In the premiere of the Intrepid Growth Partners’ podcast, we meet with Rae Jeong, the CEO of Maneva AI, a Toronto-based company whose mission is to use artificial intelligence to make factories autonomous—completely run, maintained and improved by machine-learning algorithms and manufacturing robots.
Rae joins our panel of experts:
Ajay Agrawal, co-founder and partner, Intrepid Growth Partners
Richard Sutton, Intrepid GP senior advisor, pioneer of reinforcement learning and professor, University of Alberta
Sendhil Mullainathan, Intrepid GP senior advisor, MacArthur Genius grant recipient and professor, MIT
Niamh Gavin, Intrepid GP senior advisor, Applied AI scientist and CEO, Emergent Platforms
“The ambition that we have… is to leverage some of the work that [Derby Mill host Richard Sutton] pioneered in reinforcement learning,” says Jeong, who foresees a future with zero workforce injuries and where the cost of production is simply the cost of energy.
LINKS
Maneva AI website
Maneva CEO Rae Jeong LinkedIn
A short video about Maneva’s work transforming Laura Secord chocolate production
Be sure to catch every episode by subscribing on the following platforms:
Substack // YouTube // Spotify // Apple Podcasts
DISCUSSION POINTS
00:00 Introduction
02:00 Ajay Agrawal introduces Rae Jeong, Maneva’s CEO.
05:48 Rich Sutton commends Rae Jeong's mission to make factories autonomous.
06:46 Experts explore the challenges of continuous learning.
08:04 Rae Jeong highlights the importance of human expertise in feedback loops.
09:25 Niamh Gavin suggests supplementing real-world data with novel edge cases.
11:34 Rae Jeong on the need for real-world data for optimized reinforcement learning.
14:00 Sendhil Mullainathan asks Rae Jeong how accurate reinforcement learning is at managing labels and actions.
17:45 Rae Jeong discusses the improvement of foundation models like Toyota, lean manufacturing, and the next kaizen.
20:18 Sendhil Mullainathan discusses quality control algorithms and feedback loops by asking how to make something RL-able.
22:50 Rich Sutton wonders what taking AI “to the limit” looks like in any factory.
25:52 Rich Sutton considers what it would be like to incorporate longer loops into algorithm training.
26:15 What autonomous AI-equipped factories could look like at scale.
27:47 How to transition from single-point, vision-to-action AI, to an assembly line or factory.
30:30 The biggest challenge of factory-wide credit assignments.
33:09 The way factory operations changed after the advent of electricity.
37:19 Commentary wrap-up.
Share this post