4 ms·

Average humans have a higher error rate.

by grey-area 5y ago

amelius 5y ago
If you classify an image 30 times per second, then 0.001% is quite a large error rate.
- OneTimePetes 5y ago
  Don't forget - the error might be reproduce-able. As in, it will error rarely, but at that one obstacle, with that one light config, it will error repeatedly and constantly. AI related car accidents and deaths will have special places and dates, were they will repeat annually.
- maxerickson 5y ago
  You said car, not image classifier. If errors in any subsystem surface out to the car, well then okay. But it's not unlikely that the overall system would deal with an error in the image classifier.
- 2muchcoffeeman 5y ago
  Pay wall so I can’t read the article. But can someone comment on what 0.001% error means in real world scenarios? To keep the maths easy, that’s 18 wrong frames/10min. So roughly 2/3 of a second. If these images are spread out, maybe it’s fine. If it’s failing on certain types of images, then those images could be grouped and 2/3rds of a second is more than enough time for a serious failure.
gameswithgo 5y ago
For me the question is how high is the error rate for a non drunk non sleepy non elderly non reckless human, since that is what my car would replace. in my 20s you would need to bring the reckless back in.
- taneq 5y ago
  Only if you’re the only driver being replaced.
- SuoDuanDao 5y ago
  There's also the fact though, that if you're elderly you can't really avoid being a worse driver. I know particularly in rural areas where there's no public transportation, a lot of elderly people would accept an error rate higher than what their own driving used to be. It seems pretty tough for the rest of us to deny them access to a technology like that while it's in the interstitial stage.
- rpadovani 5y ago
  Problem is, you are not alone on the road. And if somebody else does something really stupid, you can die as a consequence, without any chance to avoid it. We are definitely not there yet, but I think of myself as a decent driver, and some assistant tools on high-end models are way better (just faster, probably) than me predicting stuff. For the first year this summer, I've driven a car that was quicker than I in an emergency break situation: while I started breaking, it depressed the pedal, and I was quite surprised, 'cause the car in front of me didn't begin breaking yet.
LightG 5y ago
But, I assume, a lower completely random error rate, which an AI might produce.
ars 5y ago
No they don't, humans have a error rate of between 0.00001 and 0.000000001 depending on how you measure it.

Average humans have a higher error rate.

amelius 5y ago

If you classify an image 30 times per second, then 0.001% is quite a large error rate.

OneTimePetes 5y ago
Don't forget - the error might be reproduce-able. As in, it will error rarely, but at that one obstacle, with that one light config, it will error repeatedly and constantly. AI related car accidents and deaths will have special places and dates, were they will repeat annually.
maxerickson 5y ago
You said car, not image classifier. If errors in any subsystem surface out to the car, well then okay. But it's not unlikely that the overall system would deal with an error in the image classifier.
2muchcoffeeman 5y ago
Pay wall so I can’t read the article. But can someone comment on what 0.001% error means in real world scenarios? To keep the maths easy, that’s 18 wrong frames/10min. So roughly 2/3 of a second. If these images are spread out, maybe it’s fine. If it’s failing on certain types of images, then those images could be grouped and 2/3rds of a second is more than enough time for a serious failure.

gameswithgo 5y ago

For me the question is how high is the error rate for a non drunk non sleepy non elderly non reckless human, since that is what my car would replace. in my 20s you would need to bring the reckless back in.

taneq 5y ago
Only if you’re the only driver being replaced.
SuoDuanDao 5y ago
There's also the fact though, that if you're elderly you can't really avoid being a worse driver. I know particularly in rural areas where there's no public transportation, a lot of elderly people would accept an error rate higher than what their own driving used to be. It seems pretty tough for the rest of us to deny them access to a technology like that while it's in the interstitial stage.
rpadovani 5y ago
Problem is, you are not alone on the road. And if somebody else does something really stupid, you can die as a consequence, without any chance to avoid it. We are definitely not there yet, but I think of myself as a decent driver, and some assistant tools on high-end models are way better (just faster, probably) than me predicting stuff. For the first year this summer, I've driven a car that was quicker than I in an emergency break situation: while I started breaking, it depressed the pedal, and I was quite surprised, 'cause the car in front of me didn't begin breaking yet.

LightG 5y ago

But, I assume, a lower completely random error rate, which an AI might produce.

ars 5y ago

No they don't, humans have a error rate of between 0.00001 and 0.000000001 depending on how you measure it.