>> Jamie: We're going to get started. Lili Cheng,... and our unofficial chief social officer is going to be...

advertisement
>> Jamie: We're going to get started. Lili Cheng, distinguished engineer and director of Fuse
and our unofficial chief social officer is going to be introducing Londa Schiebinger then before
we get started I just wanted to mention that this work that Londa's going to be presenting is
applicable to all of the stuff that we do as well. As she's talking you should be thinking about
how it applies to your own work. She's very excited to hear the connections that you make as
you listen, so if you have time afterwards before running off for Memorial weekend, please
stick around and come up and share it with her so she can hear ways that you see that it is
applying to yourself. Thanks.
>> Lili Cheng: Thanks Jamie. For those of you who don't know in fuse we focus a lot on social
interaction and design, and I think one of the main things that you do in the design group is you
go talk to the people that you're designing for as the first stop. I think a lot of Londa's work
really embodies that and helps us think more of systems in different ways that we can even
expand the practice of talking to diverse people as you begin to design your projects, especially
women. For a little background, Londa is at Stanford and she is directing the Gendered
Innovation in Science, Medicine Engineering an Environment Project and I think you have over
60 different collaborators. It's a really amazing project and I'm sure we'll hear a little bit more
about that. She's one of the foremost international experts on gender in science and
technology and I think that is a topic that is really a focus of ours, so very excited for this. There
are a couple of different ways that I think we can think about women in our workforce. One is
just how do we get more women to participate in science and technology. What is really the
role of gender in culture and science? Also, for the projects and work that we do how can we
make sure that we really are designing for women as an audience. I think when you look at a
lot of our Microsoft products they sometimes reflect a less diverse audience than we think and I
think this becomes especially true as we start thinking of less platform oriented projects, but
may be things that you wear and devices. Social software obviously deals with community and
things like that. Londa has an amazing set of things that she's done. She has addressed the
United Nations on this topic. She has won numerous prizes and awards including the
Guggenheim Fellowship and the Alexander von Humboldt Research Prize, so very cool stuff.
Obviously, she has written a lot of books on this topic. Thank you so much for coming and we
definitely look forward to hearing the talk. And if people want to have smaller group
participation at the end and really brainstorm and share some of the projects that you're doing
and ways that we might do better with diversity, I think having some small group conversations
after the talk would really be great, so just stick around. Thank you and welcome. [applause].
>> Londa Schiebinger: Thank you very much. I've had a good time so far. I went out to dinner
with some of your wonderful people. I know a lot about universities and I know a lot about
funding agencies, but I don't know much about industry. When I go somewhere, I always like to
take something home with me. So please do stick around. I'd like to see if what I say has any
resonance with you and how it might apply to your work. Please do let me know. Today I'd like
to explore with you Gendered Innovations. The operative question here is how do we harness
the creative power of gender analysis to discover new things. I'm about research and
integrating gender analysis into research design. Your series here is about enhancing gender
and ethnic diversity. It's remarkably important to have diversity in teams, but is the crucial
thing to have diverse bodies in the room or is it that everybody should learn how to design for
diverse audiences? That's one thing that we can think about. First, a little bit of background.
Governments and universities have taken three strategic approaches to gender equality over
the past several decades. The first is fix the numbers of women. This means focusing on
increasing women's numbers participating in science and engineering. And in your series Marie
Klawe took this approach, how do you increase the number of women in computer science.
The second is fixing the institutions. This promotes gender equality in careers through
structuring change in research organization. Jane Margulis discussed change in these areas.
The third is fix the knowledge or Gendered Innovations. This stimulates excellence in science
and technology by integrating sex and gender analysis into research. Today, we focus on this
third strategic approach. It's the newest and hottest area. Some of you haven't heard about it
so much yet, and I think the most important for the future of science, engineering and
innovation. This is what Gendered Innovations is all about. A lot of this is most developed in
the biomedical field so I'm going to talk about that a little bit and then segue over two things
which are important to your core business. We know that doing research wrong can cost lives
and money. This is especially evident in medicine, but I think for industry as well. If you have a
product fails once you've launched it, then that is going to cost you money. For example, 10
drugs were withdrawn from the U.S. market between 1997 and 2000 and eight of these drugs
posed greater risk for women. Not only do these drugs cost billions of dollars, Forbes magazine
puts the price tag at $5 billion for a new drug because they factor in all of the failed drug
candidates. But when a drug fails it can cause death and human suffering, so we can't afford to
get it wrong. Doing research right saves lives and money. In the 1990s we had the large
governmental research project, the Women's Health Initiative, that you've probably heard of.
