There are still flaws with Tesla FSD 12.3.X but if Tesla FSD 12.4 and FSD 12.5 fix lane selection problems, speed issues, all construction zones, flashing red lights and reverse then it could a over a month to find flaws. The reported data is from the TeslaFSDtracker.com.
There is crowdsourced data that 12.3.6 can go 330 miles without a disengagement in City Driving. This is over two times better than the 119-140 miles for the best version 11 Tesla FSD.
Tesla testers are claiming large improvements for Tesla FSD 12.4 and FSD 12.5. If Tesla FSD 12.4 takes 1000 miles of driving before disengagements then it could be a month of driving for one person to find a problem. However, a hundred thousand people would still see a thousand problems per day.
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.
The law of diminishing marginal returns has kicked in hard on incremental improvements to FSD. Small improvements will require increasingly significant time & cost. We might be at 99% but getting to 99.999% could take 1000x effort.
There is no “definition of done” or certification. What is the level of risk tolerance for error? How does this bar move up or down in different cultures? We can tolerate human error and have a legal framework for civil and criminal liability. We don’t have the equivalent for machines.
This is totally contrary to my experience. I think this model is overfitted for California. Because I get only a few miles between disengagements. It makes a lot of mistakes. I’m off it now because free trial ended, so I can’t speak to the latest couple of point releases. But I can’t imagine it’s radically better
In CA and have been an FSD beta tester for 1.5 years. FSD 12.3 is now better than FSD 11, but I still disengage more than once per mile on city streets during my regular commute. The longest between disengagements with half streets and half highways with a little traffic is about 15 miles.
Poor lane selection
Stops to far from the limit line
Awkwardness at stop signs
Unable to safely make unprotected turns timely
Does not do well in dense heavy traffic. Fails to make the best decisions to get through heavy traffic. It’d easily add 10 minutes to a 15 mile commute if I allowed it to pick lanes.
Just still has long ways to go before you call it level 3
Negative Tesla article over at the BBC today:
Have the wheels come off for Tesla?
https://www.bbc.co.uk/news/business-68947020
My most often encountered issue with FSD is how it handles stop signs. It does not require a disengagement for me to handle but it is an annoying recurring issue. The car always stops before the stop sign, often before it can clearly see the cross traffic. It then creeps forward before it takes the turn. This is annoying to me and drivers behind me. I normally push the car to where I can see the traffic eliminating the first stop.
We don’t know what changes are included in 12.4. If more features are added – for instance being able to reverse- the we may see zero improvement on core metrics.