Tulsee Doshi Is A Silent Hero: She Ensures Google's AI Doesn't Have Any Bias

Today, when we pick up our smartphone and use it, the products and services created by developers like Google, and the user experience feels at home -- we don¡¯t feel that it isn¡¯t made for us or it is designed for an entirely different audience that isn¡¯t you.

Getty Images

An AI that is unbiased

Let me give you a small example. We have Google Assistant on our phones or Amazon Alexa who is able to understand our tone, accent and even languages other than English to make our user experience more user friendly.

However, how would it feel if it would only recognise US accents or UK accents? That¡¯s AI bias. and this bias is not just in the way we speak or are heard, but the applications are far more widespread than you could imagine. That¡¯s what Google and many other companies thoroughly investing in AI are trying to avoid and fix.

This uniformity, this inclusiveness in Google¡¯s products is maintained and monitored thanks to Tulsee Doshi, a Google product lead in ML (Machine Learning) Fairness and Responsible AI (Artificial Intelligence).

Tulsee Doshi: Twitter

In a conversion with Google Keyword Contributor Andrea Lewis ?kerman, she revealed how she would explain her job to someone who isn¡¯t so tech-savvy.

She said, ¡°My work includes making sure everyone has positive experiences with our products, and that people don¡¯t feel excluded or stereotyped, especially based on their identities. The products should work for you as an individual, and provide the best experience possible.¡±

Understanding bias in AI

An ML/AI platform is only as good as the data it is fed. Just like a child who¡¯s choices and understanding expands with the information given to him/her, ML/AI have bias depending on the data that is used to train it.

Let me give you a small example. For men, if you were asked to think of shoes, the first two, or even first five shoe images that would come to your mind would be all-male shoes types. Only after a few tries, you¡¯d think of female shoes like high heels, or stilettos etc and that too with limited numbers, considering they¡¯ve never worn or even thought about one. Had the ML be fed with just your data, the knowledge ML would have, would be fairly limited -- this would have caused bias.

Just like this, Tulsee too experienced a disconnect between products and the people who were going to be using these products, ¡°No matter which interesting courses I took, I often felt a disconnect between what I was studying and the people the work would help.¡±

She added, ¡°I kept coming back to wanting to focus on people, and after taking classes like child psychology and philosophy of AI, I decided I wanted to take on a role where I could combine my skill sets with a people-centered approach. I think everyone has an experience of services and technology not working for them, and solving for that is a passion behind much of what I do.¡±

Getty Images

Why is addressing bias necessary?

Today, AI and ML are in almost all walks of our life -- from shopping online, to the kind of songs you like to listen to the navigation apps on our phones or apps that help us translate languages. If these apps have a bias, they would not provide the kind of experience a user demands from them. It would make them feel that the app isn¡¯t made keeping them in mind.

Challenges in making products inclusive for all

Sure, when someone is trying to create apps that would be inclusive of everyone, things go up many notches. When asked about what challenges she faced in this work and its importance, she said, ¡°The challenge question is interesting, and a hard one. First of all, there is the theoretical and sociological question on the notion of fairness¡ªhow does one define what is fair? Addressing fairness concerns requires multiple perspectives, and product development approaches ranging from technical to design.¡±

She added, ¡°Because of this, even for use cases where we have a lot of experience, there are still many challenges for product teams to understand the different approaches for measuring and tackling fairness concerns. This is one of the reasons why I believe tooling and resources are so critical, and why we¡¯re investing in them for both internal and external purposes.¡±

Reuters

Company culture

She stated that apart from defining fairness, it is also important to understand how companies define their values to motivate their employees, ¡°We are starting to see a growing, industry-wide shift in terms of what success looks like. If organizations and product creators get rewarded for thinking about a broader set of people when developing products, the more companies start fostering a diverse workforce, consult external experts and think about whose voices are being represented at the table.¡±

She concluded stating, ¡°We need to remember we¡¯re talking about real people's experiences, and while working on these issues can sometimes be emotionally difficult, it¡¯s so important to get right."