I believe self-driving technology will eventually be fully solved as it is a bounded problem, unlike more complex scenarios.

The implementation of self-driving technology is challenging; while the ideas are easy, the actual engineering work, such as optimizing compilers and reducing latency, is what makes it difficult.

The challenge of self-driving includes solving perception problems, such as recognizing objects and predicting their movements, even when they are occluded.

In self-driving cars, supervised learning can help label the positions of other vehicles based on sensor readings.

Tesla's camera-only approach for self-driving cars is scalable and efficient, while competitors like Whimo are currently ahead in deployment experience.

There's a psychological barrier regarding the acceptance of self-driving cars, as people are accustomed to the current level of road fatalities and may resist change despite the potential for improved safety.

To advance AI, we need to focus on building infrastructure that can handle edge cases, similar to the journey of self-driving cars, which have taken years to develop due to the complexities involved.

To achieve full self-driving, we need to recreate how humans drive, which involves using optical sensors and neural networks to interpret the road environment.

The journey of Tesla's autopilot has been incredible, especially considering the challenges of computer vision and the skepticism surrounding it.