New
AI & RoboticsMore in AI & Robotics→
AI training data may become the next $100 billion race
Key Takeaways
- Andrew Ho argues many economically valuable skills are poorly represented in current AI datasets.
- His new company is targeting bioinformatics and routine lab-work datasets first.
- The thesis challenges the idea that scaling compute and model size alone will deliver broad generalization.
DE
DT Editorial Team··via the-decoder.com