Transportation

How Self-Driving Cars Will Cope With Unruly Riders


Uber will ban riders that have an undesirable rating by their driver, which for driverless cars would seem to not arise, but it turns out AI-derived rider ratings will be significant for self-driving cars.

Getty

We’ve all had occasions to rate our rideshare drivers, typically on a scale of 1 (worst) to 5 (best), but did you also know that the drivers are able to rate you too, the passenger or rider in their vehicle?

When the rating of riders by rideshare or ride-hailing drivers was initially proclaimed in the media, there was a vocal backlash by some who felt this was the shoe being put on the wrong foot. A passenger is a customer, therefore rightfully should be able to render judgment about their service provider (i.e., the driver), and opt to do so when they wish and with whatever rating they want to assign. But it seemed quite untoward to let the driver also rate the passenger.

Some felt that this put too much of an onus on the passenger, meaning that apparently you would have to bring cookies and flowers to your driver, you would need to listen when the driver blabs incessantly about their day, and you’d have to give a hearty thanks and a pat on the back upon ending the ride, lest you might get dinged in the rider ratings.

Intolerable. Unfathomable. Just plain wrongheaded.

Other than the psychological “damages” of getting a low rating as a passenger, it was pointed out that the ridesharing services didn’t particularly use the passenger ratings per se.

Sure, an astute ridesharing driver that is being summoned by a passenger could take a peek at the rating of that potential rider, and maybe decide to gird themselves for a tough and tortuous ride, possibly even electronically dumping the candidate rider with some excuse like the driver can’t get there in time, yet it was pretty unlikely that there were any serious consequences as a penalty for sour-rated passengers.

It appears the ante is being upped by a recent indication that Uber says it will start banning riders that have “significantly below average” ratings. That’s right, you might get yourself into the ridesharing doghouse and be banned from even being able to seek out rides on the ridesharing network once you’ve been blacklisted.

Once again, there has been some consternation voiced that this kind of ban is not right, and the rider ratings game is getting out-of-hand.

Problems Associated With Ratings Of Riders

Suppose a passenger has unfairly gotten dinged by drivers and ended up with a lesser rating accordingly. Some wonder, suspiciously, might the lower rating be due to something other than a bona fide reason?

Bona fide reasons are presumably things like getting into a ridesharing car and ripping up the upholstery, which is a costly untoward act, given that the drivers usually own their car, their key source of income for ridesharing, and then they have to get repairs made. Plus, the driver is potentially losing money by not taking fares while getting the repairs done. Or, the driver opts to continue driving with the torn seat, which them gets them dinged ratings-wise by passengers for having an unkempt car (thus, lowering the driver’s rating and endangering their livelihood).

A more terrifying reason that drivers might give out a lowered rating to a passenger involves situations whereby the rider threatens the driver. Maybe the rider is a nutty person and prone to violence. Maybe the passenger is a sane person but has had a rotten day and decides to take it out on the driver. Maybe the passenger was triggered by a comment that the driver made, such as when sometimes a driver misguidedly starts to discuss politics, and it gets the passenger riled up since they hold a completely diametric view on the topic.

Those that are worried about the passenger ratings are quick to point out that a driver might intentionally assign a lowered score to a passenger for the wrong reasons, perhaps because the driver didn’t like how the person looked or the clothing they were wearing. Perhaps a passenger got a lower rating because of their political positions, exhibited by what they said to the driver or maybe wearing a button or badge that indicates their political preference.

The passenger ratings might be an amalgamation of discrimination and bigotry, hiding as a seemingly unbiased numeric score for which there would seem to be no ready way to unwrap how it came to be.

Here’s an interesting question to ponder, namely, what might occur with the advent of self-driving driverless autonomous cars as it pertains to the controversial notion about ratings of riders?

Next Up Is Ratings Of Riders While In Autonomous Cars

There are pundits that suggest there will never be any rider ratings anymore once we are past the era of human drivers at the ridesharing wheel and are fully immersed in an era of autonomous cars. Thus, presumably, assuming that those biased or knee-jerk ratings by human drivers are no longer going to occur, there’s no chance of any kind of foul ratings that embed intolerable biases.

I’d say that those pundits are only half-right, at best.

Ratings of ridesharing passengers are undoubtedly going to continue.

In fact, the rider ratings will take on an even greater role than they do today.

Let’s consider the reasons for this.

A human ridesharing driver does more than simply drive a car. In a sense, the human driver is also the civility cop in the vehicle. In theory, the driver wants a passenger to remain calm and not bash or tear-up the car. The driver would also be likely to try and suppress or limit antics such as a passenger that rolls down the car window and yells obscenities at pedestrians on the sidewalk.

In short, generally, the human driver is a kind of human watchkeeper that by their very presence tends to keep passengers from, frankly, going berserk.

A passenger doesn’t even necessarily need to be told to be civil since they know overall that the human driver is a witness to any untoward acts and might either directly try to stop something troublesome or report it to other authorities.

I realize that some of you might right away be complaining that it is nonsense to suggest that ridesharing drivers are the perfection of politeness and civility. I am not saying they are. Instead, I am merely pointing out that as maybe as foulmouthed that a driver might be, they are going to usually be suppressing any way-out-there antics of a passenger (I am also not saying that all passengers are unruly, you need to though acknowledge that some are or can become so).

