They started really collecting data 2 years ago with 2000 beta testers. They collected 90 million miles by the end of the second year. Thats pretty damn epic and the improvements have been impressive. But fsd won’t be anywhere near perfect till they hit billions of miles. I think elon has been quoted as needing 6 billion miles and i wouldnt be surprised if it’s more… with 400k beta testers, we are years away. It will be gradual and iterative but the point is, it’s a matter of time. Is fsd anywhere near perfect? Hell no. Is the improvement progressing as needed to solve fsd within the next decade? Hell yes

Tell me one company who has access to the sheer amount of data required to pull off Tesla’s approach. There is a reason no one is even attempting to brute force their way to autonomy like tesla is… its a complete lack of training data

I mean with that description, you could be a 15 dollar an hr tech. What experience do you have with machine learning? What experience do you have with modeling and training?

Ha.. im genuinely curious as to the work he has done to be such an authority

There is no way he is a junior ai engineer. There is no way he is a ai engineer period

Hmm… what do you do exactly? Got a link to your github?

Elon isnt but lemme guess, you, a nobody on reddit, are the authority to say what is needed for self driving? Get over yourself