Model 3 vs. Glowforge

This is certainly an interesting read. The one issue I have with it has been mentioned before. Many of the choices that they ponder are things that can not be gauged by the vehicle doing the driving. Does it know that there is a child in the car? Maybe, it is certainly possible. Does it know that the group of 10 pedestrians has a child among them? Maybe, but that child could be occluded by an adult in the group, or a baby might be strapped to an adult and the car would not be able to tell it from a backpack.

I think the approach @johnse mentioned is the best. Can I avoid this collision without colliding with something else? Yes, then avoid the collision. No? Then have the collision.

That’s far better than the car trying to make a value decision. In court… “The car thought that it was going to hit a child in a stroller, so it swerved into the single vehicle beside it. Unfortunately that vehicle was a school bus and 10 children died in the collision. And the stroller was a cart filled with recycling for a homeless person.” - That happened in Speed right?

So yeah, Can I avoid this collision without triggering a collision at all? Yes, then swerve. No, then apply brakes.

2 Likes

Artificial Intelligence as being referred to here is where computers/robotics take on human like continuous learning independent of of initial programming. This ultimately results in machine sentience. The machine is no longer a machine. To science fiction writers and technologists over the last 75 years, that is AI

1 Like

You’re all missing the big picture in all of these thoughts though. The end goal is going to be everyone involved is in a computer driven car. Humans are the real danger, even with the little bit of AI the computers use now, they drive far more safely than your average driver. And when all the cars can discuss what to do in nanoseconds, it doesn’t matter that there are kids in a car, or a stroller somewhere, all the cars on that road can react to it.

3 Likes

That would be the Utopian vision of it, yes.

Unfortunately we won’t ever live in a utopia. Even if 100% of cars are autonomous, there will still be human elements. The first is the fact that humans programmed and designed the cars, therefore the cars/programs will ultimately fail at some point because as we all know, humans are fallible. The second is the aforementioned brat that steps into traffic. It doesn’t matter if every car is autonomous if a person gets in the path of a car that can’t stop in time. The posited situation is that the alternative to hitting the kid is hitting a wall and killing the driver and passengers in the car. What should the car do?

It’s a Kobayashi Maru… an unwinnable situation.

2 Likes

I’m just waiting for that first EMP to hit. That will be a mess, won’t it? :stuck_out_tongue_winking_eye:

7 Likes

A solar flare will do just fine hahaha.

3 Likes

like the Carrington Event.

3 Likes

Obviously pedestrians will just need to buy self walking leg upgrades that will have their own algorithms to prevent unsafe operation?

6 Likes

You’re making it an impossible scenerio to try and defend a point that doesn’t exist. The argument isn’t that a computer will never fail, it simply has to fail less than a human. The argument isn’t that a machine will have to choose between a childs life and the drivers, it’s that the machine will react to the random nature of driving faster than any human could. Accidents will drop sharply, unavoidable accidents will result in less injury, and so forth.

And all this is very far from a Utopian society, it will happen in your lifetime unless you’re 60+ right now. Simply because it’s going to be massively profitable to switch, and if there was ever a thing that motivated change, its profit.

4 Likes

That remains to be seen.

If fewer cars are crashing, that’s fewer new car sales and less cars keeping repair businesses busy. However, that may be offset by population increases. Or maybe there will be an explosion of new companies getting into ride sharing or ride hailing and driving fleet sales, but maybe because of that fewer consumers buy personal cars.

Lots of maybes floating around out there that we just won’t know until we get to that point.

I’m sure 30 years ago nobody saw the decline in brick and mortar retail we have today, or the local video rental store being reduced to a few self-serve kiosks as the masses go online for movies.

1 Like

i don’t think it really remains to be seen; there’s enough profit involved that somebody will see to it, regardless of whether traditional industries do (e.g. uber, waymo, volvo’s ridesharing effort, etc).
chances are very good that insurance companies will help drive adoption by offering discounts, too.

Every time this conversation comes up I get sad about our pathetic rail and public transportation in this country. Last year, I got Amtrak sleeper car train tickets for my family to travel from NC to DC. It was a long and painful trip, though super fun and totally worth it. Makes it a little easier that we never really put in a great rail system here.

1 Like

mm, yeah, it’s difficult to say whether they would have been similarly successful as the interstate program. it’s fun to think about what might have been, though.

Where are you guys living? lol. Uber and Lyft are both testing automated taxi service removing the single largest expense they have. It’s happening now. Every car company and arguably most of the top tech companies in the world are actively researching and developing automated driving.

Truck drivers won’t exist in the very near future, removing a huge cost and risk in shipping product. Imagine a fleet of trucks that can legally run 24/7 because it never gets tired and doesn’t demand overtime.