Shift your gaze into the future when there are self-driving driverless cars whizzing around our streets, highways, and byways. For true Level 5 autonomous cars, there isn’t a human driver in the car.

The driverless car will either have passengers or might be empty (by definition, an empty autonomous car does not have any need to rate a passenger, since there’s isn’t a passenger present).

Would it be useful to rate the passengers in an autonomous car?

Yes, indubitably.

Ways To Deal With Unruly Riders In Self-Driving Cars

Without a human driver in the self-driving car, you’ve seemingly removed the last line of defense toward catching and suppressing ill-advised human behavior of the passengers.

A lawless rider in a self-driving car might decide it’s a great place to mark graffiti, rip-up the seats, spit on the floor mats, and make quite a mess. Such untoward acts don’t have to be focused solely at the interior of the driverless cars. Riders might decide it is a good time to have a paintball fight, rolling down the car windows and shooting paintballs at innocents walking down the street.

As they say, people do the darnedest things.

Especially when they think they aren’t being watched, won’t be detected, and won’t be caught or have to atone for their actions. If this seems like a sad comment about humanity, sorry, it just seems to be the case for some, though thankfully perhaps not all of us.

I assume that you are now of the mind that maybe it does make sense to try and limit how much craziness that passengers in autonomous cars might try to undertake.

Here are the likely solutions to dealing with unruly riders when using a self-driving car:

• Inward pointing cameras will allow the video recording of passengers, similar to how stores mount cameras and warn that you might be caught shoplifting, doing so to warn everyone, even though only a small fraction of patrons is presumably going to do some thievery.

• The AI system can use the camera video to do real-time analysis and possibly detect in-the-moment when passengers shift over into being unruly, and immediately utter via machine-voice a concern to the passenger, which might then prevent further unruliness (versus merely recording video as a means of post-damage evidence).

• Using sophisticated Natural Language Processing (NLP), the AI might do more than just a perfunctory robotic-like warning and carry on a fluent dialogue with the passengers, explaining in a more natural manner what the rider is doing wrong and why they should stop their antics.

 • If needed, the AI system could potentially contact a remote human agent, similar to an OnStar-like service, and notify the agent that there’s something afoul occurring, allowing the agent to then directly interact with the passengers, possibly further convincing the rider to cease-and-desist (note that the remote agent doesn’t necessarily have access to the driving controls, though the agent might be allowed to slow down and halt the self-driving car).

• In a perhaps worst-case scenario, the AI could contact 911 and report what’s going on, presumably even opting to route the driverless car to the nearest police station.

• All or a mixture of these aforementioned aspects are possibilities.

It is quite likely that the fleet owners of self-driving cars will maintain a rating of passengers. This would allow the AI of the autonomous car to either know what to expect from a potential rider, or for the ridesharing network to algorithmically opt to refuse to dispatch a self-driving car to the requester due to a low rating. And, a rider rating serves as an additional heads-up kind of signal to prospective rowdy riders, warning them that their actions inside an autonomous car will become part of their permanent track record as a rider.

To cope with perennially rowdy drivers, some suggest that there might be specially outfitted self-driving cars that are hardened internally to accommodate outlandish acts, which could be dispatched as based on a rider rating that suggests the passenger is likely to be wayward.

Believe it or not, one proposal is that autonomous cars should never allow the car windows to be rolled down, preventing any antics involving tossing things out the window or yelling foul insults at passer-byes, but  it seems somewhat unlikely that the overall ridesharing public would accept the idea of a no roll-down as a result of the oft chance of others misbehaving (the roll-down of a  window could be based permission-based as ascertained via your rider rating).

Conclusion

Overall, the reviled rider rating is not going to disappear after autonomous cars become prevalent. Such ratings will still have merit. The owners of self-driving cars, whether fleet owners or individual owners, will want to protect their investments and will likely grab onto any viable means that will keep their driverless cars primed and ready as money making machines.

There are some added concerns underlying this matter too.

Privacy about the video recordings and AI analyses of your behavior inside a self-driving car is quite another element though that this “solution” raises, stirringly so.

Suppose you opt to take a ridesharing driverless car home after a night of fun at the local bars, and while inside the autonomous car you make a drunken spectacle. Presumably, the video and the AI caught it all. Plus, the self-driving car is likely to push this data up to the cloud database of the automaker or tech firm. Your most embarrassing moments might live forever in a database that you have no means to control.

Another concern is whether the algorithmic approach used to derive the rider rating will be truly unbiased, which doesn’t necessarily need to be the case, and could instead have hidden or other capacities that embed discriminatory mathematical elements. Uses of Machine Learning and Deep Learning can find patterns based on criteria that end-up being a form of profiling, as has already been witnessed in today’s online systems including for financial loan approval and other aspects.

For readers interested further in privacy issues and also the need for algorithmic transparency as a significant societal matter that will embroil the emergence of autonomous cars, I’ve been covering such aspects in my writings and speeches. There is a wave coming, mark my words, and we are right now standing on the beach with a chance to do something before the water crests.

In any case, if you were hoping that rider ratings are on their way out due to the self-driving car tryouts and potential eagerness for driverless car adoption, no dice, sorry to say.



READ NEWS SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.