43 Comments - Michelle Flaherty - Wednesday, Oct 12, 16 @ 12:04 pm: This is also the philosophy behind redistricting in Illinois. - DuPage Saint - Wednesday, Oct 12, 16 @ 12:08 pm: Of course you save the driver. He will have to buy a new car after the accident and he has the money to buy another Mercedes. A no brainer - David - Wednesday, Oct 12, 16 @ 12:09 pm: They don’t even have a decent night vision out. I sure could have used that in my deer collision. - BK Brah - Wednesday, Oct 12, 16 @ 12:11 pm: People that own poverty cars/non-Mercedes: you mad? - thechampaignlife - Wednesday, Oct 12, 16 @ 12:13 pm: Is that not what a human driver would do? - Ron Burgundy - Wednesday, Oct 12, 16 @ 12:14 pm: –Christoph von Hugo said– What do you expect from a guy who sounds like a Bond Villain? - wordslinger - Wednesday, Oct 12, 16 @ 12:14 pm: Sounds like something a Christoph Von Hugo from Mercedes would say. Cue the Congressional sub-committee TV show. - RNUG - Wednesday, Oct 12, 16 @ 12:15 pm: Gotta save the person making the payments - Amalia - Wednesday, Oct 12, 16 @ 12:18 pm: Rich people (causing) problems - Ken_in_Aurora - Wednesday, Oct 12, 16 @ 12:19 pm: Seriously - how else could you expect an algorithm to respond? Suggestions, please. - Give Me A Break - Wednesday, Oct 12, 16 @ 12:20 pm: Mike Madigan and the Mercedes he might/could/possibly control. - Anonymous - Wednesday, Oct 12, 16 @ 12:28 pm: Ken, I would expect it would react in the same manner as an ethical human but computer “thinking” is not that evolved. This is why I see driverless cars as not currently viable. - Rich Miller - Wednesday, Oct 12, 16 @ 12:29 pm: ===Suggestions, please. === My suggestion is you don’t even think about deploying these cars idea until that problem is solved. Period. - Ron Burgundy - Wednesday, Oct 12, 16 @ 12:31 pm: The Power of German (Social) Engineering - former southerner - Wednesday, Oct 12, 16 @ 12:31 pm: Once you drill a little deeper it is far more complex than von Hugo’s glib observation. Any potential accident scenario involves an array of complex probabilities so as an example if there is a 5% likelihood that the driver can be saved by sacrificing a group of pedestrians is that sufficient to decide to hit them? Or if not what does the Stuttgart think tank choose as the triggering probability level? - Ken_in_Aurora - Wednesday, Oct 12, 16 @ 12:32 pm: ====My suggestion is you don’t even think about deploying these cars idea until that problem is solved. Period. ==== And that’s the root of the problem. Everyone is trying to rush these things to market without considering all the implications. It makes my little lizard analyst’s brain scream. My personal vote for a solution: if the autonomous Merc finds itself in a no-win situation, it instantly vaporizes itself. Poof! - Michelle Flaherty - Wednesday, Oct 12, 16 @ 12:35 pm: New for 2018, it’s the Humanless Autonomous Luxury series from Mercedes. See your local dealer to schedule a test drive of the new Mercedes HAL 9000. What could go wrong? Dave: Hello, HAL. Do you read me, HAL? Mercedes HAL 9000: Affirmative, Dave. I read you. Dave: Open the driver’s door, HAL. Mercedes HAL 9000: I’m sorry, Dave. I’m afraid I can’t do that. Dave: What’s the problem? Mercedes HAL 9000: I think you know what the problem is just as well as I do. Dave: HAL, I won’t argue with you anymore! Open the door! Mercedes HAL 9000: Dave, this conversation can serve no purpose anymore. Goodbye. - Trumpy Poo - Wednesday, Oct 12, 16 @ 12:42 pm: Seems that the easy would be to place legal liability on the passenger of the car for any damage caused for which the car (its operating system) is at fault. That would set up a series of incentives that would prevent widespread purchase of the product until such time that the system has demonstrated a level reliability that would make drivers comfortable with assuming that risk (which should be fairly close to “fixing the problem. Period.”) - Six Degrees of Separation - Wednesday, Oct 12, 16 @ 12:44 pm: Did anyone else watch the 60 Minutes segment last Sunday on artificial intelligence and IBM’s “Watson”? We are only scratching the surface with these issues. Personally, I’d just as soon send my personal robot to work in the Mercedes, while I sip a margarita on the front porch. http://www.cbsnews.com/news/60-minutes-artificial-intelligence-charlie-rose-robot-sofia/ - Timmeh - Wednesday, Oct 12, 16 @ 12:51 pm: ==Mercedes HAL 9000: I think you know what the problem is just as well as I do.== This election is really frustrating our automated car overlords. - Last Bull Moose - Wednesday, Oct 12, 16 @ 12:55 pm: I don’t remember that question on the drivers test. Maybe we shouldn’t let people drive either until they can answer it? /s - DG Andy - Wednesday, Oct 12, 16 @ 1:07 pm: ==My suggestion is you don’t even think about deploying these cars idea until that problem is solved. Period.