Building Deep Tech with Ilir Aliu Ep 62 | Physical AI Has a Data Problem (w/ Pedro Milcent)
Apr 25, 2025
Pedro Milcent, Co-Founder of Deplace AI, shares his fascinating journey from law to deep tech entrepreneurship. He discusses the challenges of data collection for robotics and physical AI, emphasizing the need for large-scale, low-cost solutions. Milcent reflects on the vibrant tech scenes in Los Angeles and Paris and how his international experiences shaped his approach to innovation. He also highlights the importance of building expert teams and outlines his vision for the future of physical AI in Europe, showcasing his bold insights and strategic thinking.
AI Snips
Chapters
Transcript
Episode notes
Unconventional Education Path
- Pedro Milcent finished high school at 14 and started studying law in Brazil immediately.
- He later shifted to business studies in the US with a full scholarship, showing an unconventional career start.
Physical AI's Data Deficit
- Physical AI lacks accessible large-scale datasets unlike traditional AI.
- Deplace AI is addressing this by collecting diverse, multimodal data to improve robotic model training.
Defining Robotics' ChatGPT Moment
- The true 'ChatGPT moment' for robotics is when robots become generally useful and smart in varied environments.
- This requires vast, diverse training data and smart models capable of handling complex, long-term tasks.