Luke Kenworthy作者



The issue of diversity in STEM fields is something which has been spoken about for some time, albeit until recently on a smaller scale. Is the issue of diversity, starting to be recognised at a greater scale in STEM fields? Has there been a shift in the increase in vocalising the discussion and creation of movements and organisations with the sole focus of addressing the concerns of those directly affected. Although increasingly spoken about, what are we doing to address diversity and how can we practically address the issue currently present? We asked some of our friends in STEM for their practical solutions.

Georgia Gkioxari, Research Scientist, Facebook AI Research

"In my opinion, to promote diversity in the STEM fields, we need to change things bottom-up. We need to uproot the gender bias that society inflicts on women at an early age, we need to combat economic inequality, we need to guarantee equal opportunities and high quality of education regardless of zip code. This, of course, calls for a huge sociopolitical change! Top down efforts like pressure for female presence in executive roles, workshops and gatherings which promote underrepresented groups, and other such activities, are helpful and encouraging but I fear not powerful enough to establish the change we are looking for in the long run."

As a Research Scientist at Facebook, Georgia mainly works in computer vision, which is a field of AI that studies the ability to see, recognize and understand. Obviously this is an incredibly interesting field to work, however, we wanted to know if there are any other area of AI that are of interest. "I love this field, it is challenging and has potential for great impact in the world. So long story short, no! I wouldn't pick any other area to work on!"

Sarah Laszlo, Senior Neuroscientist, X the moonshot factory

Recently, as I've been growing a team and thinking about how to grow it more, I've been thinking about this a lot:  how do I get people interested in what I'm doing so that they will sign up to work on my team? The conclusion that I've come to surprised me a little bit, because I think the answer to this is the same for me now, a technical lead in industry, as it was when I was a professor trying to recruit students to my lab.

When I was an undergraduate at MIT, it was mandatory for Brain and Cognitive Science majors to take a course in either statistics or probability.  I signed up for the probability theory course — but I did not do well on the first exam, and the TA asked me to attend his office hours to discuss.  He asked me why I was taking the course, as it was offered through the computer science department and did not have many brain science students in it.  I, in a hapless manner, told him that I was taking it because a course in statistics or probability was required for my major.  He asked me which of those two things I actually wanted to learn, and I told him that I didn't know the difference between them, I just wanted to fulfil the requirement.

This was an embarrassing thing to say.  There are a lot of professors and graduate students who would have reacted to it badly, by writing me off as ignorant or as a dilettante.  This TA did not.  To his great credit, he sort of ingested that information, swallowed it that I didn't know what I was talking about, *and decided to help me get better anyway*.  About ten years later, I was a university professor, teaching a graduate nonparametric statistics seminar that no one else wanted to offer. I never would have approached understanding the seminar material without the fundamentals I gained in MIT probability theory all those years ago.

It would have been so easy for that TA to write me off.  It would have been so easy for him to have discouraged me and sent me back to my dorm upset, for me to drop out and take the (perceived as much easier) statistics course at Harvard instead. But he did not do that. 10 years later I was teaching a related course. 15 years later I still use what I learned from him. If tomorrow he called and asked me if I would work for his start up, I would consider it, and I would without hesitation recommend to him all the other women I know in the field, knowing that he would treat them the same way he treated me.

The lesson to take from it is:  if you want to cultivate diverse talent -- where diversity includes diversity of cognitive, racial, gender, and experiential factors -- you can't do it magically just in time when you have an open role on your team.  You do it all the time, always, with people of every level, cultivating talent when you see it so that 10 years later, or 15, there are all types of people in the world that have the skills you want, people that *didn't* get discouraged and quit and feel cut out of elite technical spaces.

Having diverse hires means having a diverse pipeline, and having a diverse pipeline means taking a long-term view of recruitment, spending months or years of work to find people early in their careers and cultivate them so that they are there, the network is there, when the time to hire comes.  If you see a talented person, cultivate them.  Do it even if they are very junior or if they don't know what career they want yet or if they have come from a background that has not given them the tools they need to fully make use of their talent.  Do it to help them, out of respect for all the people that have helped you along your way, but also do it to help yourself and your future teams, when you have an open role and you want the best, deepest, most diverse pool of people to fill it from.  Help young technologists, and they will be there for you when you need them because you were there for them when they needed you.

