Understanding the challenge and our interdisciplinary approach to building virtue-centered AI
AI conversational agents (CAs) are increasingly being used in counseling, coaching, and companionship roles despite having a limited evidentiary basis. This is due to several factors:
Worryingly, current AI conversational agents risk fostering moral atrophy by focusing on short-term gratification and entertainment, potentially undermining the development of essential character virtues (CVs) necessary for genuine human flourishing.
Psychology, Computer Science, Philosophy
Months to years, not days
We conjecture that nearly all extant AI conversational agents trained to enhance short-term subjective well-being (e.g., engagement, amusement) may not translate to character virtues because they:
Focus less on virtue-inducing emotions (achievement-oriented like pride, or other-directed like gratitude)
Do not consider philosophical underpinnings of long-term and collective focus (vs. immediate and individual)
Do not overcome limitations of short-term AI architecture that character virtue development requires
Our interdisciplinary approach addresses critical gaps in current AI development
How do we design and deploy AI conversational agents that are psychologically and ethically guided for character virtues?
How effective are the best-designed AI conversational agents at promoting character virtues over long-term engagements?
Combining novel interdisciplinary frameworks with rigorous empirical testing
Bringing together experts across psychology, computer science, and philosophy to develop comprehensive frameworks for virtue-centered AI design.
Proposing and implementing several technical approaches to designing AI conversational agents that optimize for character virtues rather than short-term engagement.
Conducting experimental and longitudinal studies with behavioral assessments (e.g., phone use patterns, video analysis) to obtain evidence on long-term effectiveness.
Publishing results, open-sourcing virtue-centered AI agents, sharing datasets, and creating educational materials to advance the field.
Our project will produce multiple valuable outputs for the research community and beyond
Multiple peer-reviewed papers presenting our frameworks, technical approaches, and empirical findings
Virtue-centered conversational AI systems available for researchers and developers to use and build upon
Comprehensive datasets from our longitudinal studies to enable future research
Websites, videos, and documentation showcasing key aspects of AI creation and effectiveness
Conference presentations and public talks to disseminate our findings widely
Frameworks to guide future research and public policy on AI use for character virtues and human flourishing
If funded, our work will serve as a foundation to guide future research on AI conversational agents for enhancing character virtues, inform the adoption of AI conversational agents across diverse contexts, and shape public policy on AI use for character virtues and long-term human flourishing.
By bridging the gap between short-term engagement optimization and genuine character development, we aim to redirect the trajectory of AI development toward systems that truly serve human flourishing.
This research is generously funded by
We are grateful to the John Templeton Foundation for their support of this interdisciplinary research on AI for human flourishing. The opinions expressed in this project are those of the authors and do not necessarily reflect the views of the John Templeton Foundation.