Initially, I thought the lane-keeping problem for self-driving cars was too difficult to solve, but it turned out to be manageable with the right approach.

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.

I believe that breakthroughs at the intersection of information technology and physical industries, like self-driving cars, could significantly change the landscape of innovation.

By 2028, I believe we'll have advanced AI systems capable of solving complex problems that humans can handle, thanks to significant ongoing work in AI development.

Our vision is that every moving machine will be intelligent, with self-driving being just one small part of that intelligence.

Despite the advancements, society is cautious about adopting AI technologies, as seen with self-driving cars, which are statistically safer but face public skepticism.

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.

I believe AI could potentially solve one of the seven Millennium Prize Problems, but we're still a long way from that capability.