It’s important to have people from all backgrounds (women, non-caucasian, LGBT) in higher positions in the university or workplace. When you enter a workplace/university and all the (very few) women are in lower positions (students or interns), it sends the message that you will not be able to reach an higher position

Jekaterina Novikova, Director of Machine Learning, Winterlight Labs

I find it very promising that more and more conferences (both academic and non-academic) introduce the Code of Conduct that promotes diversity and inclusion, support the diversity-related events, start thinking about child care support and try to find other ways to express their commitment to ensuring fairness and equality. Doing similar things on all possible levels in STEM fields, not just during conferences, could help a lot. Jekaterina is currently working in the field of NLP,(natural language processing) and gave it a glowing endorsement for young persons looking to enter data science. "If I was just starting it all again I would definitely choose this field again! I am fascinated by 1) the complexity of problems we have when we teach machines to understand natural language, and 2) amazing progress that is being made by the NLP community in the recent years".

Alexia Jolicoieur-Martineau, AI Researcher, MILA

I’m unable to find the source link anymore, but one nice trick is to rename “CS 101” class to something like “creative solving through computational design”.Another way, is to add programming and applied AI courses in art curriculums. There are many ways to use the tools for creativity and art.

Then, for retention, it’s important to have people from all backgrounds (women, non-caucasian, LGBT) in higher positions in the university or workplace. When you enter a workplace/university and all the (very few) women are in lower positions (students or interns), it sends the message that you will not be able to reach an higher position (implying a glass ceiling) even if it might not be true; furthermore, it means that you will not find a mentor like you. I went from statistics to AI and I also encourage others to take that path. AI is effectively statistics and statistician jobs often involve a lot of paper-writing and preparation for a PhD (and often gives you the same training although sadly without a degree).

Rana el Kaliouby, CEO, Affectiva

We need to build up the entire ecosystem of women and diverse leaders in tech, starting with having more role models for women to look up to. As a young woman, I was never able to relate to the Mark Zuckerbergs of the world - but when I met my mentor, Dr. Rosalind Picard, I saw proof that women could be leaders and pioneers in technology. Young women need to see examples of people like them, pursuing their passions, so they can be empowered to do the same. And it’s not enough just to have more women in tech - we need more female investors, founders and C-level executives to serve as role models and advocates for other women in all stages of their careers.

Today, at Affectiva we’re focused on applying Emotion AI to industries like automotive and media analytics. But another area I’m really interested in is healthcare. AI has the potential to transform the way we diagnose and treat our health from a holistic perspective, combining data on our physical health with insight into mental health and wellbeing. Today, healthcare providers can quantify all aspects of our physical health, but there’s a gap when it comes to understanding mental health and wellbeing.  For example, doctors still rely on a self-reported indication of mental health – asking how we feel on a scale from 1-10. Emotion AI could provide deeper, more quantifiable insight into our mental health and wellbeing. And if that information from AI is aggregated with data on our physical health, the result would be powerful.

Julia Rabin, Project Lead, Diversity VC

What hiring policies are in place in STEM organisations? There is plenty of diverse talent out there but are companies writing job descriptions in a way that may only attract one demographic? Then are they reaching out to diverse communities or just their own networks? These aren’t big changes for organisations to make but they could have a big impact. Certainly something to think about from Julia, especially thought provoking to consider the underlying bias which may be unknowingly or unwillingly present in the hiring process!

Samantha Edds, Senior Data Science,

Education systems promoting STEM to all different types of students with targeted outreach has become a lot more popular, and is a great way to ensure more diversity, but until those students reach working age, there is still a lot that can be done. For companies, this can be at the recruiting level, from purposely targeting diverse candidates to holding events (recruiting, meet-ups, or otherwise) with more diverse groups. There are a lot of non-profits and groups that target diverse people to help them learn more about STEM and begin to break into the field. I work with Frauenloop in Berlin as a mentor and see that everyone from small startups to Microsoft are beginning to target these groups, and if you are on the other side, and looking to break into STEM, places like Frauenloop will really help vault you into the field.