And one of the basic studies that they did was hormone therapy. Should women after
menopause replace their hormones? Should they take a lot of hormone therapy. We just have
done an analysis and the answer is no. You shouldn't. It causes breast cancer and other things.
There has just been an analysis of this study and for every dollar spent in the study $140 were
returned in diseases that didn't happen, so there's no healthcare cost for that. And we save
lots of lives. There were 76,000 fewer cases of cardiovascular disease, 126,000 fewer breast
cancers, 145,000 more quality adjusted years. While most of the results are positive, there was
the downside of more osteoporotic fractures. When you lose your estrogen your bones
become more fragile, but we can use lifestyle changes for that. Exercise, the right kind of diet
can in fact help you keep your bone. It's crucially important to get the research right and this is
the goal of the gendered innervations project. This project develops state-of-the-art methods
of sex and gender analysis and I'm going to talk about that a little bit today. We also provide 25
examples for case studies to illustrate real live examples of how adding this sex analysis or the
gender component to what you are you doing get you something better. Today I want to
introduce the methodological resources that are on this peer-reviewed website. Gendered
Innovations developed over the past five or six years through interactive interdisciplinary
workshops. We had workshops about 15 people all across Europe, the U.S. and now we're into
South Korea and these were fabulous for every technical field that we took. I'll be talking about
some of these. We had the experts from that field plus the gender experts and there were
conversations that had never happened before. It was very exciting for me personally. And
then we met in beautiful places like Berlin and Paris and it was really fun. Again, the question is
how can we harness the creative power of sex and gender analysis for discovery. I'd like to
present to you some of the outcomes. I'm first going to take an example from medicine and
then I'll come closer to your core business. These are some of the aha moments we had from
our workshops. My first example comes from cells and tissues and I want to look specifically at
stem cell therapy. Let's go back to a moment to why those 10 drugs were withdrawn from the
market. As I said, it costs a lot of money these days to develop a drug. There are many reasons
why drugs fail and fail more often for women, but one reason is that research is still most often
done in males whether it's humans, animals or cells and tissues. We have some good data on
this. This study was done in 2011 by some of our colleagues at Berkeley, and it shows the sex of
the animal used in the research in these particular areas. They took the run of the Journal for a
year and they found that mostly male animals were used in all of these fields for basic research.
The two fields that have more females are reproduction and immunology, but what I'm
interested in is this area here. I guess it turned out purple. This is where the sex of the animal
is not reported and this is data wasted to research. You might as well throw the money out the
window because you can't do any meta-analysis. It's just something people have failed to get
some of the basic data that you know. This study was done also for cells and tissues, done at
Mayo Clinic also in 2011, and here you see only gray area. The sex of the cells is almost never
reported and, again, this is research money wasted. Let's look at stem cells and I want to take
you to our website. You're welcome to use the website in any way you want. It's all public
information and free. We have our methods in buckets here. The methods are if you haven't
had this kind of information in your curriculum, if you weren't taught how to do the sex or
gender analysis, these provide questions that you can put to your materials. Then we have our
case studies in buckets here, science, health and medicine, engineering and where computer
science would be, and environment. First I'm going to go to stem cells and tell you a little bit
about this case study. Why might the sex of the cell be relevant? Research shows that there
are sex differences in the therapeutic capacity of stem cells. This slide shows very simply stem
cells taken from muscle tissue and shows that the female cells are more active than the male
cells. You can see this kind of different regenerative capacity. Yet very few researchers
consider the sex of the cell. We have a world-class stem cell researcher at Stanford who works
on Parkinson's disease. She was doing some research using all female stem cells. This is an
arbitrary decision on her part. I talked to her, quite unconscious. She just thought okay. I'll use
those. But it means that in the discovery phase she won't see anything unique to the male
stem cell, so opportunity missed, nor will she and her team detect any important differences in
the function between male and female cells. Again, an opportunity is missed. Not considering
the sex of the cell can lead to failed research. And international research team in Norway and
Australia were working with stem cells in mice and they very correctly used both sexes in the
mice. They had male mice and female mice, but again, they used all female stem cells, again,
and unconscious arbitrary decision, and the result was that their male mice died and they didn't
know why. And they thought, well, they put it on the shelf. They didn't know how to go on.
