Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Expands Fleet of Self-Driving Cars
- Thread starter MacRumors
- Start date
- Sort by reaction score
I was waiting for an Apple Car, but after doing the math on how much my gas car was costing me, I bought a tesla last summer. Honestly the EV (and especially anti-tesla) distortion field shocked me.
It is so much more 'there' than any car I've ever driven (and I've driven some nice ones).
It's like the mid-2000s jump from a flip-phone to a smart-phone. People wondered why they'd ever want to spend $500 on a smart-phone instead of $200 on a flip-phone. Then once they bought the smartphone, everybody wondered why they were paying $200 for a dumb-phone!
It's no comparison. Night and day improvement across the board, and onto boards you never even looked at for a car.
On 99% of the things you used to measure how good a car is, my electric is WAY better. But MAN it does so much more. You get back into a gas car and it's very obviously last-century thinking and last-century tech. With obvious features just missing.
If you haven't done the TOTAL cost of ownership between an electric and your gas car (gas, maintenance, depreciation, etc), then do it. It'll pay for itself. Even at today's gas prices.
And you really don't have to buy a tesla. They have the tech and interface you'd expect from a tech company (not the bolt-on poor UIs that the legacy automakers have), and they have a MUCH more seamless roadtrip experience (which is only a TINY bit of my use for a car anyway).
I can't wait to see what Apple comes out with, because I imagine I'll heavily consider switching. Until then though my advice to everybody is just do the math on an electric. They're ready. And SO much better.
--
Since I bought, I've been following the tech behind the self-driving (which is really unrelated to how good an electric is), and Tesla's just-releasing private-beta of FSD is truly is 8-10 years ahead of anybody, even Apple.
It all comes down to neural net training.
Between Tesla's current supercomputer (5th largest in the world), and their "Dojo" supercomputer coming online this month, which will be MUCH faster at NN training, Tesla has a 1-2y head-start on the infrastructure of self-driving training.
But what Tesla has that nobody is CLOSE to is the number of tagged road events harvested for training. They have the fleet of cars (even the ones without FSD) collecting a bajillion miles on every different kind of road.
But the uninteresting miles really don't help.
What they've also built is a Data Engine which uses a separate NN which runs in shadow-mode alongside the existing Autopilot (on all cars) and looks for specifically targeted events/clips which they want more training on. (They also have ANOTHER neural-net stack which auto-tags the clips, saving a ridiculous amount of human time in tagging, having them just verifying the auto-tags.)
The head of tesla's AI team mentioned they needed more work on identifying cars with canoes on the top, or identifying "Stop, except when turning right" signs. So they built a special targeting AI just to harvest any of those situations any Tesla comes across. They have 100+ of these special targets they're harvesting right now and can change them at any time.
This Data Engine, which absolutely NEEDS a fleet of cars encountering the zillion different weird-ass edge-cases is absolutely essential.
Without that NOBODY can get close to the "long tail of nines" that self-driving is required to solve.
It is so much more 'there' than any car I've ever driven (and I've driven some nice ones).
It's like the mid-2000s jump from a flip-phone to a smart-phone. People wondered why they'd ever want to spend $500 on a smart-phone instead of $200 on a flip-phone. Then once they bought the smartphone, everybody wondered why they were paying $200 for a dumb-phone!
It's no comparison. Night and day improvement across the board, and onto boards you never even looked at for a car.
On 99% of the things you used to measure how good a car is, my electric is WAY better. But MAN it does so much more. You get back into a gas car and it's very obviously last-century thinking and last-century tech. With obvious features just missing.
If you haven't done the TOTAL cost of ownership between an electric and your gas car (gas, maintenance, depreciation, etc), then do it. It'll pay for itself. Even at today's gas prices.
And you really don't have to buy a tesla. They have the tech and interface you'd expect from a tech company (not the bolt-on poor UIs that the legacy automakers have), and they have a MUCH more seamless roadtrip experience (which is only a TINY bit of my use for a car anyway).
I can't wait to see what Apple comes out with, because I imagine I'll heavily consider switching. Until then though my advice to everybody is just do the math on an electric. They're ready. And SO much better.
--
Since I bought, I've been following the tech behind the self-driving (which is really unrelated to how good an electric is), and Tesla's just-releasing private-beta of FSD is truly is 8-10 years ahead of anybody, even Apple.
It all comes down to neural net training.
Between Tesla's current supercomputer (5th largest in the world), and their "Dojo" supercomputer coming online this month, which will be MUCH faster at NN training, Tesla has a 1-2y head-start on the infrastructure of self-driving training.
But what Tesla has that nobody is CLOSE to is the number of tagged road events harvested for training. They have the fleet of cars (even the ones without FSD) collecting a bajillion miles on every different kind of road.
But the uninteresting miles really don't help.
What they've also built is a Data Engine which uses a separate NN which runs in shadow-mode alongside the existing Autopilot (on all cars) and looks for specifically targeted events/clips which they want more training on. (They also have ANOTHER neural-net stack which auto-tags the clips, saving a ridiculous amount of human time in tagging, having them just verifying the auto-tags.)
