TESLA'S BITCOIN STASH SWELLS TO $2.4 BILLION
"A major part of real-world AI has to be solved to make unsupervised, generalized full self-driving work, as the entire road system is designed for biological neural nets with optical imagers," Musk tweeted Thursday.
Although self-driving technology has made "great progress," Musk disclosed in a recent earnings call with investors that it's "one of the hardest technical problems" that has potentially ever existed.
"In order to solve it, we basically need to solve a pretty significant part of artificial intelligence, specifically real-world artificial intelligence," he said.
He said the road system is "designed for a neural net computer" and that "the entire road system is designed for vision with a neural net computer, which is because it's designed for eyes in the brain."
When it works, he said, it will be better than "the best human because it's like having eight cameras, it's like having eyes in the back of your head, beside your head and has three eyes of different focal distances looking forward."
It about developing a system that has "very good eyes" that can see in all directions.
"There's no question in my mind that with a pure vision solution, we can make a car that is dramatically safer than the average person," he said.
Although it's no easy task, Musk claimed that Tesla is well equipped to handle the challenge as the company is "developing one of the strongest hardware and software AI teams in the world."
"We built the team from scratch and have been developing what we think is probably the most advanced real-world AI in the world," Musk said. "Certainly, we appear to be able to use things with full self-driving that others cannot."
The company is already progressing with making self-driving technologies more reliant on vision-based sensors.
In a recent SEC filing, the electric car company said it is currently developing "self-driving and drive-assist technologies to rely on vision-based sensors, unlike alternative technologies in development that additionally require other redundant sensors."
However, "there is no guarantee that any incremental changes in the specific equipment we deploy in our vehicles over time will not result in initial functional disparities from prior iterations or will perform as expected in the timeframe we anticipate, or at all," the filing read.