3 ms·
This is the thing with errors and automation. A 1 % error rate in a human process is basically fine. A 1 % error rate in an automated process is hundreds of tho
by kqr 2y ago
This is the thing with errors and automation. A 1 % error rate in a human process is basically fine. A 1 % error rate in an automated process is hundreds of thousands of errors per day.
(See also why automated face recognition in public surveillance cameras might be a bad idea.)
- atoav 2y agoExactly. If your system monitors a place like a halfway decent railway station half a million people per day is a number you could expect. Even with an amazingly low error rate of 1% that would result in 5000 wrong signals a day. If we make the assumption that the people are uniformly spread out througout a 24 hour cycle that means a false alarm every 20 seconds. In reality most of the people are there during the day (false alarm every 10 seconds) and the error percentages are nowhere near 1%. If you do the math to figure out the staff needed to react to those false alarms in any meaningful way you have to come to the conclusion that just putting people there instead of cameras would be a safer way to reach the goal.
- Terr_ 2y agoAnother part is that artificial systems can screw up in fundamentally different ways and modes compared to a human baseline, even if the raw count of errors is lower. A human might fail to recognize another person in a photo, but at least they won't insist the person is definitely a cartoon character, or blindly follow "I am John Doe" written on someone's cheek in pen.
- Retr0id 2y agoHuman error rates are also not a constant. If you're about to publish a career-ending allegation, you're going to spend some extra time fact-checking it.
- atoav 2y agoCan you point to where that claim was made? I can't find it. The parent post assumes 1% for the sake of argument to underline that the impact of the 1% error depends on the number to which the 1% are applied — automation reduces the effort and increases the number. Hypothetical example: Cops shoot the wrong person in x% of cases. If we equipped all surveillance cameras with guns that also shoot the wrong person in x% of cases the world would be a nightmare pandemonium, simply because there is more cameras and they are running 24/7. Mind that the precise value of x and whether is constant or not does not impact the argument at all.
- Retr0id 2y agoI'm the one making the claim, and it does not attempt to refute the parent post, it agrees with/reinforces it. I'm also making the point that a human with an error rate of x% is not directly comparable to a machine with x% error rate, just via a different line of reasoning.
- yen223 2y agoIsn't this just saying "humans are slow" in a different way?
- kqr 2y agoYes. Slow is good in these Blackstone's ratio type situations when the cost of false positives are very high.