The head of tesla's AI team mentioned they needed more work on identifying cars with canoes on the top, or identifying "Stop, except when turning right" signs. So they built a special targeting AI just to harvest any of those situations any Tesla comes across. They have 100+ of these special targets they're harvesting right now and can change them at any time.
This Data Engine, which absolutely NEEDS a fleet of cars encountering the zillion different weird-ass edge-cases is absolutely essential.
Without that NOBODY can get close to the "long tail of nines" that self-driving is required to solve.
Last edited:
Great post. I’ve only test driven a Tesla (it will be my next car), but my experience was the same: ICE cars are as done as flip phones and blackberries were when the iPhone came along. Oh, it’ll take a while - just like it took a while to pry the flip phones and blackberries out of a lot of hands - but the conclusion is clear: the better technology has arrived, and it’s just a matter of who realizes it sooner and who realizes it later.
Last time I drove that area was some time ago so it would've been before they went vision only.
Tesla will have the advantage, yes, but if Elon runs it into the ground is it really going to matter who gets the scraps of their research? I for one am excited for a Mach-E, but it’ll be years before I can even come close to trying to afford one, by which time Tesla may have removed Musk and his incendiary BS and put someone a little more stable into the reins to actually expand on these baby steps. I can’t imagine shareholders and particularly happy with what he’s been doing lately, but at least he isn’t Bezos.
As a shareholder, I see him bringing both good and bad. A lot of weird and unnecessary crap but also a lot of innovation and free publicity. I don’t see him running the company into the ground. Maybe five years ago, but it’s too big for one person to sink at this point (maybe he could sink it with bad business decisions, but that’s not his problem - his problem is collateral nonsense. The actual business decisions are pretty sound.)
Have you TRULY done the math on an electric vs gas? Price plus, gas/electricity, maintenance, depreciation, etc? There are ways you can get a pretty low monthly payment.
In the US even with the crazy low subsidized gas prices, most medium range (300km or more) non-teslas can come in less than what you’re paying for your gas car today.
In Canada, my Tesla Y costs about the same over 10y as my 10yo Subaru Outback that was supposedly long ‘paid off’.
Also, if you’re going Mach E, wait a year or two. They’ll shake out the kinks. This first one has some… issues… compared to the other non-Teslas out there.
Test it in Ukraine. There are roads there where I have no idea what clues people about the change in rules between one section and another — and most difficult of all, you'll often be on a road marked as three lanes but people are driving 4 abreast. I remember sitting in traffic that day, thinking, "There's never going to be a driverless car on this road."
I drive a hybrid. My house isn’t built for having a charging station, and I mean I make less than half of what you’d need for the life of the loan. I cannot afford an electric vehicle with the way they are priced now, nor would I be able to get a loan to cover an insane income to debt ratio. I want an electric car a lot, but the financials don’t work out for an almost 24 year old with no monetary portfolio to speak of. I’d pull the trigger if I could. I love not having to drive five miles and refill despite driving an SUV. Go a week+ without filling up in a city environment. It’s amazing, and I want to continue it, but the Bronco Sport is my safest option, as well. If they came out with an electric option, I’d take it in a heartbeat. But my car will last maybe five more years before I have to retire her as being a cost of repair guzzler. I can’t wait to be able to save enough to get a dream electric car, though. And a charging station that can support the vehicle.
And yeah, I work at a Ford dealership. I’m fairly well aware of the issues of the Mach-E, but we’ve only ever gotten four in our inventory since they’ve come out, so I’ll just have to wait until they better. I’m more than happy enough to spend my money on long-lasting products, though.
Oh Apple - Apple doesn't have what it takes.
They don't have Elon.
They gave some middle managers who think they're better than Elon reporting to Tim Cook to efficiently get the best prices from the supply chain.
With Steve Jobs this was a different story but Apple isn't employing anyone interesting anymore. All interesting people have left. Now it's run by mediocrity and fueled by insane amounts of cash on hand.
I think this entire endeavor - and also the ill-advised venture into health, as if I want Apple to decide how healthy I am - is based on having too much chash on hand, so they just have to find places to spend money.
Cars are a no brainer to some extent but Apple doesn't have the geniuses required to build them. No Elons working there, no Ives even... it's all boring middle managers.
Tesla open sourced all their patents - Apple would be wise to clone all Tesla tech.
But instead they're the only ones driving around with silly LIDAR.
You may think it's a small detail but it shows how inadequate the tech team is.
Elon said this - all roads are made to be navigated with human eyes - all street signs all markings, all habits, everything on the road is designed so humans can drive on it, humans don't have LIDAR and adding additional sensors is a distraction - not just unnecessary but also distracting from the good solution, which is to use cameras and AI, and perfect that.
So LIDAR is a testament to doing the wrong thing from day 1.
Eventually they'll buy up one of the optical imaging startups and call it a day but not after spending endless resources on things that don't work. That's poor tech decisions that middle managers make.
