Part 9/15:
Historically, RL faced barriers like computational expense and simulation fidelity. Recent advances—particularly the advent of massively parallel GPUs, improved physics simulators, and more efficient algorithms—have lowered these barriers.
Phil Bisel highlights Elon Musk’s statement that RL techniques are now central to Tesla’s development process, evident from their internal projects at XAI and other ventures. For example, the companies are experimenting with models like Grock, scaled incrementally with the amount of RL-based training, yielding much smarter and safer AI systems at a fraction of previous costs.