Can data work be fair for all, Mr. Seshadri?

Karya is the world’s first data cooperative to bring dignified, digital work to economically disadvantaged Indians, giving them a pathway out of poverty. The startup empowers them with fair economic opportunities through completing data annotation services for AI clients. For this Karya also collaborates with the BMZ initiative FAIR Forward – Artificial Intelligence for all that supports local and responsible AI innovation in India and worldwide. Can the principles of fair AI be transferred to data workers? Questions for Vivek Seshadri, Co-Founder and CTO of Karya.
Can you still remember when you came up with the idea of Karya?

Yes, it is a long story. There is the National Rural Employment Guarantee Act in India. According to this program, the government essentially provides minimum wage employment to any rural household that asks for it – based mostly on physical labor where people are helping build road networks and other public infrastructure through this program. What we felt was, in addition to physical work, can we also provide digital work to people? Smartphone penetration was increasing, data connectivity was also increasing. So, if even people in low-income communities have a smartphone in their hand, can we provide them with digital work that they can use to gain income? That’s how it started.


That was the idea. I imagine that many questions came afterwards.

Sure. For example, what type of work can people in rural communities do on a smartphone and what will be the quality of work? How much would it cost? Is the idea feasible at all? So, being a research lab, we conducted a series of user studies to answer these questions such as can people in rural communities complete digital tasks efficiently and compete with the market to contribute value. And roughly two years ago, we decided: The idea largely has positive impact and is very feasible. So, can we scale it? We spun off the startup at that point.


And as a startup, can you tell us how many people you have reached with your app as a contractor?

It has been used by many ecosystem partners for their own deployments. Across all the deployments, we have reached around 35,000 people right now.


And what is the potential for growth? Where do you see Karya in ten years?

Our goal is to touch 100 million people through our platform. What we are seeing in the world today is being disrupted at a much faster rate than even a couple of years ago. Obviously, we are going to have to adapt to the changing ecosystem. The aim is to provide people with work that will generate supplemental income that will help them build a critical amount of savings. Simultaneously, we are exploring ways in which we can provide people with skills that they need to survive in this digital world and connect them to opportunities for long term employment.


Would these fair wages also work in a pure business model?

There’s a big gap between the value a dataset creates and the wages that people on the ground are actually getting paid to generate and curate that data. The best way to bridge this gap is to increase the wages. We could very easily support our team through the revenues that we generate. So, we strongly feel not just that this is a feasible model, an ethical world should strive to implement this model because the value is just huge.


Isn’t it frustrating that the big profits always remain somewhere else?

That’s the world that we are fighting today. There is more wealth created in the world than ever before. And a lot of the young population in general are also realizing the need to more equitably share that wealth with everybody. Even people who contribute positively to the society may not always be doing well. How do we move to a society where everybody has their basic needs satisfied so that they don’t have to worry about putting food on the plate? That’s the challenge. When people are worried about their next day, it results in a race to the bottom kind of economy where they will do anything to earn money.


How can the market change?

We stop looking at data as work, but data is more as an investment where people get continuous value from the data that they are generating, like software subscriptions. What does it mean to subscribe to a piece of software? It means that as long as this software is generating value for me, I’m going to keep paying for it. Can we create a similar world for data?


How could global economy benefit if data workers were better informed and paid? I mean, that would cost more in the first place.

There is an ethical aspect of it and there is a quality aspect of it. You pay your data workers well, you get better quality. Now moving to the ethical aspect of it, I’m just going to give you an analogy. Organizations profited out of private data of individuals, and they’ve been doing this for a long time. When new laws like GDPR said private data is owned by the people to whom it belongs, and therefore they should have the rights to control how the data is used and who uses the data and revoke permissions. This is a legal framework that decided that it is bad for companies to profit of private data of individuals for whatever reason. I think it’s the same question here.


How can data workers benefit from new technologies like AI?

Let’s take the most recent disruption. People have been using ChatGPT to improve their lives, to improve their efficiency of how they do. If I could do my job, which took one hour and complete it in five minutes with the help of ChatGPT, it has given me more time for the same work, for the same pay. The efficiency again gives access to more resources. People in low resource settings don’t have access to technology, don’t have access to resources, and don’t have access to services that can provide the same kinds of efficiencies. This results in a vicious cycle. Data work can mitigate this problem by creating more resources in those languages and those settings, which will hopefully create technology in those settings.


When somebody tells you that AI is still is a globally northern affair, would you agree?

Yes and no. Technology typically gets built for the top 5-10% of the society. Everything else is trickle down. And I think AI is no different. Take smartphones, for example: People in the highest state of society started owning smartphones and people saw the value of having a computer on your hand and the efficiencies that brought to their world, and slowly the world realized: Hey, this is efficiency I can give to everybody. And they started manufacturing lower end smartphones. Now most people have smartphones.

Vivek Seshadri is the Cofounder and CTO of Karya and a Principal Researcher at Microsoft Research India. He is a computer scientist and an entrepreneur working on problems in the intersection of technology and global development.

His vision is a world where no one lives in poverty and all people have equal opportunities. Towards this vision, he has been using his technical knowledge and expertise to enable dignified income opportunities for people in low-income communities.