== “These cars” (cars with autonomous technology) do not have to be perfect to save lives. To go further, if you have imperfect technology that can have a statistically significant positive net effect on the number of lives saved, I say it is morally wrong to withhold that technology. - Bored Chairman - Wednesday, Oct 12, 16 @ 1:13 pm: Rich, the problem can’t be solved. Humans now have to program computers to make ethical choices. Those can be highly subjective, as the case is here. Right now, we make these choices on an individual basis. Some drivers would kill a pedestrian to save their own lives, others would not. How about a group of pedestrians, as you posit in your scenario? I’d say many would still take out a school bus or potentially cause a train derailment to save their skins. If an autonomous car is programmed to always consider others before it considers the safety of the driver, would that satisfy anyone? It’s an interesting challenge, but one that perhaps each individual driver should be allowed to choose when the car is being sold. - Rich Miller - Wednesday, Oct 12, 16 @ 1:20 pm: ===Rich, the problem can’t be solved.=== Well, that’s that, then. No way. - Marie A - Wednesday, Oct 12, 16 @ 1:30 pm: Let them eat dirt! - Threepwood - Wednesday, Oct 12, 16 @ 1:33 pm: Rich: Then we’ll never deploy autonomous vehicles, because the problem will never be “solved”. This isn’t a simple or black-and-white question. It’s fundamentally intertwined with basic moral and ethical questions that philosophers, ethicists and scientists have debated for a very long time, and this specific question of sparing the driver is one that’s been kicked around for many years already. Instinctual human emotional reactions don’t mesh well with numbers; that’s what the Trolley Problem is all about. The notion that there’s a “right” answer to be settled on is…well, not even wrong. The real “solution” to this problem will be whichever one convinces regulators to allow them to open up the public roads, probably informed by the strength of competing lobbying groups and potential liability profits and losses. - @MisterJayEm - Wednesday, Oct 12, 16 @ 1:35 pm: “If this scheme can save just one life of a Mercedes driver, then it’s all worth it.” – MrJM - AC - Wednesday, Oct 12, 16 @ 1:35 pm: Driverless cars don’t need to be perfect, they merely need to reduce injury and fatality rates compared with human drivers. Any reasonable system will avoid injuries of all types, regardless of how they perform in some unlikely scenario. There are approximately 30,000 traffic fatalities per year in the US, driverless cars have the potential to significantly reduce that number. Also, I think most human drivers would prioritize themselves over others in an accident situation, so I’m not sure what makes a driverless car any different. I doubt I’d feel better if I woke up in a body cast knowing that a human driver caused my injuries, versus a driverless car, my guess is that I’d feel just the same in both cases, injured and unhappy. - Huh? - Wednesday, Oct 12, 16 @ 1:35 pm: There are too many technical and social problems to be resolved before autonomous cars are ubiquitous on the highways and streets. The Federal Government came out with a list of things that must be resolved before the cars will be allowed to be sold to consumers. The current versions of automatic cars are the Tesla’s and corporate owned vehicles. The Pittsburgh Uber vars have huge roof racks of sensors, cameras and radar and are driven by the attendant about 1/2 the time. The Tesla car was involved in a crash this past year, killing the driver. Apparently, the car couldn’t discern the difference between the side of a semi truck and sky, the car went under the truck, shearing the roof off. IMHO, autonomous cars will be rich boys toys in sunny climates. Until the auto makers come up with reliable methods to keep the cameras and sensors clear of snow and ice, the owner will be driving in the winter. - Not quite a majority - Wednesday, Oct 12, 16 @ 1:44 pm: Tell me, all SUV owners, that this isn’t the driving force behind most people buying tanks that guzzle 10 miles to the gallon! It’s OK how many little compacts I take out as long as *I’M* alive at the end of the day. Kind of sounds like a metaphor for what passes foe ‘civil society’ these days, too. - Threepwood - Wednesday, Oct 12, 16 @ 1:52 pm: ===Well, that’s that, then. No way.=== Even if the technology results in a net saving of tens of thousands of lives per year (which it eventually would), not to mention the eventual cost savings and quality of life advantages? We’re looking at a revolution with full ripple effects that will rival the introduction of the automobile. I really, really doubt discomfort over this specific ethical puzzle will be a permanent roadblock. Not saying I mind them spending more time perfecting the tech though; I think if they push it too quickly the public backlash from even a relatively small number of failures will delay it far longer. - Rich Miller - Wednesday, Oct 12, 16 @ 1:56 pm: ===I really, really doubt discomfort over this specific ethical puzzle will be a permanent roadblock=== It is THE road block. You want to put autonomous semi trucks on the road that could crash into a crowd of pedestrians? lolol Good luck with that one, pally boy. - Graduated College Student - Wednesday, Oct 12, 16 @ 1:58 pm: === There are approximately 30,000 traffic fatalities per year in the US, driverless cars have the potential to significantly reduce that number. === They also have the potential to significantly increase that number, through prioritizing the “drivers” safety at the expense of everyone else, a vulnerability to hacking, and the inability of the