Yes. But what is difficult for us humans to deal with is the fact that when Waymo does fail, it will be in a way that a human in the same situation would not have failed. Your father will be able to say, "I never would have made that mistake!" and be entirely correct. (Never mind the fact that there are far more errors your father is prone to that AI isn't.)
Yes, definitely. We have seen a lot of those failures here. They do funny non-failure things humans would never do. Obvious robot behavior. Things like: When pulling up to an intersection blocked in the side directions by construction, they will always religiously stop at the useless stop signs.
The real failures that I worry about that humans would never do are things like: Terrorists find a zero-day and make every Waymo suddenly do a hard left turn at full speed all at the same time.
I had heard about one study from Waymo, I didn't realize there are four different studies saying basically the same thing.
My parents are convinced that they're not safe enough. My wife is convinced that they're too safe because they keep blocking the two-lane causeway near our home when they're not willing to drive through sketchy construction areas.
Not much I guess, since they depend so much on the maps and on figuring out local knowledge on top of that.
It took Waymo more than a month of nearly daily attempts before it could figure out which way its cars should drive into the driveway at my apartment building. For a month straight, one of them would drive in through the exit at our building and then get stuck waiting for some remote human to rescue it. I will never understand why that happened more than one or maybe two times.
I'm still a big fan! But wow, what happened there.
Yes. But what is difficult for us humans to deal with is the fact that when Waymo does fail, it will be in a way that a human in the same situation would not have failed. Your father will be able to say, "I never would have made that mistake!" and be entirely correct. (Never mind the fact that there are far more errors your father is prone to that AI isn't.)
Yes, definitely. We have seen a lot of those failures here. They do funny non-failure things humans would never do. Obvious robot behavior. Things like: When pulling up to an intersection blocked in the side directions by construction, they will always religiously stop at the useless stop signs.
The real failures that I worry about that humans would never do are things like: Terrorists find a zero-day and make every Waymo suddenly do a hard left turn at full speed all at the same time.
Like that scene in "Leave The World Behind".
Still a fan!
I had heard about one study from Waymo, I didn't realize there are four different studies saying basically the same thing.
My parents are convinced that they're not safe enough. My wife is convinced that they're too safe because they keep blocking the two-lane causeway near our home when they're not willing to drive through sketchy construction areas.
If I measure stats from point A to point B, what's does that say about point B to point C?
Not much I guess, since they depend so much on the maps and on figuring out local knowledge on top of that.
It took Waymo more than a month of nearly daily attempts before it could figure out which way its cars should drive into the driveway at my apartment building. For a month straight, one of them would drive in through the exit at our building and then get stuck waiting for some remote human to rescue it. I will never understand why that happened more than one or maybe two times.
I'm still a big fan! But wow, what happened there.
I previously viewed some of the published stats and realized how much real world data they left out.
They are very selective as to what is included in their stats, leaving out parts of a trip that actual create problems for instance.