This is akin to removing the requirement for unicorns to wear horseshoes on the public road.
[1] https://abcnews.go.com/Technology/faa-drone-regulations-blow...
As far as the safety, these things can be indemnified and that will pave the way. Don't see any problem insurers can price accidents and injuries from autonomous cars. Also certainly plausible they will be able to be made safer than human driven.
They're probably not possible with the road system we have but we redesigned our road system to accommodate cars(arguably with terrible results for non-cars) and we could probably redesign it again for autonomous cars. For example by requiring pedestrians and cyclists to wear radio transponders.
Which parts of driving would require AGI?
Being able to read the intent of other cars and preemptively act if necessary. For instance, being able to predict that the harried mom on her phone toting 3 screaming kids in her beat up minivan is likely to change lanes without looking (I passed her preemptively), or that the car with out-of-state plates is likely to miss a confusingly signed exit and try and cut over at the last minute without looking (I slowed down preemptively, to let the car abruptly veer across three lanes of traffic ahead of me).
Both of these examples require an understanding of the world that goes far beyond mere driving skills.
An average 20 min ride increased by 10% would be 22 mins. I’d gladly trade an extra few mins in the car to 1. Be safer. 2. Get back 20 mins to check the news or play chess or worst case work.
Self driving cars aren’t just going to be a reality, in PHX they kinda are a reality. It doesn’t seem far fetched to think 5 years (high estimate - 10 years) from now Waymo will be scaled to the top 25 cities in the US.
I have the opinion we don't need AI to be perfect, just on par or better than humans from an aggregate perspective.
Only if there is a mathematical proof that eventually it will happen, someone might proof the reverse that current type of tech will never be good enough for general purpose driving and will work only in some specific conditions.
Absent souls, human drivers are existence proof that minds made of matter can drive cars.
That doesn’t stop being true even if the current preferred architecture is all wrong, as we can and do change those with ease.
So I was thinking "eventually" as somewhat in next 100 years, otherwise even if you need a soul you could say shit like "eventually we will evolve to the next plane and become gods and we could create brains".
Could it be impossible that we hit a tech limit for AI? I mean there is no sure what that we will invent something better then digital, silicon based computers. I am not that pessimist though, but is possible we get smart and realize that we could just put tracks and have self driving little trains on them to travel around: 100% safe (we could have actual scientist/engineers code the software for them) and much more green.
IMO we will probably hit the single-atom limit in consumer devices this decade, though researchers are already suggesting multiple q-bits in a single atom so perhaps we won’t even stop when we get 1Å process chips.
The main limiting factor right now is that we don’t really understand the structure of our intelligence. If you will permit an analogy: perceptrons and convolution kernels are like bricks, while current brain scans (MRI, EEG, etc.) are like nighttime satellite photos of city lights — we’ve got a gap in the middle that would tell us we need steel reinforced concrete to make skyscrapers and why want to build them in the first place.
This is a rather intuitive mathematical proof that neural nets can emulate all functions in many dimensions. Knowing integrals (Calc 102) would help.
With this information it’s either that self-driving is eventually possible or human brains are not mathematically representable (i.e. something akin to a soul).
Meanwhile, the data collection/storage is getting better (all Teslas are continually testing FSD and reporting failure cases, synthetic generation of “good and unique” perfectly labeled data[0] is only a year or two “old”). Sensors/cameras are improving (and have a lot more room to improve). The compute hardware is getting better and cheaper (analog computers[1], TSMC 2nm and beyond, Apple/Google/Teslas neural cores, etc).
Additionally, given Waymo is already in production without humans in PHX it’s arguable that we already have “autonomous vehicles”. We now just need them to improve slightly and for Waymo or similar to test in the top 25 cities in the US. Which is mostly a matter of capital and time not technology limits. Maybe weather will be an issue for a few more years/decades - but even if Waymo only operated in Summer, I’d be happy to reduce my Lyft costs - and with enough data and maybe better sensors the neural nets will be able to handle weather too. Especially if the car just goes slow.
[0] https://youtu.be/j0z4FweCy4M
[1] https://youtu.be/GVsUOuSjvcg
[2] https://www.cnbc.com/2022/01/08/heres-what-it-was-like-to-ri...
Yes, but the problem is we don't know what function we want to approximate. So it is like interpolation you give it some points and you get an approximation that is good enough around those points but probably very wrong if you are far away of the points.
We don't even know how many layers and neurons are needed so it seems to be a brute force approach, throw more data and more hardware and when stuff goes wrong maybe throw more data. But one day a Tesla might go on a different environment and maybe people have different hats and goes to shit because those hats were not in the training data, we don't have like in other sciences bounded errors.
I think of this as an unproven assumption, not a given.
Well, I mean, duh. You’d be a fool to bet against seeing a unicorn on an “eventually” timeline. Try to collect and I say it’s not eventually yet.
Unicorns can't, won't ever exist.
The idea of the existence of self-driving cars is definitely on the cards. This is en route.
Exactly.
Legacy regulations governing the occupant compartment are not yet a practical impediment for self driving cars because they haven't got that far yet.
Realistically what the removal of this regulation means is that some hypothetical future company with an autonomous yard spotter can drive the thing 500yd down the public way from warehouse B to warehouse C without the legal department getting antsy.