Centering joy in AI development and implementation
PIK Professor Desmond Upton Patton—of Annenberg and SP2—and collaborators introduce a joy-informed framework designed to initiate conversations among engineers, designers, and researchers.
2 min. read
Much of the research on artificial intelligence and human emotion focuses on potential negative implications and emotional states such as depression, with less focus on joy, gratitude, and connection. But Penn Integrates Knowledge professor Desmond Upton Patton says he and collaborators envision a future where people’s experiences—from social media to banking—are shaped by systems that seek to amplify joy.
“Current tech business models focus primarily on engagement without regard for whether such engagement stems from joy or distress, anger, and frustration,” they write in a new paper published in The Paris Journal on AI & Digital Ethics.
In the paper, the researchers present a framework for instead considering how to incorporate joy in the design and deployment of AI models. They are specifically focused on empowering marginalized communities, who experience greater bias from large language models. The goal of the framework is to assist engineers, designers, and researchers in developing AI models that promote more nuanced, compassionate AI interactions.
“We define joy as both an emotion and a tool for liberation, freedom-building, and reimagining possibility,” says Patton, who has primary appointments in the Annenberg School for Communication and School of Social Policy and Practice. “A joy-centered approach to AI challenges researchers and engineers to design models that recognize human strengths, elevate resilience, and foster thriving digital environments.”
In a study that informed the creation of the framework, researchers recruited New York City university students and Black Harlem residents to use Integrating Emotional Stories Online, a social media platform developed by computer scientists, linguists, and social work researchers to examine how people express grief and distress online.
Prompted to share their emotions and events that triggered them, the 125 users shared 949 posts from February to December 2022. The most striking theme the researchers identified was joy, with users writing about hope, gratitude, love, ease, and spirituality as coping mechanisms.
“People experiencing loss still recognize, seek, and value moments of joy,” Patton says. “That insight pushed us to consider how AI could intentionally optimize for joy, especially in a world where trauma, grief, and pain are part of everyday life.”
Their framework lays out how joy could be centered in team creation, conception, design and updates, and deployment of AI models. For example, they say that teams creating AI models should include not only engineers and computational modelers but also social scientists such as sociologists, anthropologists, and linguists.
They propose four guiding questions to support this approach:
1. What happens when joy is used as a tool for AI imagination and design?
2. What processes are needed to adapt models to identify and account for joyful concepts?
3. How can joy shape assumptions, goals, and team reflexivity?
4. What is the hope or promise of joy-centered AI?
“This framework also invites researchers and engineers to reimagine the functions and responsibilities of AI,” the authors write. “For example, what might a joyful content moderation system look like, particularly in moments of political tension or cultural conflict? How could recommendation systems be optimized to support curiosity, healing, or joy instead of engagement at all costs?”
This paper initiates a conversation on “what it means when joy is absent and how we might solve it,” Patton says, building on pioneering work at Penn in positive psychology. Looking ahead, he says researchers plan to test and evaluate the framework on JoyNet, a new machine-learning-powered platform designed to bring joyful experiences to youth who spend a lot of time on social media.
Desmond Upton Patton is a Penn Integrates Knowledge University professor with joint appointments in the Annenberg School for Communication and School of Social Policy and Practice (SP2) and a secondary appointment in the Department of Psychiatry in the Perelman School of Medicine. He is the Waldo E. Johnson Jr. Professor in Annenberg, chief strategy officer at SP2, and director of the SAFELab.
The other co-authors are Shana Kleiner of SAFELab and Penn Center for Inclusive Innovation & Technology at SP2; Shug Miller of SAFELab; Nicholas Deas, Kathleen McKeown, and Blake Vente of Columbia University; Jessica A. Grieser of the University of Michigan; and James Shepard of the University of Tennessee.
This research was supported by the National Science Foundation.
Nanoparticle blueprints reveal path to smarter medicines
New research involving Penn Engineering shows detailed variation in lipid nanoparticle size, shape, and internal structure, and finds that such factors correlate with how well they deliver therapeutic cargo to a particular destination.
A generous gift from alumni Glenn and Amanda Fuhrman brings the work of internationally acclaimed artist Jaume Plensa to the University of Pennsylvania. The latest addition to the Penn Art Collection expands Philadelphia's public art.
A massive chunk of ice, a new laser, and new information on sea-level rise
For nearly a decade, Leigh Stearns and collaborators aimed a laser scanner system at Greenland’s Helheim Glacier. Their long-running survey reveals that Helheim’s massive calving events don’t behave the way scientists once thought, reframing how ice loss contributes to sea-level rise.
