Tesla has released the latest point version of the Full Self-Driving (FSD) (Supervised) software, with an update spotted this week. In the release notes for the new v12 versions, Tesla highlighted that the updated software eliminates over 300,000 lines of human-written code that previously controlled functions such as steering, acceleration, and braking. Instead, the software transitions to rely completely on its neural network, trained by millions of frames of video from real-world driving scenarios.
This development marks a significant shift in the technology’s approach, emphasizing the reliance on neural network training over traditional human-written code. The transition to a neural network-based approach raises questions about the software’s safety and adaptability in real-world driving conditions. Furthermore, Elon Musk’s reiteration of the FSD licensing offer for other automakers adds another layer to the discussion.
It suggests that Tesla is open to sharing its FSD technology with other companies, potentially impacting the future of autonomous driving across the industry. Overall, Tesla’s rollout of the updated FSD software and the shift towards neural network reliance prompts a need for careful consideration of the implications for autonomous driving and the broader automotive industry.