July 7, 2025
How can artificial intelligence create positive social impacts? Assistant Professor Ryan Shi is empowered to do just that with a recent grant from the Google Academic Research Awards. This highly competitive program supports academic projects that aim to use artificial intelligence to create positive societal impact. Awarded under the Society-Centered AI track, Shi’s project explores how to better engage smallholder farmers using digital tools like chatbots and smartphone applications. While many such technologies already exist, they often fall short in reaching and retaining users.
"This grant is about trying to figure out ways to better engage with these farmers. The question isn't just about access to tools, it's about actual usuage and long-term interaction," said Shi.
Though this agricultural work may seem unrelated to Shi's earlier food rescue research, both of his projects are bound by a shared technical foundation: user engagement in socially-driven platforms.
Before the Google grant, Shi, a faculty member with the Department of Computer Science, worked on optimizing volunteer engagement in food rescue operations, an area akin to food banks, where food is redistributed from donors to food-insecure communities via volunteer drivers. Unlike commercial delivery services, these volunteers are unpaid and cannot be assigned tasks directly.
To make the most of volunteers’ time and energy, efficiency was essential. In this work, Shi developed algorithms to balance user attention and participation, a delicate optimization that requires understanding both human behavior and computational efficiency. The lessons learned in that space now inform his efforts in digital agriculture, where similar patterns of engagement and dropout pose challenges to scaling AI for good.
“The food rescue work came directly from my PhD,” Shi explains. “But this new project on smallholder farmer engagement is something I’ve never touched before — not during my graduate studies.”
As part of the Google-funded research, Shi initiated a new collaboration with Digital Green, a non-governmental organization focused on agricultural development. This collaboration was built from scratch as an independent principal investigator, emphasizing his move away from relying on prior networks or established research pipelines.
In stepping into a completely new application domain, Shi is broadening both the impact and scope of his research agenda. While his technical interests in user engagement remain consistent, he is exploring new frontiers, both intellectually and institutionally, that define his work as a rising academic voice in AI for social good.
"Working in real application domains doesn't just help us test ideas - it forces us to come up with better ones," said Shi.
Application Inspiring Innovation
Across both the food rescue and agricultural domains, Shi’s work reveals how real-world constraints can push the boundaries of technical research. Rather than simply applying existing tools, he often finds himself developing entirely new algorithmic approaches in response to human-centered challenges.
In fact, much of the technical innovation in his current project emerged from early field studies conducted by his collaborators at Digital Green in the Global South, which uncovered a significant barrier to engagement: limited access to mobile data and shared phone usage within households, especially among female farmers.
Shi explained that cell phone data isn’t as affordable or accessible as we assume it is in the U.S. In many families, the father has his own phone, while the mother shares a device with the children. That means there are only certain windows during the day when she can even access the chatbot.
This dynamic presented a major challenge: How do you design engagement algorithms for users who may be offline most of the time, or may not even control when they can use their phone?
Shi and his team are now developing algorithms that adapt to these nuanced usage patterns, respecting not only when users are likely to be online, but also the socio-cultural contexts that shape their digital presence. His team can’t assume that users are always reachable, therefore they are working to build systems that are smart and sensitive to these real-world limitations. For Shi, these kinds of insights are not peripheral, they’re core to his research ideologies.
"We can't assume that users are always reachable. So we're working to build systems that are smart and sensitive to these real-world limitations," Shi said.
Balancing Academia and Purpose
As a junior faculty member, Shi faces the familiar academic hurdle of time management: juggling research, collaboration, mentoring, and publishing. But the nature of his socially rooted work adds an extra layer of complexity.
He noted that one of the biggest challenges is deciding how to prioritize, as he is constantly choosing between making real-world impact and advancing technical research. While Shi’s work often finds the sweet spot between these two goals, he acknowledges they don’t always align. In many of his partnerships with nonprofit organizations, for instance, technical infrastructure is minimal, limiting what’s feasible.
“Sometimes they just want a dashboard,” Shi explains. “But then I have to ask, what will you do with this dashboard? How will it actually support decision-making?”
Unlike tech giants, nonprofits often lack the systems and training to turn raw data into actionable insights. This disconnect means Shi’s work must not only generate novel algorithms but also navigate the realities of deploying them in environments that aren't built to support complex technology.
Still, for Shi, these challenges are what make the work meaningful. His approach, grounded in community collaboration and responsive innovation, is shaping a new vision for AI research: one that’s socially aware, technically rigorous, and unafraid to tackle the messiness of real-world problems.
"One of the biggest challenges is deciding how to prioritize. You have to say no to a lot of things - and even when you say yes, you're constantly choosing between making real-world impact and advancing technical research," said Shi.