|
|
|
|
|
|
by cscurmudgeon
1823 days ago
|
|
|
> It’s already about 10x. It is not from your data alone. It maybe, but the data is not from a controlled experiment. Because: 1. Autopilot disengages in dangerous/ambiguous situations.
2. The set of users with autopilot is different from those without. |
|
|