[hellllllllooo]

There's a lot more problems involved than just data collection. The algorithms do not exist to safely and reliably solve these problems even with all the data in the world. Simply put Tesla is declaring it can do something with a restricted set of sensors and compute that the current state of the art in research doesn't support. Even Waymo who have no real production restrictions and can spend >$100k in equipment per car and have a lot more compute is not claiming anything like this will be feasible in the next year. He is claiming cheaper, faster and better than anyone else.

The problems just get harder and harder as you move from a nice controlled environment like a freeway and even there Tesla has shown there are many issues with the system where a driver needs to pay full attention for it to be safe.

From first hand experience with these problems and their complexity I fully expect a slightly better version of autopilot to be delivered and Musk to declare success despite it not being anywhere close to an autonomous fleet of Tesla robo-taxis and justify the difference with the caveats he quietly snuck in that he knew wouldn't make headlines.

That might be enough for his customers and investors so from a business sense it might pay off.

[mrep]

Can you give some examples of those problems because you are coming off as very hand-wavy about it?

[hellllllllooo]

Simple case that Waymo etc still struggle with is making a left turn through an intersection or an unprotected left. Even something as common and predictable as crossing oncoming traffic that humans do every day is still causes disengaments for these cars. Cruise has a team just focused on issues at intersections. Throw in some pedestrians and this is very hard. Humans can intuit a lot from a pedestrians behavior that these cars just cannot handle. Is someone standing by the corner or about to walk into the street? People do a lot of weird unpredictable things in downtown areas. Tesla has done ok with the easiest part of driving, freeways, with the support of an attentive human but three are still many reported problems. Moving on to complex environments with multiple moving actors at such high reliability that it is a hands free production-grade system in 1-2 years is not likley, but good luck to them.

[mrep]

Isn't that the exact type problem that massively benefits from having tons of real world data because they will have examples to train their neural nets for practically every possibility?

[hellllllllooo]

Not really. Neural nets are good at detection and classification problems for objects in the environment but they are not a panacea. Assuming a neural net is good enough to detect everything a car will see (for pedestrians this isn't the case) path planning while predicting behavior of other actors is still a very difficult unsolved problem.