Chanuki Seresinhe, Visiting Researcher, The Alan Turing Institute and Lead Data Scientist, Popsa

How can we promote and increase the diversity issues present in STEM? By empowering those from diverse backgrounds who enjoy STEM fields to pursue careers in this field in the first place, with an emphasis on finding ways to encourage and support those people to stay in those careers even when they hit the inevitable roadblocks along the way.

I think exposure is a big part of promoting diversity; in particular, by making children and youths aware of different areas of STEM

Julia Kroll, Data & ML Engineer, Amazon

To further promote diversity in STEM fields, anyone who acts as a gatekeeper to the next stage of developing a STEM career – college admissions officers, professors, recruiters, hiring managers, entrepreneurs, and venture capitalists – should strive to encourage a diverse pool of people to pass through their gate. Is this something which is too engrained in process culture to change, however? Within the field of AI, What are they doing at Amazon to increase diversity? Amazon has many programs to increase diversity and inclusion, such as Amazon Future Engineer for underserved and underrepresented K-12 students, and Amazon’s Student Programs that offer internships to a diverse group of students. Visit our diversity website for the latest information about our workforce: Diversity and Inclusion at Amazon.

Bianca Curutan, Software Engineer, Postmates

I think exposure is a big part of promoting diversity; in particular, by making children and youths aware of different areas of STEM that are available to explore. STEM covers such a broad range of fields and expertise that it's difficult to say where the limit (if any) is, so there is a high likelihood that their interests intersect with STEM opportunities in some way or another.

Dej Mejia, Senior UX Designer, Adobe

Start early by celebrating the contributions of women and under-represented minority inventors, educators, solutionists and technologists so that students can see themselves as capable future contributors. Participating in events that champion diversity, like speaking to minority groups at universities to share career stories and successes, can have a big impact. Helping future diverse STEMsters understand possible career paths, and how to bridge the gap between STEM and their key interests, can foster a sense of promise in pursuing careers in STEM.

As a UX designer, I have had the opportunity to specialize in different technologies throughout my career. AI has been the most intriguing field to solve for, but I previously designed user experiences in the identity monitoring space. Ideating with product and development teams to discover new features that can help users manage and protect their identities can be a satisfying experience

Afsaneh Fazly, Director of Research, Samsung Toronto AI Lab

When asked what Afsaneh thought was the answer to diversity problems in STEM, it was a short, simple but incredibly powerful one. Simply "By recognizing what women bring to the table". This may seem as though it is too straight-forward, however, could it be the case that sometimes we overthink the potential solutions? There can be many efforts to .  create movements for the future and incentives for current female students, but do we fully appreciate those currently working to advance the industry? Afsaneh currently works in the field of computational linguistics which she suggested gives her a small window into human cognition! Could this field be for you?

Next week, we will be asking contributors  to answer 'Recent reports have suggested that Science graduates are increasingly looking for careers away from STEM fields, how do you think we can increase the number of graduates interested in choosing stem jobs?'. I know, our questions aren't getting any easier! Interested to contribute? Email me on

Interested in being part our mentorship scheme to support women in AI? Either as a mentor or a mentee? See more here.

Not yet read the first Diversity Diaries blog? Hear what our contributors gave as their biggest piece of advice you would give to a young woman starting her career in AI/Data Science.




Julia 是MIT设计的一个面向科学计算的高性能动态高级程序设计语言,项目大约于2009年中开始,2018年8月JuliaCon2018 发布会上发布Julia 1.0。据介绍,Julia 目前下载量已经达到了 200 万次,且 Julia 社区开发了超过 1900 多个扩展包。这些扩展包包含各种各样的数学库、数学运算工具和用于通用计算的库。除此之外,Julia 语言还可以轻松使用 Python、R、C/C++ 和 Java 中的库,这极大地扩展了 Julia 语言的使用范围。