I think AI vehicle piloting tech might help considerably.
As someone with many years of professional driving experience, I'm very skeptical that fully autonomous AI-piloted vehicles will ever meet reliability and safety standards to the point where they can entirely replace human drivers and navigators. Too many unique condition sets--"unknowns"--for non-human cognition to accurately and safely interpret, often regarding situations that human intelligence can hack routinely in order to find workarounds.
However, I think it could be lifesaving for more automobiles to be equipped with AI to assist human drivers, as a co-pilot. One of the functions that would be most worthwhile would be the ability to monitor the driving responses of a human driver to detect signs of exhaustion or impairment, such as lane weaving, clearly slowed reflexes, poor use of the acceleration and brake pedals, etc. AI monitoring might signal an initial warning to the driver, say, and then continue monitoring to find out if their driving returns to a competent standard. But if the human driver continues making errors, the AI program could find the shoulder of the road or another safe place to pull the car over to a stop, and then shut it down for a couple of hours. And if the driver starts up the vehicle after that rest period and they're still unable to drive safely, AI could shut it down again.
It's my impression that the technology to accomplish this level of AI piloting is already available. I'm mystified about what makes fully autonomous AI vehicle piloting such an essential goal. This truly looks like a case where the perfect is the enemy of the good. The hurdles related to many modes of partial AI-assisted vehicle piloting have already been achieved, including those that are in my opinion the most important features- like maintaining safe distances between vehicles all the way around, and proper braking in traffic. Whereas completely replacing human drivers would probably entail an interconnected grid system, a technology that's both potentially dystopian and unacceptably fragile. A single failure at a critical junction would be potentially catastrophic. We're much better off with the flexibility and fast response of decentralized decision-making than centralized interlinked control. This is not Legoland, it is not SimCity, it can't be modeled entirely on a screen as a flawlessly performing closed system.
I'm not making the argument against fully autonomous "driverless vehicles" as Luddite gainsaying in order to preserve human jobs. I'm actually not that worried about the effects of UAVs on employment, because driving is almost always only part of the job duties of "driving jobs." Humans still perform tasks like inventory accountability, safe and efficient cargo loading and unloading, passenger boarding and exit, pedestrian assistance for cab passengers, impromptu changes in navigation, and other tasks that have nothing to do with vehicle piloting much more responsively, safely, efficiently, and reliably than any practical robot technology. No AI assistance is going to load a suitcase into an automobile boot, for example, or keep the interior cleaned up between rides. It isn't going to set up the tracks to unload cargo into a warehouse. Etc., etc.
But I'm confident that an AI program is able to detect unsafe vehicle piloting, and safely take steps required to find a safe place to park it and disable the driving functions in the event that it's required.
I've been pointing this out in story comment discussions for quite a number of years. All I need is one person- or, I don't know, maybe an AI program- to read, pay attention, and take my suggestions seriously.
Ric- this is an important topic. I’m glad you covered it. I’ve also heard that the rate of accidents are higher around Daylight Savings time? I appreciate the reminders here.
Thank you for this, Ric! Collecting DUIs is treated almost like a sport out here, and an arrest for a 4th/5th/6th might make page 2 or 3 in the paper. Maybe. The Tavern League is extremely good at lobbying, and the rest of us all pay for it.
Forgot to talk about the interlock/breathalyzer. Those things aren’t terribly reliable. Sometimes you blow and blow (lol) on the thing but it still won’t listen. So your husband gets to come and pick you up. And then you take the car to a place that can fix your breathalyzer. More $ more pain in the ass.
I drove drunk once. I was about 18 and came around a corner too fast, encountering an older man dressed in rags. I was able to come to a stop and the man man misunderstood. zhe climed into the car; I handed the keys to a sober friend and haven[t driven drunik since then. That was about sixty years ago. I haven't touched alcohol for the last five years since I gave up driving because of seizures. I was ,smoking in my car in our driveway/carport. Giving up driving meant giving up the two triggers for smoking, alcohol and coffee. Of the three, the only one I still miss is coffee.
I think AI vehicle piloting tech might help considerably.
As someone with many years of professional driving experience, I'm very skeptical that fully autonomous AI-piloted vehicles will ever meet reliability and safety standards to the point where they can entirely replace human drivers and navigators. Too many unique condition sets--"unknowns"--for non-human cognition to accurately and safely interpret, often regarding situations that human intelligence can hack routinely in order to find workarounds.
However, I think it could be lifesaving for more automobiles to be equipped with AI to assist human drivers, as a co-pilot. One of the functions that would be most worthwhile would be the ability to monitor the driving responses of a human driver to detect signs of exhaustion or impairment, such as lane weaving, clearly slowed reflexes, poor use of the acceleration and brake pedals, etc. AI monitoring might signal an initial warning to the driver, say, and then continue monitoring to find out if their driving returns to a competent standard. But if the human driver continues making errors, the AI program could find the shoulder of the road or another safe place to pull the car over to a stop, and then shut it down for a couple of hours. And if the driver starts up the vehicle after that rest period and they're still unable to drive safely, AI could shut it down again.
It's my impression that the technology to accomplish this level of AI piloting is already available. I'm mystified about what makes fully autonomous AI vehicle piloting such an essential goal. This truly looks like a case where the perfect is the enemy of the good. The hurdles related to many modes of partial AI-assisted vehicle piloting have already been achieved, including those that are in my opinion the most important features- like maintaining safe distances between vehicles all the way around, and proper braking in traffic. Whereas completely replacing human drivers would probably entail an interconnected grid system, a technology that's both potentially dystopian and unacceptably fragile. A single failure at a critical junction would be potentially catastrophic. We're much better off with the flexibility and fast response of decentralized decision-making than centralized interlinked control. This is not Legoland, it is not SimCity, it can't be modeled entirely on a screen as a flawlessly performing closed system.
I'm not making the argument against fully autonomous "driverless vehicles" as Luddite gainsaying in order to preserve human jobs. I'm actually not that worried about the effects of UAVs on employment, because driving is almost always only part of the job duties of "driving jobs." Humans still perform tasks like inventory accountability, safe and efficient cargo loading and unloading, passenger boarding and exit, pedestrian assistance for cab passengers, impromptu changes in navigation, and other tasks that have nothing to do with vehicle piloting much more responsively, safely, efficiently, and reliably than any practical robot technology. No AI assistance is going to load a suitcase into an automobile boot, for example, or keep the interior cleaned up between rides. It isn't going to set up the tracks to unload cargo into a warehouse. Etc., etc.
But I'm confident that an AI program is able to detect unsafe vehicle piloting, and safely take steps required to find a safe place to park it and disable the driving functions in the event that it's required.
I've been pointing this out in story comment discussions for quite a number of years. All I need is one person- or, I don't know, maybe an AI program- to read, pay attention, and take my suggestions seriously.
Ric- this is an important topic. I’m glad you covered it. I’ve also heard that the rate of accidents are higher around Daylight Savings time? I appreciate the reminders here.
Thank you for this, Ric! Collecting DUIs is treated almost like a sport out here, and an arrest for a 4th/5th/6th might make page 2 or 3 in the paper. Maybe. The Tavern League is extremely good at lobbying, and the rest of us all pay for it.
This was an informative and interesting post.
Forgot to talk about the interlock/breathalyzer. Those things aren’t terribly reliable. Sometimes you blow and blow (lol) on the thing but it still won’t listen. So your husband gets to come and pick you up. And then you take the car to a place that can fix your breathalyzer. More $ more pain in the ass.
I drove drunk once. I was about 18 and came around a corner too fast, encountering an older man dressed in rags. I was able to come to a stop and the man man misunderstood. zhe climed into the car; I handed the keys to a sober friend and haven[t driven drunik since then. That was about sixty years ago. I haven't touched alcohol for the last five years since I gave up driving because of seizures. I was ,smoking in my car in our driveway/carport. Giving up driving meant giving up the two triggers for smoking, alcohol and coffee. Of the three, the only one I still miss is coffee.