Allow me to discard your opinion, unless you're an expert in AI and self-driving, your opinion is merely based on hopes and dreams.
For clarification, are you an expert on AI and self driving?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
-
Tesla's Supercharger Team was recently laid off. We discuss what this means for the company on today's TMC Podcast streaming live at 1PM PDT. You can watch on X or on YouTube where you can participate in the live chat.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
For clarification, are you an expert on AI and self driving?
2daMoon
Mostly Harmless
Do you know with any degree of certainty that 500 wasn't also 1 in 10 for this group?
I can see how a more narrow focus could see it the way you presented it. However, with no idea what percentage of the whole Supercharger group that 500 would be, I wouldn't have enough data to make the connection between "deci" and that percentage.
Though, it wouldn't surprise me one whit if, worldwide, there were 5000 or more people in the overall Supercharger group. If it were found to be true that including all aspects of Supercharger operation (planning, permitting, installation, service, manufacturing) across several continents the "deci" connection might actually be valid when applied there as well.
Anyway, the statement was about the funny coincidence based upon how the layoff was stated as being 10%. The number that everyone was familiar with being a known reference point. Either way, those 500 were part of that 10% which Elon had said would be laid off.
jeewee3000
Active Member
To add to this, current FSD 12.3.x is not feature complete since for example reversing/switching gears is not in the set of possible actions FSD can perform.
In a few months this will be included (if we can believe Ashok/Elon tweets) which will adress many interventions due to the car getting "stuck" after being unable to complete a certain turn due to lack of space.
Once FSD is "feature complete" (i.e. from door to door or parking space to parking space, including reversing) it is more reasonable to compare with Waymo and the like (except that FSD is general versus the mapped areas).
Regarding the intervention rate, I'm curious why Tesla's intervention data seems so out of touch with the loud voices on X.com/TMC stating otherwise.
I'm guessing it has to do with the fact that the general happy impressed customer is less likely to express an opinion online versus the unhappy users.
Also when I see some critiques regarding say AI Driver videos, the negative comments are often nitpicky and not critical for FSD to work. (So many comments regarding the system being too slow. I'd say better safe than sorry. When you're a passenger you won't mind the 1 minute added to your trip versus being scared ****less in a racing FSD-car.)
/endofrant. FSD is making steady improvements.
If you give a "yes/no" answer, like the poster you're advocating for, then it means your knowledge and status in the field is such that no other explanations are required. I always added extensive arguments to my points.
Given the reporting: "Multiple sources have now confirmed that Elon Musk has shuttered the division responsible for the Supercharger network, resulting in layoffs for the entire team—approximately 500 employees.", it seems clear it was near 100% of the (US?) team (there are still Supercharger things happening, so not 'everyone' everyone. Were it 5,000 initally, the reporting would be different. Also doubtful the worldwide team would be 9x the size of the (US?) given China has a similar number of sites and rest of world is the final thirdish.
willow_hiller
Well-Known Member
Waymo seems to be pretty good at maintaining maps, so we have very little evidence about what their cars would do. This video is just about 2 years old now, so it's more than possible updates to the Waymo driver will make it irrelevant, but we do have one example of what happens when Waymo exits an area that they have mapped:
At timestamp 30:20, the Waymo employee says "I don't know why this construction zone is not taken off the map." Meaning, at that point in time, they had to manually geo-fence off any new construction zones, or else risk the vehicle disengaging like it does in the video.
I think driverless rollout would happen within a year of rollout with safety drivers. So we are talking about less than 2 years from initial rollout to full autonomy rollout. As Ashok said, "It is the beginning of the end!"
And to clarify, since they'll need to report disengagements to the DMV, your expectation is that, as soon as they start rollouts with safety drivers, their official disengagement data will be on par with Waymo?
Because if a robotaxi messes up then it will trigger headlines all over the world. Tesla and/or Elon - media loves to write FUD.
But if waimo or cruise messes up it's only in the local papers. Almost nobody cares.
So Tesla want FSD to be as fool proof as possible before launching.
Seems like they can now handle un-mapped construction zones without having generic issues (I'm sure there are situations when it still fails):
Knightshade
Well-Known Member
Autopark does this today- and is a subfeature of FSD.
Tesla has an interesting choice here.... they could do their pilots outside of CA.... say Nevada, Florida, or any of the other dozen or so states it's legal and there's little to NO reporting required at all. Thus we'd have no insight into disengagements or other metrics other than what Tesla chooses to release to the public.
OR.... they do it in CA-- which has actual regulatory hurdles to get on the road at all-- and requires a decent bit of reporting of data.....
Normally the first one seems to make tons more sense to avoid bad press.... but we also know FSD is most capable/overtrained in CA so it's most likely to get to the best performance results there the fastest too.
willow_hiller
Well-Known Member
At the beginning of the video, she says construction can make maps insufficient or invalid. But at no time during the demonstration videos does she state that these are unmapped construction zones.
Two cherry-picked clips in a promotional video from Waymo themselves are not proof that it can robustly handle construction zones.
High probability he is not an expert, however, he might have stayed at a Holiday Inn last night though.
This is not true. It shutting down was posted in articles all over, proof:
New York:
LA is not Local to San Fran:
Cruise sidelines entire U.S. robotaxi fleet to focus on rebuilding 'public trust'
In the wake of California pulling Cruise's permit to operate robotaxis in the state over safety concerns, Cruise is suspending all U.S. operations.
www.latimes.com
GM’s Cruise Halts All US Robotaxi Service After Suspension for Pedestrian Who Was Dragged
Cruise suspended its driverless robotaxis nationwide two days after losing its self-driving permit in San Francisco for an incident in which a pedestrian was trapped under an autonomous vehicle.
California DMV Shuts Down Cruise Robotaxis Due To Dragging Of Pedestrian
A person was hit by another car and thrown in front of the Cruise. Then it stopped and dragged her while trying to pull over. The DMV says Cruise hid that.
www.forbes.com
Do you think Waymo would go as far as saying their cars can handle construction sites if they couldn't in any situation? I don't think the potential benefits of them saying this outweigh the risks. Secondly, I don't think it's feasible to imagine that they're able to map construction areas in detail. They could be working on one lane in the morning and another lane in the afternoon, completely changing the topography of the site.
No, missed that. I did notice the discourse has been more civil last few days.
Similar threads
- Locked
- Locked
- Locked
