AI and Social Good – Intro

AI and Social Good – Intro


[MUSIC] Okay, let’s get started. So it’s good to see so many
friends and colleagues here. Special thanks to all of you
who have traveled from far and wide to come here. For those of you who don’t
know me, I’m Sriram Rajamani. I’m the Director of
Microsoft Research India, and I’m very pleased to
welcome you all. The attendees for this workshop
include academia from all over the world, labs,
NGOs, and startups. And you’ll hear from all of this
cross section of people over the next, today and tomorrow. So this workshop started in
a conversation with Narahari, Milind, and myself. In fact actually, you held a similar workshop
earlier this year.>>April.>>April and then this was follow on
conversation from that. And Amit Sharma, he joined us,
he has a PhD form Cornell and he just joined Microsoft
Research two weeks ago. And he wanted to take a break
between his stay in the US and here. And we asked him to organize
his workshop during the break. [LAUGH] And he thankfully
did a wonderful job, so thanks to Amit for doing that. And then you all must have
met Satish, Siddharth, and Chris who would have
been in touch with you. So as far as I’m concerned, my
goals are to really get to know you, get to know each other and
each other’s work, and establish collaborations, particularly
interdisciplinary ones. And I’m very grateful that
the gathering here consists of not only computer scientists,
but social scientists, and even within computer science, various
disciplines of computer science. And I’m hoping that
this gathering will set the foundation for
future meetings. And more broadly, that we can
get at least some conversation started on framing this topic on
what AI and social good actually means and come up with an
ambitious agenda for ourselves. So that’s actually
my hope from this. And I’m hoping that for most of you that’s the hope in
you coming to this workshop. The format of the workshop
is there’s gonna be brief presentations and
panel discussions. All talks are gonna be short. So we wanna focus
more on discussion, rather than presentations. And these are focus areas that
the company has put together. I think there’s fascinating
talks and each of them have talks with people from very,
very diverse perspectives. I wanna say a little bit about
Microsoft Research India’s own journey in this space,
though we didn’t call it AI for social good. [LAUGH] And
we’ll say more about that later. So we have been working on
technologies for development for over 12 years now from
the inception of the lab. And societal impact has been
in a core DNA of the lab from its founding. Here are examples of things
we have done over the years. We have experts thinking about,
even understanding users, actually even thinking about
interventions of projects. We have people who study users
in its own right, right? We have some of the world’s
leading work on how to think about user interfaces for
illiterate and semi-literate users,
and people in maybe. More recently we have people
like Jacky O’Neil who studies workers in modern work
platforms like Ola and Uber. And this is just ethnographic
studies where people just sit with these users
instead of buying what they are facing, right? More like pure science that’s
what I think of this work. And we’ve done work in health. Many of you might know about
abilities as well as 99Dots. Transparency is an area that
we’re increasingly getting into education, employment,
agriculture. DigitalGreen was an NGO that
was incubated out of a lab. The connectivity and so on. So this is our background
in this space. And in this workshop,
you will hear the dots that are highlighted
in red, the automobile one. Karya was just dignified
the digital labor, Colin Scott will give
the talk FarmBeats, which is a technologies for
precision agriculture. Is gonna give the talk, you’ll hear some of her
work in this workshop. I think of particular
fascination to me is actually how to scale
these things, right? It’s one thing to actually do
a research project in the lab, and the other one
is to deploy it. And when that happens
we learn a lot more than just being in the lab. And I’m very proud to say
that our lab is scaled, there’s always a ratio between
this list and this list and that’s by design. I mean it’s very hard to think
about several things but only a small number of them we can
even attempt and even a smaller fraction of them actually
make it and become scale. So DigitalGreen was spun off
into NGO in 2008, it actually runs in thousands of villages in
India and all of the world now. Rikin Gandhi is the CEO
of the organisation. 99Dots spun off into a forward
profit company, Everwell, interesting and you could
actually ask Bill as why he spunned off a forward profit
company, I don’t know. I think Bill will walk
in I think later today. MEC was an online learning
platform that we built and it’s being productized by Microsoft
and it’s called Sangam. They have integrated
with LinkedIn, and even though we started it
for undergraduate education, it is now being productized for
skilling and blue-collar labor. IVR Junction, MultiMouse. So these are things that
we have managed to scale. This is one of the reasons
why we invited entrepreneurs. Because part of the scaling
involves not just science but entrepreneurship. I wanted to say a little bit
about what we have learnt in the process. We just thought
of step back from all these projects we have done. Here are insights which
is in some sense right, our view has shifted from
how can new technology help to how can old and
existent technology help. And particular, Bill East has coined this
term as frugal innovation and he thinks about things that,
like everyday things like paper as technology to help solve
these very difficult problems. And this is actually a nice
perspective where we sort of got away from thinking about
illiteracy as just an inability to read and write to a lack
of cognitive skills. Which is a much more deeper
understanding of the problem if you want to reach and
be inclusive about technologies. And I won’t go
through all of them. Another thing of particular
interest is actually thinking about who as just not
consumer’s information, that just simply receive