They thought oh postdoc probably made a mistake. When in doubt you blame your postdoc.
Eventually through a gendered innovation project in Norway the team realized that they should
consider also the sex of the stem cell, not only of the animal, but the sex of the stem cell. They
found that for their project male to male worked well and female stem cell to female animal.
But you have to consider all of the possible combinations before you ruled them out. You'll be
interested to know that the same is true for organ transplants. I hope you never do need a
heart, but if you do you better find out what the studies say about which sex combination
works well. And then for heart transplants it becomes a little bit complicated because we know
that it is sex matching. It is the thing that works best, but women donate more organs, so your
surgeon may not be able to choose the ideal organ for you. So we have a sex issue here and we
have a gender issue that interact in that particular example. Even for the stem cells, so the
method I was talking about here is analyzing sex. You see it's about the fourth one down. But
you have to consider also factors intersecting with sex and gender when you're doing this. If
you look around the room here all women aren't the same in the room and all men aren't the
same, so we have to consider differences within those groups If we're doing that. So with
humans there are all kinds of differences that are important and even in these stem cells we
have to look at the environment that they're in. We have to look at the hormonal environment,
immunological issues and you have to look at the type of the stem cell and the disease being
treated. You can't just analyze sex or just analyze gender. You have to look at these
intersecting issues and that's what our methods try to get at. Now I'm going to come to my
second example and I'm going to come closer to home for you. I'm going to look at an example
from computer science and specifically machine translation, so natural language processing.
Let me start with a little story. Couple of years ago I was in Spain and there were a lot of
newspaper articles about me. I don't read Spanish so when I got home I zoomed it through
Google translate and I was shocked that there is a masculine default in both Google Translate
and Systran which is the big European system. So Londa Schiebinger, he thought, he wrote,
occasionally it considered. How can such a cool company as Google make such a fundamental
error? My agency was wiped out. I worked very hard to get where I'm going and I just thought
this is ridiculous. So Google Translate faults to the masculine pronoun, because he set -- here
comes our gender analysis. Because he said is more commonly found on the web then she said.
And here's the interesting part. We know from Ngram, also a Google product, down where I
live you can't live without Google, but you can pretend I didn't say that. We know from Ngram
that the ratio of he said to she said peaked in 1968 at 4 to 1 and then it declined by the year
2000 to 2 to 1. So what else happened in this period? We had big social change. There's the
women's movement, the NFS starts increasing budget to increase the number of women in
engineering. We have all kinds of changes. Presidents are using inclusive language. Television
commentators are using inclusive language. Everybody is using inclusive language, but the
machine is defaulting to the masculine pronoun when it's doing translation. So with one
algorithm Google wiped out 40 years of very hard thought cultural and social change. And they
didn't mean to. So we invited Google tool workshop, one of these fun workshops that we have
and we also invited a natural language processor from Stanford and they listened for about 20
minutes and they got it and they said oh. We can fix that. I mean Google was appalled. They
had no idea. It's unconscious bias again. They just hadn't checked for that. So fixing is great,
but constantly retrofitting for women is not the best road forward. I had to ask myself how is it
that those Google engineers got out of Stanford where a lot of them are educated without
knowing the basics of gender stuff. We teach a lot of gender courses at Stanford. I teach a
course on gender innovations, a bit like what I'm telling you right now, but it's an elective and
kids have to opt for it. I think what we need to do and I want to work now with the new Dean
of engineering to fix the engineering curriculum, the computer science curriculum. And I was
discussing at lunch how we might incorporate some of the basics about gender issues right into
the core curriculum so that people will get out and won't make these silly mistakes. Why does
machine translate default to the masculine pronoun? The machine gets a score of 68 percent if
it chooses the male pronoun, so it gets a higher score than if it did not. And another issue is
that the translation program can't understand context, so even though Londa is the 1543rd
most popular woman's name in the United States, the machine can't connect that fact which it
can very well know to the rest of the content in the paragraph or the sentence. This masculine
default has some unintended long-term consequences. If we keep adding translations to the
web that just defaulted to he, you're going to be creating the future and the machine will even
more often choose he in any of its translations, so we are building a future where we're
reinforcing gender inequality when we don't intend to. I think that we are much smarter and
that. I don't have to pick on Google. There are other companies with unconscious gender bias
and it means that these unconscious assumptions, it means that we're not developing the very
best product that we could for the greatest number of people. Apple released its health app in
September and if you have an iPhone, which I do, it downloads it automatically for you. You
can't miss Apple's health app. And the health app tracks everything. It tracks your steps. It
tracks your blood alcohol content. It tracks your intake of sodium and magnesium and your
heartbeat, your blood pressure and everything. So what doesn't it track? The menstrual cycle.
