Using Data for Good: How Data Scientists Are Using Information to Create Positive Societal Change

In 2020, data is being collected, synthesized and collated to inform business and policy decisions around the world. And with an ever-increasing wealth divide in the U.S., many data scientists are wondering how they can create sustainable systems to use data insights for good.

In a September 2020 panel discussion hosted by Trilogy Education Services, a 2U, Inc. brand, three data scientists discuss how they’ve used their professional backgrounds to help organizations systematically and intentionally utilize data to benefit communities.

Below is a transcript of part of the question and answer session with Anjali Samani, Data Science Director of Data Science Apps and Data Intelligence for Salesforce; Anusua Trivedi, Senior Data Scientist Lead at Microsoft’s AI for Good Research Lab; and Collin Cunningham, Data Scientist of Amazon Web Services (AWS) and Co-Founder of SpringForward and ATLFamilyMeal; hosted by Alison Abbington, Industry Engagement Manager, on behalf of 2U, Inc.

What does “data for good” mean to you?

Anusua Trivedi: This is a very new team we have formed [the AI for Good Research Lab at Microsoft], just two years back, and we’ve mostly been working with enterprise customers. The kind of asks we saw starting to evolve were, “How can we apply this kind of knowledge to more of a nonprofit or academic sector?” One where you’re not focused on making a profit, but trying to utilize data to make policies which can advance the betterment of a society. So all of our work is actually driven by policy changes.

Anjali Samani: Historically, I haven’t been a fan of this term because it’s so broad. Does it mean for profit or not for profit? Does it include government or non-government agencies? What does “good” mean in this scenario? I prefer to get more specific about it.

First, think about what you’re passionate about. If you feel passionately about climate change, there are for profit and nonprofits addressing these issues. I prefer to identify the cause that speaks to me, then get really specific about the kind of work I want to do within that space.

Collin Cunningham: Those were thorough answers, but the one thing I’ll add is about using AI for good. It’s about the metric you want to use. Take corporations for example, their first mandate is usually “I want to make as much money as possible.” With AI for good, that metric for success is not value for a corporation or economic value, it’s specific societal benefit. The usefulness and implementation of the model is seen by evaluating how impactful it was: Did people actually need this service? Did it give them functionality they didn’t previously have?

How did you fall into data for good?

Anusua Trivedi: I started out as an engineer at Microsoft then began working in the AI for Health space. Healthcare is very different in the sense that it has a lot of monetary support when you are actually building products, but it also is hard from a community-user perspective because everything is so expensive.

So while I was working on products in healthcare, we started putting out things for free on the side, this little app, an assessment or an assistant, and we started building models of sustainability.

That gained a lot of traction from individual researchers and academics who approached us asking, “How can we use this?” That’s when we realized we needed a system where individual researchers can come in and try out these problems. So we started our lab, which is completely grant-based, where we give out grants to educational institutions, hospitals and individual researchers.

Up until now we’ve sponsored about 1,200 individual researchers working on COVID[-related products], which have been fully productionalized for researchers who would not normally have the resources to create it. These grants help them democratize their products for societal good.

Anjali Samani: I also fell into it. I started seeking out real data projects for profit and nonprofit companies and offering data services to them. I realized that many nonprofits were looking for help in the data space but had no idea where to begin. They didn’t even know what data they had. It would be in docs and spreadsheets. They wanted help using the data to structure a marketing campaign or fundraising plan, so I began doing pro bono education and consulting work to help them understand how to use their data to answer their most important company questions.

First, let’s figure out how to use data to keep the lights on and grow, then let’s be more intentional and figure out how to go about collecting that data while bearing in mind cost-effective ways to help them go forward.

It’s not necessarily about building out a fancy database they can’t maintain. I call this “enablement”. It’s helping these organizations, who don’t have a lot of money, still be able to benefit and leverage insights from data in a low-cost, low-effort kind of way.

Collin Cunningham: I don’t have professional experience with AI for good, for me it was more about having something outside of work I was passionate about. My first entry into AI for good actually had nothing to do with AI.

In Atlanta I founded a charity called Spring Forward that went into schools in underserved communities to help level the playing field when it came to college assistance. We went to schools and complemented their guidance counselors by providing services to teach students about filling FAFSA and college applications. Then COVID hit and kind of destroyed our charity because of the in-person and one-on-one interactions with students, but we’re planning to bring it back eventually.

When Spring Forward took a pause, a friend and I started ATL Family Meal, which is kind of like Uber Eats, but free. After my friend in the restaurant industry witnessed all restaurants being forced to lay off their workforce, he founded this charity.

After getting investments and donations, we fund restaurants who cook food, then we have rideshare partners pick up and deliver the meals to those who need it. I built the data science portion of it, building the optimization so we didn’t have to hire a company to build it, which can be very expensive. I helped him get that up and running and in this past week we’ve hit 100,000 meal deliveries.

Over that time I’ve gained an interest in AI for good. It’s finding a mutually beneficial thing where you can find something that has societal impact and it’s valuable for other people and you also learn to do it yourself.

Note: This Q&A has been edited for brevity and clarity

Want to learn how to harness the power of data for good? Find out how Columbia Engineering Data Analytics Boot Camp teaches the skills that shape the way businesses make decisions.