Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autonomous Car Progress

This site may earn commission on affiliate links.
There are already some signs of a second "AI winter". Expectations have been too great, and progress is more difficult that the optimists were expecting. I expect that we're going to see slow evolutionary progress over the next 30 years. And by mid century we may have L5.

I think we will have L5 relatively soon, but it won't be based on strong AI. It will be more like what Waymo does with algorithms doing most of the work.
 
I think we will have L5 relatively soon, but it won't be based on strong AI. It will be more like what Waymo does with algorithms doing most of the work.

I don't think we even have a tiny grip on the legal and regulatory issues. True L5... the unrestricted "take me here"... not a chance. You might have growing pockets of L5 available before that, where the vehicle can operate L5 on a set of restricted access, well surveyed roads that are limited to properly equipped vehicles.

But to operate in a typical urban setting with pedestrians all over the place, construction and who knows what else going on? I don't see it. It's one of those situations where we will get to 99% quickly. But getting to 99.999 - which is required for true L5 - will be a really tough slog.
 
  • Helpful
Reactions: Brando
Last edited:
Yeah, this is really something that I think is woefully behind and not being considered enough (unless it's all happening behind the scenes or isn't easily accessibly via public information...)

AV folks need to get a consortium together and come to an agreement about V2V communication standards, working with various DOTS on signage/road maintenance/striping, etc. before it becomes a free-for-all with competing "standards" (cue the XKCD on standards)

That'll lay the groundwork when there are only AVs cruising around, but then they can work backwards and discuss how they're going to (more) reliably integrate AVs when they're a small, but growing subset of vehicles that have to interact with non-AVs.
 
Yesterday I heard somebody comment upon hearing that the Chinese are tinkering with human DNA:
Now even man is "Made in China".
Existentialist question as far as 'driverless' concerned:
race%2Bsimulator%2B%25283%2529.jpg
 
When George Holtz was young (about just a little more than 2 years ago), he challenged Elon Musk that he could develop a better system than what Tesla was getting from MobilEye! He was talking about autonomous development at that time too.

It's now only 2 years working hard on his own company comma.ai, I think it is interesting that he is no longer promising autonomous vehicle goal anymore!

Shocking!

Is autonomous vehicle system that hard? Or does he just quit dreaming about it because of lacking in funding?

Maybe he just cant do it with a cellphone and camera. Its to hard for that. Until cells phones get more powerful. Hah.
 
Progress in AI can be exponential, and if that is the case (and given that humans are useless at guesstimating exponential growth) it will be "sooner than you think" :)

Or much later than you think, because people suck at guesstimating exponential growth.

If an ancestor of you put 5 dollar at a 3% interest on a bank account at the Bank of New York, back in June 9th 1784 when the bank first opened. And you‘d it find out by accident, you‘d be pretty disappointed that you‘d only have $5050 on that bank account, despite 234 years of exponential growth.
 
Or much later than you think, because people suck at guesstimating exponential growth.

If an ancestor of you put 5 dollar at a 3% interest on a bank account at the Bank of New York, back in June 9th 1784 when the bank first opened. And you‘d it find out by accident, you‘d be pretty disappointed that you‘d only have $5050 on that bank account, despite 234 years of exponential growth.
Except technology evolution started thousands of years ago. In 1820 we invented electricity. In 1890 we invented the first electric car. Airplanes in 1900. Jet engine around 1941. First computers that could fit within reasonable space around 1960. Atari games around 1980. Internet in 1990. Facebook and first public video streaming service YouTube in 2005. Smartphone in 2008. Computer generated low-res pics using neural networks in 2014. Photo-realistic text-2-image computer generated images in 2018.

It's more like we deposited 3$ back in year -2000, but we are now starting to get in the steep end of the exponential curve.

Took 60 years to get from first room-sized computers to Atari. Took 30 years from Atari to Samsung Galaxy Note 9.
 
Except technology evolution started thousands of years ago. In 1820 we invented electricity. In 1890 we invented the first electric car. Airplanes in 1900. Jet engine around 1941. First computers that could fit within reasonable space around 1960. Atari games around 1980. Internet in 1990. Facebook and first public video streaming service YouTube in 2005. Smartphone in 2008. Computer generated low-res pics using neural networks in 2014. Photo-realistic text-2-image computer generated images in 2018.

It's more like we deposited 3$ back in year -2000, but we are now starting to get in the steep end of the exponential curve.

Took 60 years to get from first room-sized computers to Atari. Took 30 years from Atari to Samsung Galaxy Note 9.

There is no steep end... it just depends on where you are right now. Growth 1000 steps ahead is always way more, than 1000 steps before.

If we deposited $3 at 3% in 2000 BC we would have 6.7 x 10^ 51 and 10 years later we would still only have 9 x 10 ^ 51.

And to get back to my original deposit example, $5 for 234 years at 1.05% would be $454,000. That’s why I say we are bad at exponential growth. It can be 5k, or almost 500k, if you expect 5k and get 450k you will be surprised, but also the other way around.
 
There is no steep end... it just depends on where you are right now. Growth 1000 steps ahead is always way more, than 1000 steps before.

If we deposited $3 at 3% in 2000 BC we would have 6.7 x 10^ 51 and 10 years later we would still only have 9 x 10 ^ 51.

And to get back to my original deposit example, $5 for 234 years at 1.05% would be $454,000. That’s why I say we are bad at exponential growth. It can be 5k, or almost 500k, if you expect 5k and get 450k you will be surprised, but also the other way around.
What I think you meant to say was that if in your example you used 5% interest instead of 3%, the value would be $454k instead of 5k. And if it grew at 7% it would be $37M. Compound interest is a beast!