Ah yes, the "more data will magically solve self driving" argument.

More data makes it easier to train and evaluate systems but will not fix an inadequate approach.

Humans can drive with vision alone because humans are way way smarter than glorified regression models. We have a higher order model of what is happening. I am skeptical that you can replicate this with current AI. The approach of using LIDAR to compensate for dumber AI with more and better data seems more sound.

“Seems” That’s all it is to you LIDAR fanboys.

You simply do not know what you are talking about.

also Google being short on data given they've been working on this problem for 12 years is a pretty poor argument. If anything, it should tell us how are the problem of self-driving really is.

He didn't say that. You're putting words in his "mouth."

I mean, the argument literally was Tesla supposedly having more data than Waymo and nothing else. So I'm not sure how they didn't imply what I said.

So on one hand, since humans can drive using only vision - lidar isn't needed.

But on the other hand, even though humans don't need that much training data suddenly vast amounts of data is absolutely required?