Happy International Day of Women and Girls in Science! Today is a great day to not only celebrate women and girls in STEM, but also to consider the ways in which we can work toward an even playing field for #WomenInScience! To start this conversation, here is my attempt to breakdown stereotypes in Artificial Intelligence.
I believe AI can be a powerful tool to elevate human lives. But not the way it stands today. Today AI, which learns by examples, is built predominantly by one group of people — male, white, aged 25-45. It inherently learns their preferences and biases, and thus cannot effectively serve the diverse needs of the general population. In many cases, it is detrimental to already-marginalized groups; think racist chatbots, facial recognition systems that cannot detect dark-skinned people, or gender-biased credit cards.
My audacious ambition is to address this bias problem in AI by changing the face of the AI workforce, one young person at a time. The goal I am aiming for is AI that can powerfully and positively transform human-machine partnerships because it has been built by a workforce that represents a diversity of perspectives, concerns, experiences and skills — our collective intelligence — thus putting human-human partnerships at the very core of AI’s success.
I have mentored young people in machine learning and AI for ten years now. Three years ago, I did something bolder: with my boss, Rana el Kaliouby’s encouragement, I developed my company, Affectiva’s EMPath (Emotion Machine Pathway) internship program with a central goal of dispelling the “techie” stereotypes deeply ingrained in AI recruitment, and building up the next generation of AI leaders representing the rich diversity of our world.
The field of AI is often intimidating to newcomers. Most are under the impression that participants have to have years of experience in tech to even join. I contend that this broad generalization is false and part of the (harmful) myth that keeps many out of the field. So to prove this point, the only technical requirement for being accepted to the EMPath intern program is one introductory high-school programming class. This simple requirement as opposed to a laundry list of technical acronyms in the job description brings us interns who would not typically be found in other tech intern programs: young high school students, women, people of color, immigrants, students who didn’t go to ‘pedigreed’ schools.
In the program, the interns are provided with one-on-one daily mentoring from experienced AI researchers and engineers, who teach them technical skills through individualized projects that are both interesting for the intern and impactful for the company. Everyone from the company’s CEO to the newest research and engineering employee volunteers their time with the interns.
Our interns work on various AI projects ranging from building models for detecting human emotions, to using AI to understand social phenomena like political polarization, to testing methods for mitigating algorithmic bias. These projects are regularly showcased in peer-reviewed conferences, in front of industry leaders at my company’s yearly summit, and digitally through the company’s website. And in all venues, the response is the same: appreciation and respect mixed with incredulity; people are still not used to seeing groups that are predominantly women, predominantly non-white and extremely diverse achieving success in AI. But to that I say, “World, get used to it!”
Breaking down the stereotypes of technologists is at the forefront of my career ambitions. Revisioning who can work in tech is not only essential to leveling the playing field for women, for people of color, for immigrants, for the differently-abled, but it is crucial for AI solutions to work robustly for all of us and deliver on its promise to transform the world for the better!
What are some other ways we can break down stereotypes of scientists or create a more level playing field for women in science? What can we do to on a small, or grand scale to increase representation of women in tech, science and AI?
This Article was originally published on LinkedIn on February 11, 2020.