it so comfy. So, you know, we talk at DataKind a lot about doing data for good and using AI in the service of
humanity and the thing is we say the best of intentions
be
cause so many people out there are realizing they
can give their skills back. There's opportunities to help
nonprofits and governments use data to predict where
famines are gonna occur or get clean water to people more safely but it's actually tricky to do, you know, sometimes we try our
best and get out there and say we're gonna
volunteer some time out there but doing it right can
actually take a lot of work and a lot more practice
than you might expect so we're always trying to
help people do
it better whether it's with DataKind
or in their own companies. - Yeah and talk to me like
maybe through a sample thing or just kind of that
process of how do you make that more efficient, how
do you make that effective? Because, again, you probably don't know the unintended consequences of trying to get into a space but
what does that look like? - Yeah so the way that we
work at DataKind, as you say, we're a volunteer organization,
we get data scientists to go team up with
nonprofits to help so
lve their data science machine
learning challenges. And so actually a lot
of people think great, just give me a project, let's just jump in. but, like you said, we don't always know the right thing to build, we don't know how it's gonna be used and, importantly, with this whole conversation about ethics in the AI space, we want to make sure that we're thinking really critically about how we're building algorithms that don't cause harm
so what we'll often do is we have a pretty long kind
of get t
o know you process, a scoping process where
we'll bring in NGO's that say we think we
could use machine learning or AI and then data scientists sit down and sort of listen to their challenges, knock on some data, see
what we can actually build and I'll actually say that's probably one of the most enriching
parts for both sides because a lot of
nonprofits come in and say, you know, my AI need is
a I need a new database 'cause that's like sort
of what they know, right, just like a corporate client
might say I think I need Hadoop but instead, a good data scientist may step on back and say well what are the
challenges you're trying to solve? We talked to one group that was trying to prevent fires that were occurring. Red Cross was saying we want to stop fires from happening where
they can be preventable so data scientists step back and say well how do you do that? They say well we just kind of go out there and put out brochures and
after kicking it around, they said ya know, the
data you'v
e got, yeah, the data you've got and
data we see out in the world could be put together to build
a kind of predictive model to show you where you should go and that was really kind of
the watershed moment for them. Scales fell from their
eyes and they're like oh we didn't even think
about predicting the data, well now we can do this and
that and the other thing so it's really that process
of talking with people, getting to their real
needs and understanding what the data can do to help
you build
a kind of AI solution that's really gonna be useful for humans. - That's good and that's kind
of like the reactive stuff but you talk about the proactive stuff where finding problems can actually be harder than finding solutions. Talk a little bit about that. - Yeah, that's right. - That stuff that's kind of unseen. - Yeah, so that's a great
way of putting it, unseen. I think a lot of times we see folks who come in and say I want to go build a satellite imagery tool that's
gonna help find poach
ers. You may hear that from either side, NGO's but a lot of the time, that's not as easy as just getting a bunch
of satellite imagery and running it, right, you
know, and just getting into that because you don't know who's gonna use it or how they're gonna use it or if the data can even support it. So while a lot of people
think donating their time involves just going to a nonprofit and saying let me work on this project, I think it's cool or working with you, we actually find that
finding the r
ight problem to solve takes this kind
of six component process we talk about where you have to find the right problem, you
gotta find the right data, you gotta find the data
scientists, you gotta find the person who's gonna act on it, fund it and even subject matter
experts who can say hey, that poacher thing
that you're building actually is a good thing or oh, actually if you tweaked it this
way, it wouldn't just work for the World Wildlife Fund
but a lot of organizations and so that's actually
, I
think, the trickiest part is not just saying who's the nonprofit I'm gonna work with and what do I solve? It's saying how do I really
get all of those components, the data, the problem, the
stakeholders, all around the table so you can really
design collaboratively to find the right ones so when we say finding solutions, or–
finding problems is gonna be harder than
finding the solutions, that's what we talk about,
setting the right thing up. - And it's a good way
of positioning it, too. By t
he way, most people,
when they list six parts to a solution, only get through like four, maybe four and a half. I don't have the data ahead of me but I'm pretty sure
that's about the right. - [Jake] Yeah, yeah. - You talk a little bit about how, what is the single biggest kind of barrier for nonprofit organizations? Because it could be scale, it could be a lot of things but what do you see? - I think, right now, what we see is there's so much energy. People want to be able
to use machine learnin
g and AI and they have
really great interventions. I mean some of these
nonprofits are transforming the way that we think about everything from sanitation to
disposal to curing poverty. The challenge they have
right now, I think, is one, resources, just having the money to actually engage with
data science and AI is incredibly expensive. - Right and convincing someone that that's the best allocation
of the money, right? - That's exactly right. You know, for people who
haven't worked in nonprofit
s, you don't get money up
front, right, people want to give you money to say I
want you to go save the whales and I want to know where
every dollar is gonna go. - Right 'cause there's an
accountability that you have as a nonprofit organization
that you don't have as a corporation because
those stakeholders want to know is 92% of
this, is 94% of this, is 45% of this going to
actionable stuff, right? - That's exactly right
so whereas a startup, they may say okay here's some seed money to figure ou
t some stuff
with, it's very hard to get that in a nonprofit
space so to go to a founder, or excuse me, a funder, and say hey, we need a half a million dollars to build an AI team, you get blank faces, like what is that gonna do,
why would I give my money? So that, I think, is one
of the biggest challenges and it's really a problem with the space and the way funding works
in the nonprofit space so that's why we offer
our services pro bono so that we can get over that barrier and you can just sta
rt working with expert Googlers right away. - Well that's the double edged causality of being a nonprofit is
that people will want to work with you because it's cause-based and it applies well to CSR but they also want to know exactly what you're doing. - Yes, that's a big challenge. - To close it out, because
this is really concise advice but what do you see the
potential for nonprofits in the next five years,
kind of leveraging it? What do you think is actually achievable? - I start to get a l
ittle
optimistic over this. - I like optimistic. - Oh okay great. - We've talked and we've been rooted but give me your most optimistic future. - Well in my mind, it's
that we don't just look for good values and data, we use
AI for our own human values. Almost every field could be
touched by AI in some way whether it's, again, transporting
water more efficiently or cracking down on vaccines,
cracking down on diseases or getting vaccination
rates up, it's everywhere and really what I'd like to se
e though is that it not be on corporations alone to say we have to give back to do this and not be on nonprofits alone to say we have to solve this problem but that we actually start creating spaces where the two come together. Nonprofits who say we know
the space innovation we need but we don't have the
resources can sit across from corporations who say
we've got great human capital, we've got data that no
one else has ever seen, we can provide some services. - Find those perfect mergers. - Yea
h, so to me, that's
my big kumbaya moment is that we're all coming
together to think about no just how can we make
little drips and drops of improvements on these problems but how can we really
push together forward to use data for social impact? - But we're at a conference
so it's always good to end on a kumbaya moment. - Oh good, good, I tend to always so yeah. - Well that's perfect. Thanks so much, Jake. - Oh yeah, my pleasure,
thank you very much. - And follow us along at #AdobeTT. Tweet us.
Thank you.
Comments