Solving the Empathy Gap in Technology
by Arathi Mani
I sat in a wind tunnel for the first time while attending a summer camp as a middle schooler. Twelve-year-old me thought it was the coolest thing ever, sitting there as the wind whooshed across my face, feeling like I was nearly levitating in the same air used to test aircraft levitation. That experience single-handedly convinced me, at that time, that I was going to be an aerospace engineer when I grew up.
Well, you know what they say about the best laid plans, especially those laid out by 12-year-olds…Five years later I chose to pursue computer science. My plan on becoming an aerospace engineer was derailed by my great enjoyment of taking my first computer science class in high school. Then, trying to decide between the two areas, I figured I wouldn’t be shut out of the world of aerospace if I pursued my love of computers, but the reverse might not be true. It was a solid (and rather astute) bet for a 17-year-old to make, even though I still have not yet sat in another wind tunnel (the dream lives on!). I have been extremely fortunate to be able to apply my computer science knowledge to a vast array of applications — from mental health, to natural language, and now, in my current role (Senior Software Engineer at the Chan Zuckerberg Initiative) to single-cell biology.
The Job Gap That Keeps Growing
My senior year in high school marked an interesting inflection point for computer science in the U.S. — it was 2010 and the start of a sharp growth in the number of students who graduated with a degree in Computer Science. In parallel, an external narrative seemed to be mirroring this growth: there are more jobs available in CS than there are qualified workers.
This line was inculcated in me not only by my school counselors, but also generally by the news. For me, it was probably some combination of this external narrative as well as that one computer science class I had the privilege of taking that spurred my interest and, in a happy collision, deviated my path away from aerospace engineering. Interestingly, a decade later, the external narrative to convince more folks to pursue computer science jobs has continued to remain roughly the same. The gap has been projected to be quite dramatic in the upcoming decade, with the U.S. Bureau of Labor Statistics projecting a 21% increase in demand for Software Engineers between 2018 and 2028 (given 1,365,500 jobs in 2018) and only 284,100 people available with the skills necessary to fulfill these jobs.
These statistics as well as the allure of cushy, well-paying jobs at companies like Google, Amazon, and Facebook continue to fuel an exponential growth in the conferral of computer science degrees. For example, of the undergraduate degrees conferred at Stanford during the 2018–2019 academic year, a full 17% were degrees in computer science. The pattern continues even beyond Silicon Valley with the number of conferred degrees in computer science-related fields growing steadily upwards from around 43,000 in 2010 to over 71,000 in 2017.
While this growth is to be celebrated, I see two big problems. One is that the gap between jobs and jobseekers still isn’t closing and instead growing larger as time passes. This is unsurprising as more of our lives become entrenched in technology, generating even more technology-based jobs.
The second is a growing subplot around diversity in more than the demographic-sense, threatening the technological advances that we so eagerly want in our world:
Computer scientists desperately need people with deep domain knowledge about all the other fields of study.
And I really do mean all. We need people who have studied African American Studies and Computer Science, we need folks who have pursued Social Work and Computer Science. We need people who are experts in Agriculture, in Politics, in Law, in Microbiology, in Philosophy and Computer Science. We are badly lacking diversity of knowledge in technology, I believe, partially perpetuated by the external narrative pushing CS studies and not multidisciplinary studies.
Diversity of knowledge, stemming from diversity of study, background and perspective, is not only beneficial but paramount to the success of our communities and our world.
Lack of Representation Leads to Inequity
This past April, I attended the International Conference for Learning Representation, which was the first major machine learning conference ever to be held in Africa. It was ultimately held virtually rather than in Ethiopia due to the coronavirus pandemic. One of the most impactful and powerful keynotes I heard during the conference was delivered by Professor Ruha Benjamin, who described the way in which many current technological advances — data science in particular — hide modern-day social injustices by failing to address historical biases. Oftentimes, these technological advances misguidedly celebrate automation as progress while creating a society that unconsciously reinforces past prejudices by ignoring lack of diversity in the data itself, in our methods, and more broadly, in ideation.
One life-altering example is a recent effort to predict recidivism, the likelihood that a prisoner might reoffend. Without accounting for the fact that historical human-generated data reflects many of the human rights atrocities of our generations’ pasts, we end up amplifying the biases into the future in ways that clearly discriminate against underrepresented groups, especially Black people. Another example is lack of representation in state-of-the-art machine learning research; 30% of all living languages today are languages spoken in Africa, but the majority of machine translation research has traditionally focused on translations of Indo-European languages (i.e. French, German, Spanish, etc.). Eventually, successful research will find its way into products, and so lack of representation early in the pipeline bodes of unequal opportunities later down the line. In fact, — not one of the languages available for Siri real-time translation is a language spoken in Africa.
Dr. Benjamin said during her talk “computational depth without historical and sociological depth is superficial.”
We can all do better to understand the ways in which we perpetuate unconscious bias but we also need to take action to address this superficiality.
One way we can do this is by championing the recruitment of a workforce that is not only demographically diverse, but also multidisciplinary.
A Multidisciplinary Approach
To ensure we are creating a world that deeply values the health of our communities, we need a set of tenets that emphasize learning from our past and curiosity into the nuances of the problems we are trying to solve. Both of these principles rely on diversity of knowledge, which point to a strong need for diversity in cultural background, in race, in domain knowledge, in gender, and in numerous other ways. The conjecture becomes nearly a tautology: if you have diversity in people, then you will likely also have diversity in thought.