And when you see the release of the product, the videos of the release, it's like we track
everything. So there are lots of reasons why women might want to track their menstrual
cycles. Again, this product can be fixed. There are lots of apps that in fact do track the
menstrual cycle and it can be added, but couldn't we consider gender from the outset? What
radically new things could we conceive? Now the Gendered Innovations project is launching a
series of tech roundtables. I'm going to be inviting maybe people like you from Microsoft
Research to come and join up. I'd like to understand how you think about gender in your
products, in your design, and learn more about what's going on there. We hope to be
launching some of these in the fall. We have a partner in the White House Office of Science
and Technology Policy. She's going to do them on the East Coast. I'm going to do them on the
West Coast and then we'll think if we get good stuff we'll have a gender summit that kind of
summarizes some of these things and excites people about it. I'm going to go for a final
example because I want to hear from you as well. This is assistive technology for the elderly
and believe me, the robots are coming. World population will age dramatically by 2050 and
you can see that it's a problem especially for Europe. It's a problem for the U.S. It's a problem
for Japan and some of the Asian countries as well. Large, elderly populations will place a
growing strain on human caregivers and our health and social systems. This case study explores
markets of assistive technologies for the elderly and looks at the value added at considering
both sex and gender when designing these technologies. What we want our designs that will
look at sex differences between men and women. We want people to look at the sex-related
physical needs that elderly people would require in these technologies. Women, for example,
tend to live longer, but they usually have more debilitating disease and men, for example, tend
to lose their hearing earlier. Sometimes these are occupational issues. They have worked with
loud construction objects or something like that. We not only need to look at the physical
needs of this population but also what the gender related social needs are and how that should
fit into the technology. We see that there are different gender patterns between men and
women. Men and women tend to have different partnering patterns when they're elderly.
Women very much live more often alone than men do. And men and women have different
experience in household management because maybe the men haven't taken care of homes so
much and they lose their partner they have different receptivities to technology. For instance,
elderly women tend to live alone more often. Our engineering checklist here and our terms
and our methods encourage researchers to analyze how sex and gender interact in individual
men and women so they can design the most effective and marketable assistive technologies.
Designers want their products to be useful and appealing to both men and women. Gender
issues become more important as the assistive technologies become more personalized. I
should have better pictures of the robots. This is the one that the European Union was using,
the companionable and we had a workshop with roboticists and they all claim that they have
no gender in their robots, but then the guys who developed this told me that they
spontaneously started calling the robot Hector. Not only did they conceive it as a male, but as a
gladiator. There are a lot of robots that are developing and one is a robotic nurse named Cody
that can bathe people. Bathing is a very intimate relationship between human and machine
and so designers will really want to make sure that both men and women of different social
backgrounds are comfortable with these machines as we move forward. Another assistive
robot is Herb, the home exploring robot butler who can fetch household items for you. Herb
can fetch you a cup of tea. He can remind you to take your medicine or even clean up the
kitchen. If there's a robot that can clean up the kitchen, I'm ordering it right away. As these
robots enter our lives we humans will gender them. You see, I've changed the direction now.
