I almost always avoid watching any video “evidence” in news stories, such as that posted yesterday by Tempe, Arizona Police showing Uber footage of its Autonomous Vehicle striking a pedestrian. My avoidance is based in part on a strong inability to deal with video violence in any setting. For instance, I’ve never quite understood the eagerness to repeatedly show, and watch, replays of athletes being gruesomely injured. It’s just not for me.
More important in terms of news, however, is that video is perhaps the single most easy medium in which to manipulate perception and understanding, while it is also, rather inexplicably, seen as the single most “convincing” form of evidence for most humans. Taking this human set of traits for a bit of a ride, what is the allure of watching magicians like David Blaine on TV?
You’ve simply got a magician manipulating perception right in front of you? Okay, that can be interesting, I guess. But videotaping magic tricks? That’s like dropping atomic bombs of misperception, one after the other, just to make the eye/brain “rubble bounce.”
So, as with all of us, it is with with my own understanding, perception and bias that yesterday afternoon I did watch the Uber video footage as released by the Tempe Police.
Tempe Police Vehicular Crimes Unit is actively investigating
the details of this incident that occurred on March 18th. We will provide updated information regarding the investigation once it is available. pic.twitter.com/2dVP72TziQ— Tempe Police (@TempePolice) March 21, 2018
And then I spent the rest of the afternoon and evening in something of a deliberately protective coma, a bit like the body going into shock immediately after a traumatic injury.
It’s only this morning, after having spent a long half-sleepless night rolling it all around that I can write about it this morning. No, I’m not made of the stuff that can watch such things, and yes, my strong understanding of the power of video to deceive has me doubting if any, even one iota, of certainty can be gained by watching this video.
Except one fact.
The victim was in the right driving lane at the time of impact, heading right. Not the left lane nearest the desert landscaped median. Not having just “darted out” from there.
The right driving lane. Heading right.
Looking at Google Streetview on Tempe’s Mill Avenue in daytime, impact occurred in the right lane right about here:
Or more probably, having perhaps ruined my day by looking again at the video this morning, just up the Mill a bit about here:
Yeah, impact is almost certainly here, as you can make out the beginning of the left turn lane. All that open viewing space between the median, left turn lane, left driving lane, and point of impact. Note, too, the streetlights in the Streetviews above.
Again, my understanding and bias about video “evidence” is too strong to make any argument beside this: the impact was in the right driving lane.
And that fact, admittedly, video manipulation/misperception capacity notwithstanding, is damning and enough, in my view, to stop ALL autonomous vehicle usage EVERYWHERE until shit like this doesn’t have ANY chance of happening.
This beta-testing on humans, even if they are lowly pedestrians (and let’s face it, those walking our streets instead of driving are generally perceived as lesser humans), must stop.
Their are two ways Arizona can “solve” this problem.
1) Before redeploying the robots,make sure the AI steps up it’s game sufficiently to be, at the very minimum, a clear safety upgrade over human drivers. Preferably to the point where Asimov’s 1st law of robotics is in effect, but that might be too high a bar.
2) Build a giant fences around vast swathes of public right of way, designating it as not for access by pedestrians, and basically using the “Jay-walking” solution. It’s always the pedestrian’s fault.
I wish I was more hopeful that a state like Arizona would choose #1.
LikeLike