Hacker News Reader

Comment by TheAlchemist | Hacker News Reader

Similar to when Andrei Karpathy left Tesla. Tesla was on the verge of 'solving FSD' and unlocking trillions of $ of revenue (and mind you, this was already 3 years after the CEO said that they will have 1 million robotaxis on the road by the year's end).

Guess what ? Tesla is still on the verge of 'solving FSD'. And most probably it will be in the same place for the next 10 years.

The writing is on the wall for OpenAI.

View on HN

The original saying of "fake it till you make it" has been changed to "say it till you make it" :-)

I follow the latest updates to FSD and it's clear to me that they are getting closer to robotaxis really fast.

That's what was said years ago.

Yeah I follow it too. There is progress for sure, one have to wonder if the CEO was very consciously lying 5, 8 years ago when he said they are less than 1 year away from robotaxis, given how shitty the system was.

They are on a path of linear improvement. They would need to go on a path of exponential improvement to have any hope of a working robotaxi in the next 2 years.

That's not happening at all.

FSD isn't getting "solved" without outlawing human drivers, period. Otherwise you are trying to solve a non-deterministic system with deterministic software under a 0% error tolerance rate. Even without human drivers you still have to deal with all the non-vehicle entities that pop onto the road from time to time. Jaywalkers alone is almost as complex to deal with as human drivers.

LOL this is BS. We have plenty of deterministic software being used to solve non-deterministic systems already. I agree that 0% error rate will require the removal of all human drivers from the system, but 0.0001% error rate will be seen as accepted risk.

I hear they will have FSD by the end of the year.

which year exactly is TBA

Reply on news.ycombinator.com