We gender things. Studies of machine voices, that is to say synthetic voices, machine
generated voices, show that human listeners tend to gender machine voices. That is to say we
interpret them as the voice of a man or a woman even when the designer intended that voice
to be neutral. This is the work of the late Cliff Nast, really interesting work. Apple's Siri, the
original iPhone voice is interesting in this regard. As a party game I used to ask Siri why she's a
woman when she was only a woman. One of her responses was I was not assigned a gender,
implying that it's not Apple's fault that you the listener ascribe gender to her. That's your
problem. And as soon as we humans interpret a voice as masculine or feminine, we tend to
apply all of our cultural stereotypes to that machine. We have this really want to understand
the world in terms of gender and that's one of the things that we do. Here we need to come
back again to the method that I talked about earlier, and that is analyzing factors intersecting
with sex and gender. So when we are talking about designing assistive technology for the
elderly and looking at men or women, we have to say which men or which women. We have to
think about differences in ethnicity, differences in physical ability, that sort of thing, so that we
can in effect do a better job for a larger group of people. For example, Siri can't understand my
friend's French accent at all and my Jamaican friend can't get any kind of voice that would
match his. So there are many factors that we need to consider. Considering sex and gender
when designing new assistive technologies will be one important factor to ensure that products
are successful with all users. On our website we now have about 25 examples. You can click
into any of them and see them. I encourage you to go look at them. We always have an
abstract. You can learn a lot in a little bit of time. We have our full case study, which is not very
long and I also made a popular version for journalists because so many people were calling and
that's called In a Nutshell. You can read that as well. Designing sex and gender analysis inter-
research and innovation is one crucial component contributing to world class science and
engineering. I want to end with just a little bit of attention to policy. Policy is one driver of
innovation and I know it's not so important to you in companies because you don't respond to
those funding things, but it's important for you to know the pressure that is on your colleagues
in other sectors. We incorporated all of the policies that we know for granting agencies.
Granting agencies can ask applicants to explain how sex and gender is relevant to their
proposed research as a requirement for funding. This is where it gets really interesting. In
December 2013 the European commission which is the NSF, NIH and NEH for Europe developed
its next funding framework for the next six or seven years. They identified 137 areas of science
and technology where they will be looking for sex and gender in the proposal. If you don't say
how it works, you're not likely to get a high score for your funding. These fields include
computer hardware and architecture, nanotechnology, oceanography, geosciences, organic
chemistry, aeronautics, space medicine, biodiversity and it goes on and on. They haven't
included things like theoretical physics where we don't know how gender works in those
projects, so it's really just areas where they have evidence already that this would be a value
added proposition. Then since 2010 the Canadian Institutes of Health Research have ask for sex
and gender in their proposals and they're currently giving teeth to this policy. So far they have
just asked that you included, but now they're looking to make it 10 percent of the grade for any
proposal. The Bill and Melinda Gates Foundation since 2008 has asked for gender as a
requirement for funding development work. And what's so great about their program is that
the program officers will help the researcher if the researcher doesn't have background in this
area to understand how they should do the work. Now, the U.S. National Institute of Health,
this is what's coming. This is the big one and there's a lot of activity and excitement about this.
In fall 2014 the U.S. NIH announced that it would be requiring sex analysis in preclinical
research so that's in cells, tissues and animals. As you probably know, since 1993 we've had a
law, a federal law that women and underrepresented minorities must be included in clinical
trials. That has always been a problem because you're going to do the research on human
women, but you haven't done it on mice and rats earlier. You haven't included the females
there, so what kind of scientific research is that? It doesn't seem very safe. That's policy at the
beginning of the research process. The granting agencies encourage researchers to include sex
and gender analysis. And then also at the end of the research and I would be interested to
know if any of your journals are beginning to ask for this. But journals are now requiring that
sex and gender be included in the research or it's not excellent research, and so the reviewers
would have a problem with it. I'm very interested to find out what industry can do. Products
and systems that incorporate the smartest aspects of gender and open new markets, products
that meet the needs, the complex needs of diverse user groups enhance global competitiveness
and sustainability. So I'll be watching to see what Microsoft does. You can be a leader and I'll
watch for the fireworks to the north. Thank you [applause].
Download