It’s a big part of why universal income is being discussed more frequently, the robots are coming and they are better than us. :smiley:

5 Likes

I’m not doubting any of that.

I’m saying, today, automakers make profit in part because people crash and destroy their cars. They profit when somebody totals their car and needs to buy a new one. They profit when someone just bangs up their car and takes it to their service dealer for repair. Autonomous vehicles will shift that, and yes they will either adapt their business to suit, or ask for more government handouts.

2 Likes

they will adapt. i mean they may also die, this is a period of rapid upheaval but they’re already adapting and making investments into side businesses (autonomy, ride sharing, etc). they have to simply because a lot of these changes are coming from outside of the traditional auto making business. reminds me a lot of chipmakers suddenly getting into storage…now that software (beyond engine control chips, etc) is finally considered important, we see sw-rich industries moving into the space.

1 Like

Truck driving is the number one occupation in the majority of states. It will be brutal.

4 Likes

Coming in late on this thread, but the self driving topic is so interesting, I want to backlog some responses in there :slight_smile:

As for the comments about not wanting to ever drive the speed limit… I am from Montana. Even in a place where “the next town over” is 4 hours away, speeding does not buy you a notable amount of time at the end of things. Within your city, it means FAR less. I long ago realized that increasing the probability of your and other’s deaths is not worth a few moments. Meanwhile, the productivity or relaxation you can fit in during the full time of a commute? THAT is worthwhile.

As to difficulties needing to be overcome for self driving cars to function… Roads are a recognizable form, and don’t tend to move around much. You can pretty easily program in every single road with accelerometers and GPS combined in such a way that the car knows precisely where the road is, regardless of complete fog and white out or snow cover. The one thing which can be different about a road is construction, which is clearly marked, and can be equipped with additional electronic broadcasting to update self driving cars.

For the ethics concerns… those don’t come up. Like… EVER. And once again, the computer vision and added sensors mean that the self driving cars are even less capable of ever being placed in such a situation. It can know there is a kid behind a row of cars moving erratically, and slow down. Absolutely any situation I have read where the trolley problem is brought up, there is a logical error in assuming the self driving car has to wait until the proposed moment to make the impossible decision.

The idea of making self driving cars broadcast/identify themselves to human drivers… I actually can dig that. Gut reaction was “HELL NO!” But, reflection on the idea reveals that the analysis here is all wrong. People said that human drivers would take advantage of the self driving cars. But… I think law enforcement is more likely to take advantage of the cameras on the road able to record and report bad driving instantly. And in this light, just as a police car is clearly marked because of the danger of legal repercussion… you would need the self driving car marked.

For society accepting self driving cars… look at the history of the automobile being adopted. Roads used to be primarily foot traffic. It took no time at all for the convenience of speed to completely reconfigure what we consider a city/town to be.

Yes… eventually a self driving car will happen to result in a death or disability, in spite of the sensors and algorithms. And at that time, there will be outcry, especially if a child is in ANY way connected (even just their parent being severely injured). But, once again… cars equipped with cameras can be leveraged to build social capital. Make sure to show everyone every single life saved because of autonomous driving. There won’t be many, because most cases that kill people now simply won’t happen. But let us see the car which avoids running over the idiot child running into the street a few dozen times before the inevitable accident in a case no human could have avoided either. Make sure people understand that the cars are better, and when they finally show they are not flawless, emotions will be tempered.

As for the later in the thread discussion about job losses… those are rampant. And it is why automation and assured minimum lifestyle are so big in conversations lately. Having a basic living guaranteed is possible with automation widespread enough to support people without any people being involved in the process. And once we CAN do that, guaranteed minimum lifestyle will remove the need to consider job loss as a concern. It will also enable people to get up and move to a place where the local politics fit their preferences, rather than wail and moan about the situation they are “forced” to stay in.

4 Likes

I agree with you. I was merely pointing out what some are saying will be a potential roadblock to autonomous vehicles. In the end the car only needs to make the choice of “can I avoid an accident”, if the answer is no, then the kid gets hit. But the parents of that kid don’t care that deaths via vehicle have dropped by 20% (made up number), they will only see that a self driving vehicle “chose to murder their baby”. Even so, that won’t stop them from implementing it as soon as they can.

1 Like

It isn’t hard to imagine a scenario where the car has no data to predict an interaction with a human. I mean a young child could easily be 100% occluded by the vehicle it is behind, so the car has no way of knowing that object even exists, let alone is on a trajectory that will cause a collision. There is also the potential person standing still or walking parallel to the road. The car has no reason to predict that the person will suddenly bolt out into the street.

People have to accept that innocent people will die as a result of any vehicular activity, and realize that the number of vehicle casualties will be drastically reduced as autonomous vehicles roll onto the streets.

2 Likes