Adobe Think Tank: The Importance of Diversity & Inclusion in Artificial Intelligence (AI)
- Articles, Blog

Adobe Think Tank: The Importance of Diversity & Inclusion in Artificial Intelligence (AI)

– Welcome to Adobe Think Tank again. Follow along and join the
conversation using #AdobeTT. And with me is Olga Russakovsky, assistant professor at Princeton. Thank you so much. – Thank you for having me. – And, you’ve got the swag on right now. – [Olga] Absolutely! – (laughs) Talk to me a little
bit about why you founded AI4ALL and, more importantly,
diversity inclusion, you know, at the K through 12 level too. Because I think that’s
incredibly important. – Yeah, absolutely. So, AI4ALL, we started it together with Fei-Fei Li and Rick
Sommer in 2016, actually. 2016? 2017, last year, last year. It’s only a year old. – Time’s flying. – Yeah, time’s flying, we’re already 2018. So it’s a foundation to
increase diversity inclusion in artificial intelligence, in AI. And we started because there
is really, really a problem in AI right now, and I would argue, actually, the biggest problem that is in AI right now is the fact that it’s being built by a very
homogeneous group of people. So there’s statistics like at NIPS, which is the leading AI conference. Last year, about 10% of
the authors were women. And that’s really a problem,
because we’re building this technology that’s, in theory, supposed to benefit everybody. It’s supposed to reflect everybody and that just can’t
happen if we are excluding large groups of a population systemically. – That voice is getting
lost in that conversation and then the creation
and the implementation and it just kind of gets,
it’s a chain effect, right? – Exactly, absolutely, absolutely. And so, AI4ALL really
tries to remedy that. And there’s several things we do. One of the big activities
is we start summer camps in partnership with universities. And every summer camp actually targets a particular underrepresented group. So we started with a camp at
Stanford for high school girls. Then we started a camp
at Berkeley last year for local low-income students. We’re starting four more this year, including one at Princeton
for racial minorities. And so, each of these
camps brings together a group of students to come
to campus for the summer for between one and three weeks, learn from AI professors,
from PhD students, and also get to know each other and build a support community amongst themselves. And then AI4ALL brings
all of these students into its alumni network,
provides continuous education, provides continuous support to make sure that it’s not just a three-week program, that it’s really has impact beyond that. – What age do you think is
the best to kind of start introducing students to AI? – Yeah, so we’ve talked about this a lot. So the Stanford camp, we
get girls after ninth grade. Talked about this a lot,
it’s sort of that time when they have enough
mathematical foundation to actually appreciate
some of the AI technology, but are sort of starting
to think about electives and starting to kind of
shape their career path. At Princeton, we’re
doing camp for students who just finished 10th grade. And the reason for that is
we’re looking at AI technology and policy, and we want students to have taken US Government, and that’s usually taught in 10th grade, so we want another year of
history under their belt before we bring them to camp. So, it kind of varies, but
these camps are largely sort of after ninth or 10th grade. – I mean, this is a
potential pitfall, obviously, in the, you know, the ecosphere of AI, but are there any other
big, potential pitfalls for, you know, the whole of AI as a field? – I think this is the biggest. I think this is, by far, the biggest. I think this causes bias in
algorithms, bias in systems. – This is humans programming machines, so that bias is always there. – Absolutely, and this is
humans, more importantly, who are deciding what problems to work on. So, largely, the problems
we choose to tackle as researchers are the ones
we’re passionate about. That is influenced by our
experiences as humans. How we choose to find
solutions to these problems, that depends on, kind of,
what our experiences are, what we think are the right solutions. Where we look for data,
that’s influenced by, again, our experiences. It’s influenced by our location. Like if we send a robot out
in the world to collect data, a lot of the time, what’s
the easiest thing to do? They will run around our office and the initial data collection
will be in our, like– – Yeah. Yeah, it’s what’s seen, what’s apparent. You know, talk to me a little bit about developing things in AI, right. So, what can you do to kind
of be a leader in this space and kind of create that
from the inside out? And create more leaders, right? – So that’s a really,
really good question. I mean, there’s many things you can do. Well, first of all you can support AI4ALL. That’s a big thing you can do. – That’s incredibly easy, alright. – Exactly. I think it’s very important to educate the next generation of students and I think that’s really
where the impact comes from is in how we do education,
how we do mentorship. So, making sure when
you’re educating students, so, certainly don’t paint AI as a field that only geniuses can be a part of. So AI research is really a
skill, just like any other skill. It can be–
– It’s a spectrum, right? – There’s a ton of things you can do inside of that spectrum. – Absolutely, but any of
the things we want to do, really, it’s all a skill. So you, you know, learn
about different technology, learn about different algorithms, you learn about different
coding frameworks, and you can build up that skillset and become an expert, become
a leader in this field just by, you know, it’s a
step-by-step learning process. And I think it’s very important
to emphasize to students that it’s not that we’re born doing AI, or we’re, you know, born
being AI researchers, or there’s something special about us that makes us uniquely
qualified for this field. No, it’s just that,
you know, we studied it for many years and now here we are. – Right. – That’s one of the big
things that I think, unfortunately right now,
part of the AI conversation is about, is particularly
with deep learning. Sort of, there’s an emphasis on, wow, this is a really special, unique field and there’s a little bit of a feel of it’s only accessible to the select few, and I think that that’s a problem. – Yeah, so it’s more about
making it accessible, obviously giving people opportunity, and then giving them ability, over time, to become leaders, right? But you have to start at a, you know, a beginning point, an introduction, right? And that introduction
is incredibly important. – Absolutely, and you have to continue supporting these students. You have to continue making them feel like they’re a part of the community. – Because if it’s just a one week thing, and then you drop off,
now you’ve lost them, so– – Exactly, exactly. – And so how can people get involved? – Well, there’s several ways. So one of the things we’re doing, well, of course, if you’re
a high school student, you can attend one of the camps. We are also working on releasing some online educational content, because really the other problem is that AI is actually not accessible to many people right now, and there’s a lack of
knowledge, awareness, understanding of how
to use this technology, and if we’re trying to build something that will change the lives
of everybody for the better, how can we do that if a
lot of people are afraid of this technology or don’t
understand how to use it? And so we’re developing
some online content with the hope of actually
reaching millions of people with this and this is both
kind of introductory modules to some of the technology, teachers’ support
materials, things like that. And we’re hoping to launch
that later this year. – That’s awesome. Thanks so much, Olga. – Thank you very much for having me. – And to learn more, again,
go to

About Ralph Robinson

Read All Posts By Ralph Robinson

1 thought on “Adobe Think Tank: The Importance of Diversity & Inclusion in Artificial Intelligence (AI)

  1. Someday a computer will give a wrong answer to spare someone's feelings, and man will have invented artificial intelligence.” ―Robert Brault

Leave a Reply

Your email address will not be published. Required fields are marked *