Finding the scaling laws of agent: interviewing Guohao Li and his Camel AI

https://is1-ssl.mzstatic.com/image/thumb/Podcasts125/v4/00/fc/e7/00fce7b3-27d4-c9ea-537f-17c0fba6e734/mza_10735902237024048181.jpg/600x600bb.jpg

the talktank

LSE SU Entrepreneurs

87 episodes

3 days ago

You’ve reached the talktank - the official podcast of the LSE Entrepreneurs Society. We delve into the minds of those who think, live, and breathe outside the box, by interviewing thought-leaders about their work, experiences and stories they've collected along the way. Through our four series called "Beyond Profit", "Artistpreneurs", "Unwrapped" and "Bits and Bytes" we explore diverse areas including social entrepreneurship, the creative arts and technology. Our aim is to expose our audience to diverse perspectives and inspire the next generation of change-makers.

Entrepreneurship

Business

RSS

All content for the talktank is the property of LSE SU Entrepreneurs and is served directly from their servers with no modification, redirects, or rehosting. The podcast is not affiliated with or endorsed by Podjoint in any way.

Entrepreneurship

Business

https://d3t3ozftmdmh3i.cloudfront.net/staging/podcast_uploaded_episode/2681391/2681391-1739984496085-d8d9157e22554.jpg

Finding the scaling laws of agent: interviewing Guohao Li and his Camel AI

the talktank

53 minutes 1 second

8 months ago

Finding the scaling laws of agent: interviewing Guohao Li and his Camel AI

In this episode, Dr. Guohao Li, founder and CEO of Camel.ai, discusses the vision behind his open-source project aimed at developing multi-agent systems.

Last year, Camel and Microsoft AutoGen were involved in an academic "conflict" due to AutoGen quoting their early works on Multiagent yet didn't give enough credit to the Camel team. It would be possible to say, that Camel is the earliest organization and research team to step into multiagent systems, almost half a year ahead of Microsoft.

About Dr. Li,

Guohao Li is an artificial intelligence researcher and an open-source contributor working on building intelligent agents that can perceive, learn, communicate, reason, and act. He is the core lead of the open source projects CAMEL-AI.org and DeepGCNs.org.

Guohao Li was a postdoctoral researcher at University of Oxford with Prof. Philip Torr. He obtained his PhD degree in Computer Science at King Abdullah University of Science and Technology (KAUST) advised by Prof. Bernard Ghanem. During his PhD studies, he was fortunate to work at Intel ISL with Dr. Vladlen Koltun and Dr. Matthias Müller as a research intern. He visited ETHz CVL as a visiting researcher. He also worked at Kumo AI and PyG.org with Prof. Jure Leskovec and Dr. Matthias Fey as a PhD intern. His primary research interests include Autonomous Agents, Graph Machine Learning, Computer Vision, and Embodied AI. He has published related papers in top-tier conferences and journals such as ICCV, CVPR, ICML, NeurIPS, RSS, 3DV, and TPAMI.

Now, he is building his startup to continue the research to find the scaling laws of AI in the multiagent system, listen to his story in the new TalkTank podcast!

00:00Introduction to Dr Guohao Li and Camel.ai

02:47The Vision Behind Camo.ai and Open Source

05:58Open Source vs. Closed Source: A Business Perspective

09:06Building a Community: Contributors and Users of Camel.ai

11:51Use Cases of Camo.ai in Startups

15:13Challenges in Building Reliable AI Agents

18:02The Importance of Multi-Agent Infrastructure

22:04Timeline for Multi-Agent Infrastructure Development

24:54The Future of AGI: Perspectives and Predictions

30:11Scaling Laws and the Role of Multi-Agent Systems

33:50Camel.ai's Initiatives and Achievements

37:06Balancing Research and Commercialization

40:51Collaboration with Mosaic ML and Competitive Landscape

43:04Academic Influences: Learning from Professor Philip Tull

48:56Advice for Aspiring Entrepreneurs

54:01Closing Thoughts and Future Aspirations