I noticed too much use of the word "safety", like the LLM was told to emphasise it, so I did a little test: randomly scroll and move the mouse without looking, is there "safety" in there? I did it for 4 times and every time I found it. Ctrl+F -> 136 results.
I personally found this NVIDIA move very interesting. Automakers generally do not want to become frontier AI infrastructure companies and they love technology standardization.
The real technical challenge is rappresented by edge cases: a software that is excellent 99.9% of the time can still be unacceptable if the remaining 0.1% contains rare but catastrophic scenarios. And that's why we still don't see many self-driving vehicles on the roads today.
However, NVIDIA has a credible shot because it controls much of the loop - hardware, training infrastructure and simulation environment. If it works they will impose a huge vendor lock-in, difficult to replicate for other competitors.
I'm very exciting for Nvidia to meaningfully enter this space. I know they've been working on autonomous vehicles for a while now, but it seems like they are approaching a real product. Hopefully, they produce something that can be used on consumer vehicles. We really need good competition in this space. The US market is limited to Tesla FSD and no other manufacturer is even close. I'm not confident individual manufacturers could meaningfully develop their own solutions. A strong third-party option is a great direction for the industry.
18,600 engineering years sounds impressive to someone because it is the bulk of a career for 1,000 engineers. But it is less than two years for 10,000 engineers. The depth of understanding really hinges on which version is closer to reality.
Meta Horizons World probably puts up similar numbers if you sum up the hardware/software tech stack to get this number.
Probably the only way to catch up to Waymo's technical lead is for every other player to collaborate. The world dearly needs another self-driving car option.
I don't think it's LOC for productivity, it's LOC which have passed safety scrutiny. We're talking about the kind of code which would pass muster on something like NASA's safety assessments, probably. Takeaway: it's a huge codebase which has been audited for safety.
If it means what I think it means, you take every engineer working on it (and maybe the years of research involved) and add it all up. Say you have a room with about 10 engineers with 10 years of experience per developer, you can claim there's 100 years of developer experience between all of them (maybe the overlaps not unique enough and its more like 30 to 50 years? but in this case I think they're rounding up, and I assume it means thousands of engineers involved in the project) that's how I took it.
My first interview in tech I was asked what the heck I was even doing with the D programming language, followed by the remark that in the next room (where all the devs were) there was at least 100 years of experience between everybody there, and not a single one knew what D was, my manager clearly did, which cracked me up.
However, it's one of those metrics that tends to be kind of meaningless. Vehicle safety team uses GPUs, so lets bill all GPU driver teams to the metric... that sort of thing
You missed the "21+ billion safety transistors safety assessed" gem.
I don't even know what that was supposed to mean. Hopefully all the safety transistors in the safety graphics card of my safety-PC were safety-assessed, too /s
> Are we seriously using LoC as a measure of productivity again?
Yes, sadly. Because its how everyone justifies LLMs. "Look at how much code it writes!" is the only measure they can come up with to sell its usefulness, completely forgetting that it'll be more useful if we started talking about how much code they remove.
The real technical challenge is rappresented by edge cases: a software that is excellent 99.9% of the time can still be unacceptable if the remaining 0.1% contains rare but catastrophic scenarios. And that's why we still don't see many self-driving vehicles on the roads today.
However, NVIDIA has a credible shot because it controls much of the loop - hardware, training infrastructure and simulation environment. If it works they will impose a huge vendor lock-in, difficult to replicate for other competitors.
Meta Horizons World probably puts up similar numbers if you sum up the hardware/software tech stack to get this number.
> 18,600+ Engineering years invested in vehicle safety to date
What does this even mean?
> 7,000,000 Lines of safety-assessed code
Are we seriously using LoC as a measure of productivity again?
Not to mention the em-dashes
If it means what I think it means, you take every engineer working on it (and maybe the years of research involved) and add it all up. Say you have a room with about 10 engineers with 10 years of experience per developer, you can claim there's 100 years of developer experience between all of them (maybe the overlaps not unique enough and its more like 30 to 50 years? but in this case I think they're rounding up, and I assume it means thousands of engineers involved in the project) that's how I took it.
My first interview in tech I was asked what the heck I was even doing with the D programming language, followed by the remark that in the next room (where all the devs were) there was at least 100 years of experience between everybody there, and not a single one knew what D was, my manager clearly did, which cracked me up.
It means over 18,600 engineering hours have been spent working on vehicle safety. This is a pretty common metric.
I don't even know what that was supposed to mean. Hopefully all the safety transistors in the safety graphics card of my safety-PC were safety-assessed, too /s
Hot take here, but personally I feel they should safety assess the danger transistors, reducing the need for so many safety transistors.
Yes, sadly. Because its how everyone justifies LLMs. "Look at how much code it writes!" is the only measure they can come up with to sell its usefulness, completely forgetting that it'll be more useful if we started talking about how much code they remove.