Interconnects Interviewing Andrew Carr of Cartwheel on the State of Generative AI
Oct 31, 2024
Andrew Carr, co-founder and chief scientist at Cartwheel AI, is on a mission to create innovative text-to-motion models for creative fields. He dives into how generative AI can enhance creativity through niche applications, like AI-generated poetry. Andrew shares insights from his time at OpenAI and discusses the fascinating interplay between AI and art, emphasizing the need for human oversight. He also explores the evolving AI landscape and the importance of fostering a positive research culture in tech companies to drive impactful innovations.
AI Snips
Chapters
Transcript
Episode notes
Blender Experiment
- Andrew Carr's interest in text-to-motion AI stemmed from his work on Codex at OpenAI.
- He experimented with Blender and its Python interpreter, realizing the potential for animating 3D characters with code.
Cartwheel's Differentiated Approach
- Cartwheel AI uses its own hardware, models, and labeled motion data, avoiding Blender due to its instability and limitations.
- Cartwheel's approach focuses on generating motion data directly, enabling realistic character animation with kinematics.
Motion vs. Robotics
- Cartwheel's text-to-motion AI differs from robotic foundation models by focusing on storytelling and emotive characters, rather than physical accuracy.
- The lack of physics and focus on visual appeal simplifies generating fantastical movements.