Have had the beta (11.3.6) for the last few days and been doing a lot of driving with it. One thing Ive noticed is it won't always repeat incorrect behaviors. An example being at first when leaving my apartment complex it would go into the right lane which very quickly merges into a single lane and then has to correct for this. Now it consistently will turn into the appropriate lane initially. There are other instances Ive noticed it no longer doing incorrect behavior. My question then is, is FSD capable of learning on its own and improving and taking from previous disengagements to improve, separate from actual software updates? Or is it simply just making different decisions due to the sheer number of variables there can be when driving?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD constant improvement ?
- Thread starter Jacob Lamczyk
- Start date
No, all learning is in the neural net/mothership. Every time it see something it is the "first time" it has ever seen it. So it can react differently every time since it is all" new data" for it to analyze.
Your car can make different decisions because off changes in traffic data or even invisible small Map Data deltas updates than can be sent to your car.
Last edited:
Similar threads