It falls out naturally then that pursuing a multidisciplinary background is invaluable, and the cultivation of cross disciplinary teams across all work sectors is vital.
But how do we encourage this? One possible solution is to shift the way we teach computer science in grade school; instead of presenting CS as a class to take in high school to vet the major for collegiate study, we can start earlier and begin teaching basic concepts in elementary and middle school through programs like Scratch and Alice.
Another important shift that I think needs further emphasis is the inclusion of CS in general education courses. Instead of presenting CS as one of the few majors with extremely high job prospects, we should encourage students to pursue any major, knowing that a CS class or two will be included. By doing this, we ensure that students are better prepared to address the technological shifts in any industry. Who knows, perhaps if this had been the case back in 2010, I would be working with wind tunnels today! The good news is that I think there are signs that both shifts are already beginning to occur; in fact, regarding the second shift, I believe think we are at the cusp of a new wave of multidisciplinary tertiary studies, called informatics.
Defining Informatics
Looking back in history on the rise and fall of various studies, it was only in 1958 that the first definition of aerospace engineering as a discipline appeared; the term ‘engineering’ emphasized the application of mathematics in a particular field, in this case aerospace.
In many ways, I see the future of computer science study mimicking the way mathematics is taught today; as a general purpose tool that serves nearly all disciplines in an applied sense.
Today, computer science has a strong theoretical component to it because of its siloed study. While theoretical study should always be strongly supported, applied study is also very valuable and helps unlock additional knowledge in the field in which it is applied.
‘Applied CS’ generally takes on a different name in the list of pursuable studies: ‘Informatics.’ We are already seeing a trend towards this with the emergence of various Informatics fields — Bioinformatics, Business Informatics, and even Music Informatics. It makes sense: given that the term ‘Informatics’ has been loosely defined as “the science of processing data for storage and retrieval,” and given that data is now being produced everywhere, we need more multidisciplinary tools, that are also domain-driven, to process it.
The Final Piece of the Puzzle: Cross Disciplinary Teams
Beyond educational reform, companies need to start forming cross disciplinary teams with a strong focus on hiring a diverse workforce. Here as well, we are already beginning to see positive change in this direction. For example, the Stanford Institute for Human-Centered Artificial Intelligence (HAI) aims to build a multidisciplinary community involving AI researchers and domain experts to develop solutions that are human-centered at its core. Other Diversity, Equity, and Inclusion (DEI) efforts throughout the world, like SACNAS, make huge strides towards this goal by focusing broadly on diversity of thought. Make no mistake, we aren’t done yet and the road ahead is still long; there are going to be times where we as a society feel like we aren’t making progress. even though we’re investing time into having tough conversations or hammering down what seems like an indestructible wall. I can personally attest to having felt these feelings in my own role, working at the intersection of single-cell biology and technology.
While my own background is in tech, tech, and some more tech, my work requires me to collaborate with scientists to ensure we build technology solutions that address the core concerns that scientists are facing and not make up a solution based on what I see at surface-level. I work on helping drive the creation of a comprehensive cell atlas; the vision is an amalgamation of many single-cell datasets generated by many different labs across the world, spanning donors of all ethnicities, genders, ages, and other demographics, integrated in such a way that we have a more holistic understanding of the human body that can accelerate cures for many diseases. Without the incredible support of the team of computational biologists that I am very lucky to work with, I would not be able to understand the nuances of the problems the tech needs to address. They help me answer questions like how does data need to be integrated such that biological variations are retained? What sort of metadata do I need to store such that meaningful downstream analyses can be performed on the corpus?
There’s something really beautiful about being able to have heated conversations about product decisions but come out the other side with not only a better understanding of the other person, but with a little better understanding about the world we live in.
The very act of trying to adopt a mindset that genuinely attempts to understand others’ perspectives is a step in the right direction and will bring us closer to creating a more equitable society. Coming to compromises and building technology that is for biologists and accelerates their work is truly such a rewarding experience and over time, the conversations that once were quite hard, become easier.
All of these efforts are absolutely worth it. We risk so much by choosing to separate computer science and “everything else.”
We risk building a split society of the haves and have-nots and not solving the real crises of today. The Governor of California, Gavin Newsom, succinctly summarized this growing dichotomy in his keynote address at the opening of HAI: “There is an empathy gap… in technology.” I am very hopeful that this empathy gap can be solved, but we need to invest, promote, and encourage diversity in technology, more fervently than ever. We can take one small step towards this goal by at least addressing in our educational pursuits that the choice isn’t Computer Science or. It’s Computer Science and.
About the Author
Arathi Mani is a Senior Software Engineer at the Chan Zuckerberg Initiative working on the Single Cell technology team within CZI Science. She is currently leading an effort to build a platform that enables the publication, discovery and exploration of interoperable single-cell datasets with the goal of eventually creating a map of all the cells in the human body. Prior to CZI, she worked as a Software Engineer at Google and was a Lecturer at Berkeley. She has a Master’s degree in Computer Science from Stanford University and a Bachelor’s in Computer Science Engineering from The Ohio State University. In her free time, she loves to hike, ski, and bake far too many cookies.
Twitter: @ArathiMani • Instagram: @arathimani • LinkedIn: Arathi Mani • Medium: Arathi Mani
