AI Expert Predicts Tesla's Full Self-Driving 12.X Will Surpass Human Drivers

Artificial intelligence continues to advance, and with it, the capabilities of autonomous vehicles are reaching unprecedented levels. AI expert James Douma has analyzed the developments in Tesla's Full Self-Driving (FSD) technology and suggests that the upcoming version 12.X will not only match but exceed human driving capabilities. According to Douma, this new version of Tesla's FSD will require significantly fewer interventions from human drivers, potentially lowering them to 1% or less compared to the earlier version 11.

Tesla's Pursuit of Full Autonomy

The journey to fully autonomous driving has been a challenging road, with each new update bringing Tesla one step closer to this goal. The FSD version 12.X is expected to benefit from enhancements in data processing and computation power. With the introduction of more than 15,000 Nvidia H100 chips, Tesla has expanded its training compute to over 50 Exaflops, providing the kind of power needed to manage the complexities of real-world driving.

Intervention Rates and Progress

Intervention rates are a critical measure of autonomous driving systems' success. Douma highlights that if interventions were previously occurring once every drive, a reduction to 1% would mean drivers might only need to intervene once every three months to a year. Additionally, reports suggest that Tesla is making 5 to 10 times faster progress each month. If these trends hold, Tesla could reach the 1% intervention rate by as early as June 2024, paving the way for the FSD system to exit beta testing and rapidly adapt to various global driving conditions. Following this trajectory, the advent of Tesla robotaxis could be as close as 6 to 24 months away, contingent on continued rapid progress and verification of Douma's predictions.

Tesla, FSD, autonomous