information from us. But as really producers
of information. And for
us to learn from each other, and not just disseminate
information to them. And projects like DigitalGreen, IVR Junction, are really about
hearing from this population rather than just telling
to that population. And thinking about technology
not only in terms of deeds like health and education which is
altruistic in us to help people. But really think about
the population as one which is aspirations, I’m thinking
about their wants things like entertainment, employment which
are really much more wants instead of having that
distinction in mind. And really creating a scientific
discipline in this space is also extremely important and I know Milind has more to
talk about that as well. And so the community
that we are apart of, we started a conference
call ICTD which MSR India was integrally
involved in creating. And the next ICTD is going to
happen in Pakistan next month and it started out
in Berkeley in 2006. It was held in Bangalore,
in 2007, and so on. And this is the community
where we come from, alright? So I want to say something
about my use of the term AI. Where sort of, I sort of
think about this as tech for goody following into cloud for
good and drum rolls please, I think now we have
sort of AI for good. But I think, but
in reality I also think that the AI has significant things
to offer in this space. If I think about speech natural
language understanding, I think they are just so
key to reach this community. Even I when I hear a million
percent about his work on social network analysis,
recommendation systems. I see actually so many possibilities for
algorithms from AI. And it’s already happening and
there is a lot more potential. And also I think AI has become
such a core discipline, that I think it has become not
only just, its algorithms and logical reasoning and so on,
and statistics and so on. But I sort of view AI as
a more interdisciplinary area, which interacts with all
areas of computer science as well as social sciences. So in my part, I hope you get
you forgive sort of my broad use of the term, and I know that in
conversations with Milind, Erik and Narahari, I think we all
share that passion, that this is more an interdisciplinary work
rather than AI in its core. And I wanted to
show this slide for why I think such a workshop
is timely right now. [LAUGH] And why you’re
holding it in India, because it’s been now 12 years or so
since I returned from the US. And I have seen technology
adoption in India accelerating. Digital technology is pushed by the government
in a massive scale. That are now has about a billion
people enrolled even though a lot of us have concerns about
security, privacy, and so on and there is 1 billion people
enrolled in Adar and there is something
called India Stack. So these are all things that are
societal scale technologies that the government is pushing. And if I look at businesses
right, gone are the days where our country was mainly doing
services for the best. But we are now building
consumer products, targeting the same population
that we want to work with. And intense competition there,
and during this process of actually thinking
about AI and social good, I have actually been talking
to certain impact investors. And right near my office, there
is actually Omidyar Network, and several others, who do something
called impact investing where their goal is to actually
not get just returns but also social returns which they
measure in terms of impact, it’s very interesting. Another thing that is
interesting is that there is India there is a law
called company’s act which mandates that every company give
away 2% of its profits and if we don’t give it to the government,
we decide how to spend it. And that I think has also spawned off lots of
activity in this area. So I wanna say a little bit about possible models
of collaboration. This is based on our
own experience working with community. So in the past, we used to have summer schools
where we used to get lecturers from all over the world mainly
to help educate students. We are now thinking about how
to transform that into not just a one way lecture style, but
more as a collaboration. And last year,
we did a summer school in IOT. I think at ISC,
it was done in ISC. But in addition to the lectures,
what we did was we recruited some of the students
to not only after the lecture, after the lectures, come spend
the entire summer with us. So people, a subset of
the students who attended the summer school has spent
an entire summer with us and in fact, pulled the projects
that you’re going to hear today, the one on Road Safety and
Precision Agriculture where it actually started with those
students in the workshop. This summary went
even a step further. And we did a summer
school on artificial social intelligence, right? Where, as part of the summer
school, as the call for summer school. In addition to students, we invited proposals from
faculty to do projects. And there was a company that
review those proposals and we selected four of those
projects, and the faculty and the students as a team
spent an entire month with us at MSRI in addition
to having lectures. And then on top of that
actually after one, one we had a jury and I think
was kindly third of the jury. And we reviewed those projects,
and we picked two to fund later, right? So Pretty Jorti’s project on
accent adaptation from ASR system she’s from IT Bombay and
this project by Supta Shi on utilizing social media,
we continue to fund those projects and
they collaborate with us, right? So I’m saying these things and
it’s another example of things that we’re doing is actually a
virtual center, suddenly, I find there’s a lot cryptographers in
India, I think that’s good, so we decided to form a Crypto
center where it’s. And all of these, we put our money behind
all of these things, so then we sort of not just
think about dissemination but really collaboration. So these could be models
of projects that come out of this discussion. There are various mechanisms for
us to support them, right? So that’s all I have, and
I just wanna end with goals. And I’m really looking
forward to learning, yeah. Thank you so much, and
thank you all for coming, yeah.>>[APPLAUSE]