They don't have Elon.
They gave some middle managers who think they're better than Elon reporting to Tim Cook to efficiently get the best prices from the supply chain.
With Steve Jobs this was a different story but Apple isn't employing anyone interesting anymore. All interesting people have left. Now it's run by mediocrity and fueled by insane amounts of cash on hand.
I think this entire endeavor - and also the ill-advised venture into health, as if I want Apple to decide how healthy I am - is based on having too much chash on hand, so they just have to find places to spend money.
Cars are a no brainer to some extent but Apple doesn't have the geniuses required to build them. No Elons working there, no Ives even... it's all boring middle managers.
Tesla open sourced all their patents - Apple would be wise to clone all Tesla tech.
But instead they're the only ones driving around with silly LIDAR.
You may think it's a small detail but it shows how inadequate the tech team is.
Elon said this - all roads are made to be navigated with human eyes - all street signs all markings, all habits, everything on the road is designed so humans can drive on it, humans don't have LIDAR and adding additional sensors is a distraction - not just unnecessary but also distracting from the good solution, which is to use cameras and AI, and perfect that.
So LIDAR is a testament to doing the wrong thing from day 1.
Eventually they'll buy up one of the optical imaging startups and call it a day but not after spending endless resources on things that don't work. That's poor tech decisions that middle managers make.
Exactly what I just said!!
Elon destroyed the need for LIDAR in a 2 minute speech. All they had to do is _listen_. That's how easy it was. But no ... 🤦♂️
You need to make more money.I drive a hybrid. My house isn’t built for having a charging station, and I mean I make less than half of what you’d need for the life of the loan. I cannot afford an electric vehicle with the way they are priced now, nor would I be able to get a loan to cover an insane income to debt ratio. I want an electric car a lot, but the financials don’t work out for an almost 24 year old with no monetary portfolio to speak of. I’d pull the trigger if I could. I love not having to drive five miles and refill despite driving an SUV. Go a week+ without filling up in a city environment. It’s amazing, and I want to continue it, but the Bronco Sport is my safest option, as well. If they came out with an electric option, I’d take it in a heartbeat. But my car will last maybe five more years before I have to retire her as being a cost of repair guzzler. I can’t wait to be able to save enough to get a dream electric car, though. And a charging station that can support the vehicle.
You can work on the ford dealership as long as you have to (and pick up all the interesting skills you can learn there, no sarcasm, there's a lot open ears and eyes can pick up, just figure out who the best guy is that works there and observe him, all you need to do)
Then figure out your life and realize you can make infinite money, but also do something you love.
Work towards freeing yourself from salary slavery, it never gets better.
Also buy bitcoin obviously...
Apple needs to buy him up ASAP, even if it costs billions.
Been following his progress, this is the kind of spirit that can take on Elon.
In reality, otherwise, Apple will waste billions on a sub par solution.
Their REAL advantage is they are attracting the best talent in the world with the opportunity to do the coolest, most advanced work that will also change the world.
The upcoming Autonomy Day is going to be crazy exciting for us nerds, but Elon said its goal is 💯 focused on recruitment.
First off, GM's super cruise is the equivalent of Autopilot not FSD. As for whether it's as advanced as Tesla's Autopilot is debatable, GM's super cruise only works on highways that have been mapped unlike Tesla's Autopilot that works on every highway.
Some people argue Apple should host apps in the most profitable-per-square-feet app shopping mall on Earth for free (or allow anyone to just set up a stall in the parking lot), forgoing 13 years of build up to this in terms of research, development and delivery (in terms of hw, sw, security, speed, customer retention, etc.). Go figure.
Ah yes. Because when it’s Apple and their App Store and APIs, everything should remain Apple’s because they “worked so hard for it and need to pay engineers”. But when it’s Tesla, all of those concerns go away! Should Tesla be forced to give away everything for free and at the same time Apple allowed to keep everything for themselves to profit off of?
Thank you for the advice, but Bitcoin is far too voliatile an investment as it stands and crypto as a whole is probably going to be taxed to high heaven once the IRS figures out what it classifies as, at least in the US. I love administrative stuff: being in the background keeping things from going kaput. Unfortunately there’s not a lot of $$ in it at my level. Ah well, just keep learning skills.
(That was the joke. In light of the “technoking”’s impromptu remarks about the Epic v. Apple trial, tweeted from the porcelain throne.)
The entire "level" system is stupid and unhelpful for measuring progress.
The thing that matters is what percentage of miles driven require human interaction. This is somewhere around 10% for what Tesla has released to all customers so far and well below 0.1% for their FSD beta. It's around 1% for Waymo in their beta and well above 20% for everyone else.
OK, my car is 15 years old, paid off, and still gets 30 MPH. With the way the world has changed, I've cut my annual driving by over half. My insurance rates are one-fifth what they would be with a Tesla, as is my registration.
So please...tell me how much buying a $100,000 Tesla is going to save me. Tell me about all the wonderful features which make increasing my carbon footprint with that initial purchase worth it.