The press has really gone into hyperdrive - the L.A. Times just copied the New York Times and has released its own "editorial board" opinion piece saying we should "hit the brakes" on autonomous driving:
It's time to tap the brakes on self-driving cars
Also, after a brief break in front page coverage because of the police shootings issue, Tesla is back on the L.A. Times' front page tonight.
We who pay attention to little details all realize a much more powerful autopilot is on the cusp of release - one which will probably much reduce the likelihood of another accident like Joshua Brown's.
But Tesla's "Ta Da - you never expected THIS did you????" method of dropping new products and technologies prevents Musk from telling everyone "Hey Autopilot went 130 million miles with no deaths AND in just a couple months we have a new-and-improved version which will be even safer."
Tesla is so secretive with its fleet learning methodology that nobody outside the autopilot team actually knows what the system has learned, how it accomplishes this and how quickly. Just knowing that there is some kind of neural network being trained on top of Mobileye's own trainable neural network, and that exceptions by drivers are sent back to HQ is NOT a deep explanation to anyone of how the system works. I don't need one myself - I trust Musk. But a lot of people seem not to.
I understand the reasons for both of these strategies - but it does seem they also are preventing a more vigorous defense of Tesla's rapidly improving abilities to avoid accidents.
Monday evening front page of latimes.com:
It's time to tap the brakes on self-driving cars
Also, after a brief break in front page coverage because of the police shootings issue, Tesla is back on the L.A. Times' front page tonight.
We who pay attention to little details all realize a much more powerful autopilot is on the cusp of release - one which will probably much reduce the likelihood of another accident like Joshua Brown's.
But Tesla's "Ta Da - you never expected THIS did you????" method of dropping new products and technologies prevents Musk from telling everyone "Hey Autopilot went 130 million miles with no deaths AND in just a couple months we have a new-and-improved version which will be even safer."
Tesla is so secretive with its fleet learning methodology that nobody outside the autopilot team actually knows what the system has learned, how it accomplishes this and how quickly. Just knowing that there is some kind of neural network being trained on top of Mobileye's own trainable neural network, and that exceptions by drivers are sent back to HQ is NOT a deep explanation to anyone of how the system works. I don't need one myself - I trust Musk. But a lot of people seem not to.
I understand the reasons for both of these strategies - but it does seem they also are preventing a more vigorous defense of Tesla's rapidly improving abilities to avoid accidents.
Monday evening front page of latimes.com:
Last edited:
