So I just got FSD beta and I have to say that I'm not all that impressed. I do think the problems I experienced can be solved - but I believe at least some of these problems have been around a while in some shape or form. I'd really be curious to know why they haven't been fixed yet. My comments after a handful of short test drives in the city.
- Steering needs to chill out. I had multiple instances where the car decided it needed to change direction NOW. Not smooth or confidence inspiring. Lightly holding the wheel it nearly caused me to lose grip it started spinning so fast once. You definitely get knocked around a bit.
- Lane choices and choice of position on road are weird at best: Normally you expect the car to pull into the bike lane when making a right turn. In traffic if it doesn't pull into the bike line, you want to take over to avoid impeding traffic. Sometimes it does, sometimes it doesn't. When it does, see #1. Had an instance where it chose the left of two turn lanes when an immediate right 100 meters down the road needed to be taken. At other times, it's cut left-corners at too high a speed in a residential neighborhood without lane markings - most people slow down and take the corner wider to avoid possible oncoming traffic.
- Accelerator - There needs to be some better choices made in terms of how closely to follow speed limits - on residential roads, it should reduce max speed and accelerate more gently. It seems to be a bit too cautious coming to a stop at stop signs, then very slow to resume moving. When it decides to go, though - hang on tight!
The visualization effects that come with FSD beta are phenomenal - this style of graphics definitely needs to be included in standard builds. If nothing else, it's a FSD marketing tool.
Honestly, I feel like just adding some damping on directional changes (steering, acceleration and braking) would go a long ways towards giving passengers more confidence on what the car is doing or going to do. As it is now, it is highly unpredictable and it does not induce trust that the car is going to do the right thing.
So having "tried Tesla FSD" - I am highly skeptical it will be ready in a year given the pace of progress. Most of these issues have been around a while based on experiences with EAP and early FSD beta feedback. Maybe Tesla has opted to focus on other areas of the stack and decided that these are minor issues to be resolved later. Hard to say, really, but I can honestly say that now having finally tried Tesla FSD I believe Chollet's timeframe more than Elon's...