4 ms·

My point is that Waymo isn't a good example of Google's success considering that after decades of development it still can't drive on the right side of the road

My point is that Waymo isn't a good example of Google's success considering that after decades of development it still can't drive on the right side of the road in broad daylight on a clear day.

dageshi 2y ago
Which is no different to a whole host of bad human driving as well. Mistakes are inevitable in something like driving, they happen to even the best human drivers. All that matters is that over time the self driving mistakes decrease to the point they are far below the human average. I don't see the point of picking one example of a mistake and pretending it's indicative of a failing endeavour when literally no other company is even close to what Waymo is doing right now.
- PostOnce 2y ago
  Several different comments in this sub-thread call this a "mistake". Machines don't make mistakes. If a crane collapses, it didn't make a mistake. If a robot car drives into oncoming traffic, it's not a mistake, it's an engineering failure, a bad design. This wasn't a fuckup in a prototype, this was a life-critical fuckup 15-20 years into the project which is still confined to a small number of well-mapped cities with good weather. So my point stands, major fuckups decades into a project when conditions are fairly close to ideal are not acceptable and are indicative of a failed project. As an aside, it's also not the case that "better than human" is all that matters. The kind of fuckup also matters because people vote, and votes change regulations. If people don't like it, it won't be legal. Driving into oncoming traffic in broad daylight with 20 years of engineering progress is not something people are going to say "oh okay well sometimes my grandma does that too" to. You guys are wearing rose-tinted glasses.
dekhn 2y ago
The car almost certainly made a reasonable decision based on the information available. They have all the technology and logic required to make decisions based on ambiguous and vague signage, and if they hadn't done this, we'd be reading articles about how Waymos would stop in situations where humans would figure out what to do. There is a lot of press trying to "gotcha" Waymo but from what I can tell, their systems are actually quite good.
Workaccount2 2y ago
Waymo is a failed experiment because this one time one of the cars made a mistake. Do you hear how naive you sound?

My point is that Waymo isn't a good example of Google's success considering that after decades of development it still can't drive on the right side of the road in broad daylight on a clear day.

dageshi 2y ago

Which is no different to a whole host of bad human driving as well. Mistakes are inevitable in something like driving, they happen to even the best human drivers. All that matters is that over time the self driving mistakes decrease to the point they are far below the human average. I don't see the point of picking one example of a mistake and pretending it's indicative of a failing endeavour when literally no other company is even close to what Waymo is doing right now.

PostOnce 2y ago
Several different comments in this sub-thread call this a "mistake". Machines don't make mistakes. If a crane collapses, it didn't make a mistake. If a robot car drives into oncoming traffic, it's not a mistake, it's an engineering failure, a bad design. This wasn't a fuckup in a prototype, this was a life-critical fuckup 15-20 years into the project which is still confined to a small number of well-mapped cities with good weather. So my point stands, major fuckups decades into a project when conditions are fairly close to ideal are not acceptable and are indicative of a failed project. As an aside, it's also not the case that "better than human" is all that matters. The kind of fuckup also matters because people vote, and votes change regulations. If people don't like it, it won't be legal. Driving into oncoming traffic in broad daylight with 20 years of engineering progress is not something people are going to say "oh okay well sometimes my grandma does that too" to. You guys are wearing rose-tinted glasses.

dekhn 2y ago

The car almost certainly made a reasonable decision based on the information available. They have all the technology and logic required to make decisions based on ambiguous and vague signage, and if they hadn't done this, we'd be reading articles about how Waymos would stop in situations where humans would figure out what to do. There is a lot of press trying to "gotcha" Waymo but from what I can tell, their systems are actually quite good.

Workaccount2 2y ago

Waymo is a failed experiment because this one time one of the cars made a mistake. Do you hear how naive you sound?