Chapter
The Future of Self-Driving Cars
19:35 - 27:01 (07:26)
The development of self-driving cars requires a 100-fold improvement in computer power, data, and software algorithms. However, the amount of available data to train these cars is limited, leading experts to predict a focus on well-mapped areas rather than unexplored territory.
Clips
The continued progress of AI is dependent on improving computer power, data, and software techniques, but there may come a time where the internet may run out of human-generated data to train on, leading to a ceiling effect in AI advancements.
19:35 - 23:33 (03:57)
Summary
The continued progress of AI is dependent on improving computer power, data, and software techniques, but there may come a time where the internet may run out of human-generated data to train on, leading to a ceiling effect in AI advancements.
ChapterThe Future of Self-Driving Cars
Episode#141 – Erik Brynjolfsson: Economics of AI, Social Networks, and Technology
PodcastLex Fridman Podcast
Andrew Eng believes that autonomous vehicle companies should focus on level four, where the car only goes in well-mapped areas, rather than tackling every possible scenario.
23:33 - 27:01 (03:28)
Summary
Andrew Eng believes that autonomous vehicle companies should focus on level four, where the car only goes in well-mapped areas, rather than tackling every possible scenario. Waymo has embraced this approach by allowing the public in Phoenix to ride in their self-driving car without a safety driver.