Crowdsourced results show that Tesla FSD 12.3.6 should have the best disengagement results for any version prior version of FSD. FSD 12.3 had slightly better city driving results but usually the reported disengagement results improve from the first days. It would be a very solid result for 12.3.6 to get over 300 miles for city and highway driving.
Tesla should release FSD 12.4 by the end of next week. If FSD 12.4 gives speed control, intersection creep and improves lane choices, then half of the disengagements would go away. If FSD 12.4 gives integrated reverse then the car would be able to get out of parking spaces for a far better end to end experience. It would go from parked, drive and parked again for almost no need for driver interventions. 600 miles between disengagements would be only one or two disengagements per month.
The 12.3.X systems are at one disengagement every 3 or 4 drives and usually only lane choice and annoyance with speed disengagements.
Elon says FSD 12.5 has been tested internally for months and is next level. Next level would be feature complete (reverse, flashing lights, emergency vehicle response, avoiding pot holes, not driving in blind spots) and another halving of the disengagements. It could go from averaging one critical disengagement per month to one or two per year and one regular disengagement per month.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
At a certain point `disengagements` become a weak or even negative metric
We know for instance that a good modern chess computer can outplay any human BUT how many players sitting at a board would overrule (disengage) the computer if given the chance, why, because we are not good enough to understand the reason, I am a chess player in the top 1% of hobby players and I am often bemused initially by a computer move thinking it bad, I only trust the computer as I objectively know its better than me, the same will apply to a driving computer.
Effectively we will be Dunning- Kruger you need yourself to be as good as the computer to understand the why, and most of us will not be that good, I am certainly a safe driver a former trained police response driver, who drove emergency vehicles in London but I only consider myself slightly above average.
Level 5 is zero disengagement in your entire life. We won’t get there in a few months. Get real.
He’s not saying that. Level 5 is years away. But FSD is impressive and shows the future where driving is headed.
Until the insurance rates come down on FSD Teslas, below other cars, even non FSD Teslas, it’s a meaningless statistic. Insurers don’t play games when there’s real money at stake and they probably know more about the real risks than the NHTSA.
A no-fault insurance fund covering all FSD Teslas, regardless of whether a driver is present, will be the only way robotaxis can ever work.
It’s going to get really difficult to keep users vigilant when disengagement drop to once every few weeks or months. Expect FSD accident rates to skyrocket as FSD is improve into the infrequent disengagements range.
What makes FSD dangerous is that there is not a hand off like in a level 3 system. rather there is a disengagement which means essentially a failure without any warning. What this means is the disengagement rate needs to be such that it’s way less than a once in a lifetime event. Think of it like a front tire falling off your car. It can be a theoretical possibility but so long as it doesn’t really happen, it’s OK. Whereas if it happened say once in 10,000 miles on average, the car would be unusable.
Gary, please explain to me the difference of driving on the interstate with a Model 3 with FSD, and a Mercedes EQS with Drive Pilot.
Is Drive PIlot the system that only works on a very few freeways in traffic jam when a car is ahead of you, up to 45 mph, when sun is shining and you must be able to takeover within 10 secs when the system alerts you? That is what every assistance system could do and even my 3 yo Kia could handle easily, seen from specs. FSD is way ahead.
For one, Drive Pilot is limited to specific freeways in CA and NV and under 40mph among other conditions like perfect weather during the day. Basically, it would only work on an interstate in stop and go heavy traffic. Apples and oranges compared to FSD.
I’m not Gary, but I’ll take a crack at this. Drive Pilot operates only at speeds below 40 MPH in traffic jams on precision-mapped limited-access divided highways in Nevada and California, during daytime on dry roads in clear weather. Meanwhile, FSD (Supervised), operates on every road, street, highway or freeway in the US and Canada, paved or not, mapped or not, at any time of day or night, in most any weather, at any speed up to 85 MPH. However, FSD does require a hand resting on the wheel or other periodic action to show engagement, and requires the driver’s gaze to be outside; while Drive Pilot does not require a hand on the wheel, and allows the driver to look at the display or a phone, as long as their gaze is forward.
I think you don’t understand what disengagement means in this context. It’s is drivers doing the disengagement, not the vehicle. I usually disengage because it is too slow (cautious) navigating a tricky intersection, as to not get horns honked at me.