James Douma is an AI expert and he believes that Tesla FSD V12.X will achieve better than human driving with more data and compute to refine the system. It will need 1% or less interventions than FSD V11. If interventions have been at 1 every drive then 1% interventions would mean one intervention every 3 months or 1 intervention every year.
Tesla has installed over 15000 Nvidia H100 chips to reach over 50 Exaflops of training compute.
There are other reports of 5-10 times faster progress each month. This would suggest that the 1% intervention level could be reached by June 2024. This would be enough to get FSD out of beta and to have systems that would rapidly reach good performance across the world by adapting to all driving areas. The robotaxi level might take another 6-24 months, if James Douma and the reported faster progress is correct.
If this estimate of 2023 Nvidia H100 chip purchases is correct then Tesla has at least 15,000 Nvidia H100s. H100s are 12X Nvidia A100.
This would means that Tesla would have installed those chips by now and be at about 65-70 Exaflops of training compute. If Tesla was able to buy another 10,000 Nvidia H100 chips then they would reach the 100 Exaflop target from last year.
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.
Better than human is a very low bar, especially in Florida.
It has to be far better than just “better” to be accepted.
It wouldn’t hurt to have them smart enough to know when they need to call police, fire, tow truck or ambulance, even if they aren’t to blame.
totally agree here. its sort of like the airplane vs car safety rating. no one would fly if airplanes were only as safe as cars or twice as safe. it will have to be orders and orders of magnitude safer to have adoption.
FSD level 5 without any intervention, and an accident every trillion miles, won’t happen this decade. It will happen when we reach ASI, in the 2040s as Kurzweil stated. People with unrealistic timelines are just delusional.
Let me know when it reaches 0 intervention. Then and only then will I be interested.
What is meant by an intervention? If that means person needs to take the wheel immediately to prevent a crash, it could be more dangerous as the driver gets too comfortable with it over the months of nothing. Yes, you have to keep hands on the wheel and be alert, but I see it causing issues.
“This will be it”
Yawn, we’re hearing this each year, since like mid 2010’s?
People are tired
Sadly Pat is correct. When Tesla decided to discuss FSD was far too early for mainstream people that don’t want to be involved with alpha & beta type testing with a technology in its early phases. Most companies hide that development until it is nearly ready for acceptance testing from most of the consumer base. How many really want to donate their time to a tech project they don’t get paid for? Well, there is some people that really want to be part of that process which is why you hear about this constantly, these people demanded that FSD be opened up to the public though it was Elon’s mistake in the first place to allow for the unfinished products purchase. Sure it gets Tesla its driving data but exposes people that don’t have the mindset to go thru a very long development process. My suggestion Pat, detach from the Driving AI space until the next U.S. President comes and goes. Then check back as the product should be ready for you to START to follow.
No need to pick up figures, make gods out of them, dream their dreams look for experts that support Their actions. This is China, not the next big future that we are waiting for.
How can it be better if it still needs interventions.
Fatality!
By the standards of FSD development – most human drivers would get pretty regular “interventions” for breaking some of the formal or informal rules of the supervising drivers. Sometimes it’s about not being aggressive enough on a left turn, or too aggressive about lane changes. It doesn’t imply there would have been an accident or that FSD wasn’t already driving at the level of many humans. The standard discussed is every instance of FSD being better and safer than pretty much any human driver – because even the best human drivers would have bad days and be more inconsistent or inattentive sometimes. The point is, it doesn’t matter whether critics think it’s taking too long or are tired of the projections – there is no more competition and when FSD gets to unambiguously better than human it will eventually be accepted by regulators and have the revolutionary effects predicted.
While I actually think V12 is finally going to deliver actual FSD, most do not. Elon has spent the past decade saying FSD is coming soon, and now people are treating him like the boy who cried wolf.
That’s OK. Critics say the same things about Starship or his other projects. FUD and politics have short term impacts but SpaceX and Tesla have the necessary resources and are just continuing to execute. The facts will speak for themselves and revolutionary changes will occur whether critics are “tired” of waiting for them or not.