Speaker
Description
Many grand challenges like climate change and pandemics emerge from complex interactions of millions of individual decisions. While LLMs and AI agents excel at individual behavior, they can't model these intricate societal dynamics.
Enter Large Population Models LPMs: a new AI paradigm simulating millions of interacting agents simultaneously, capturing collective behaviors at societal scale. It's like scaling up AI agents exponentially to understand the ripple effects of countless decisions.
AgentTorch, our open-source platform, makes building and running these massive simulations accessible. It's optimized for GPUs, allowing efficient simulation of entire cities or countries. Think PyTorch, but for large-scale agent-based simulations. LPMs are already making real-world impact. They're being used to help immunize millions of people by optimizing vaccine distribution strategies, and to track billions of dollars in global supply chains, improving efficiency and reducing waste.
In this talk, we'll dive into the underlying technology of LPMs, provide an overview of AgentTorch, and explore how you can contribute to both our research at MIT and the open-source project. We'll also discuss intriguing topics like prompting LPMs versus LLMs, opening up new avenues for AI development. Whether you're interested in complex systems, AI, or tackling global challenges, LPMs offer exciting opportunities for innovation and impact.
Session author's bio
Hardik is a final year undergraduate student at IIT Kanpur. He has huge intertests in Artificial Intelligence, Language Models and Multi agent systems. He has been closely working with the MIT Media Lab for the past 6 months and has helped in developing and improving AgentTorch. Hardik also completed the GSoC'24 successfully under the organisation C2SI. Since then, he has been an active open source developer and has contributions in various organizations.
Any other info we should know?
A similar talk was presented in the previous edition of OOSC
| In Person Attendance | In-person |
|---|---|
| Please confirm that there are included headshots of all speakers in their profiles | Yes |
| Level of Difficulty | Intermediate |
| Agree to Privacy Policy and Notice | I agree |