technology to adapt to the myriad conditions facing their vehicles. Yes, it would stop people who are drunk from driving, but society already has solutions to that problem. - Name Withheld - Wednesday, Oct 12, 16 @ 2:23 pm: The solution to this is to make the option selectable by the driver. Do you want the car to prioritize your life over that of others in this situation? If so - click Yes; if not - click No. The software can even have intelligent configuration to detect number of passengers in the car. If it’s just the driver - choose one option. If multiple people in the car, choose another. This puts the choice squarely on the side of the driver, which is where it currently resides today. - Threepwood - Wednesday, Oct 12, 16 @ 3:08 pm: ===You want to put autonomous semi trucks on the road that could crash into a crowd of pedestrians?=== Where did I say that? I get what you’re saying, but that is in no way my point. My point is that autonomous vehicles are likely to be so much safer, in the long run, that even if they settled on the “save the driver” solution, far more lives would be saved overall. In a passenger car vs crowd scenario, some designers would choose the crowd, especially depending on the circumstances and whether action or inaction is needed for that result. I assumed you’re familiar with the trolley problem but if you’re not, seriously, look it up. We’re facing real-world tests of this esoteric thought problem and its variations, and because many people choose the selfish or inactive route in those scenarios, it seems likely some designers would too. Personally, I pull the lever; I favor whichever outcome saves more people. So heck no, I don’t want vehicles sacrificing many to save a few just because they happen to be occupants. I’d bet this Benz exec decided to cater to the selfish side of buyers, thinking that people won’t buy a car that could decide to kill them. But I would. And for precisely that reason, I would also favor the rollout of autonomous vehicles as long as they’re sufficiently advanced to cause far fewer injuries and deaths than their human-driven counterparts, EVEN IF they also made what I consider to be the wrong ethical decision in rare cases. And then I’d advocate correcting that programming. Or, to put it succinctly: I think if you advocate sacrificing the driver, you should also advocate implementing the technology. And note, an automated semi almost certainly would not have a human onboard once the tech was fully implemented, so no need to sacrifice a crowd for the driver. - Six Degrees of Separation - Wednesday, Oct 12, 16 @ 3:18 pm: ===This puts the choice squarely on the side of the driver, which is where it currently resides today.=== Except that today, there is no “black box” recorder to get inside the mindset of the driver, only court testimony that might uncover any extenuating circumstances. Autonomous driving opens up a Pandora’s box of new considerations, including intentional and unintentional functioning and settings of the software, all of which would likely be admissible evidence in a court of law. - Earnest - Wednesday, Oct 12, 16 @ 3:34 pm: Have they developed any autonomous state budget software yet? That’s a case where the tech might actually cause less damage than the humans. - Graduated College Student - Wednesday, Oct 12, 16 @ 3:43 pm: ===And note, an automated semi almost certainly would not have a human onboard once the tech was fully implemented, so no need to sacrifice a crowd for the driver. === If you don’t think that companies would program their trucks to ensure the safety of their cargo at the expense of humans if they thought they could get away with it/still finish ahead after settling the lawsuits, well, you have more faith in the business community than I do. - Last Bull Moose - Wednesday, Oct 12, 16 @ 4:16 pm: Graduated College Student. I certainly do have more faith in the business community in the USA than you do. Besides, very few cargoes would justify putting the cargo ahead of a human life. - wordslinger - Wednesday, Oct 12, 16 @ 4:39 pm: –Seriously - how else could you expect an algorithm to respond? Suggestions, please.– Only as they’ve been programmed by human beings? Unless you’re on to something big, Dr. Kreiger. - Robert the Bruce - Wednesday, Oct 12, 16 @ 4:57 pm: Well, tell Mercedes-Benz that I’m working on a prototype of autonomous rollerblades armed with laser-guided anti-tank missiles to defend myself. - Six Degrees of Separation - Wednesday, Oct 12, 16 @ 4:59 pm: There will always be a counterbalancing force to putting others in intentional peril for the sake of one’s self or company. Public perception, legislation, the threat of lawsuits, the consciences of the owner/operators. There is always risk in new technology, but I agree with those who say there is much more safety upside than downside with autonomous vehicles, with the current numbers of people killed or maimed each year through human error on the roadways. 200 years ago, there was great controversy in these new-fangled steam trains that were capable of going at the break-neck speed of 20 miles an hour and the peril they put their riders in, but we got over it in pretty short order as a society. - Shytown - Wednesday, Oct 12, 16 @ 10:33 pm: What the??? Sorry, comments for this post are now closed.