Siri's Creator Challenges Texting-While-Driving Study

waderoush writes "A rash of media reports last week, reporting on a study released by the Texas A&M Transportation Institute, implied that using voice-to-text apps like Siri or Vlingo while driving is no safer than manual texting. But Adam Cheyer, the co-inventor of Siri, says journalists took the wrong message from the study, which didn't test Siri or Vlingo in the recommended hands-free, eyes-free mode. In the study, researchers asked subjects to drive a closed course while they held an iPhone or Android phone in one hand, spoke messages into Siri or Vlingo, proofread the messages visually, and pressed buttons to send the messages. Under these conditions, driver response times were delayed by nearly a factor of two, the researchers found. 'Of course your driving performance is going to be degraded if you're reading screens and pushing buttons,' says Cheyer, who joined Apple in 2010 as part of the Siri acquisition and left the company two years later. To study whether voice-to-text apps are really safer than manual texting, he says, the Texas researchers should have tested Siri and Vlingo in car mode, where a Bluetooth headset or speakers are used to minimize visual and manual interaction. 'The study seems to have misunderstood how Siri was designed to be used,' Cheyer says. 'I don't think that there is any evidence that shows that if Siri and other systems are used properly in eyes-free mode, they are 'just as risky as texting.''"

Comments

"proofread the messages visually"

[kruach aum]

Really all you need to know.

one more distraction while driving

[bloodhawk]

I thought it has been long established through research that even a hands free cradle talking on the phone is a dangerous distraction while driving, Can't see how this can be less of a distraction than that even if it is better than manual texting. People have enough accidents without additional distractions.

Re:one more distraction while driving

[prelelat]

Yes and talking to someone in the car is distracting too. So is having kids kicking you in the back of the seat, changing radio stations. Billboards with flashy lights are distracting or they wouldn't have them. Oh I'm sure I'm missing a few more things.

My point is, is that there are a ton of distracting things going on around us as we barrel down the road. The question is, is one more safe than the other. It would be logical that if you can speak to the device instead of type it would be safer. Having your head down and hand off the wheel or if your driving a standard no hand on the wheel or some form of wtf. Having these studies are important for trying to understand how safe something is so we can judge if it's within an acceptable margin. I think texting manually falls into being unsafe and I don't want to share the road with people that are doing it.

It would appear speaking into Siri or other applications that do speech to text hasn't been studied enough to make a final decision, but I think it's going to end up OK. This study is a piece of garbage though and falls into bad research, as the software wasn't used as intended in the car.

Re:one more distraction while driving

[s.petry]

I think the complaint is based on the perception that visual distraction is way worse than any other form of distraction while driving. The problem of course, is that we know that visual processing is not the only inhibitor.

Think of all the times you almost caused an accident when emotionally distressed for example. I can name a dozen easily, where I was so (insert emotional state like angry, excited, sad) where I simply lost focus on the road. Talking while driving may invoke a similar emotional responses, even if a service is translating the speak to text.

That example out of the way, there are numerous other situations where I have been similarly distracted. Thinking about a problem I needed to solve, sometimes coming up with that solution while driving.

So I agree with the person that visual distraction bad, but he neglects the fact that visual distraction is not the only distraction people face while driving.

Re:one more distraction while driving

[icebike]

Yes and talking to someone in the car is distracting too.

Actually, its not that distracting.

Other people in the car are aware of traffic conditions, they actually stop talking, they even point out dangerous situations (even fi from the back seat). Talking to a person in the passenger seat may actually be beneficial to driver safety.

Having a conversation on the phone, that requires concentration, can certainly be distracting, but even the simplest text message is far more distracting. All of the tests of this kind of stuff were done asking people to solve simple math problems or word games on the phone while driving over a challenging course in an unfamiliar vehicle.

Yaking on the bluetooth about nothing in particular while stuck in stop and go traffic simply isn't that dangerous as long as its hands free. The studies suggesting talking on the phone (hands free) is dangerous simply isn't born out by accident statistics. Texting while driving is born out by accident statistics.

Re:one more distraction while driving

[blackraven14250]

It depends on the person talking too. 2 kids, or 2 individuals that have never driven, are likely to not be paying the kind of attention needed to give that feedback, while 2 experienced drivers will.

Re:one more distraction while driving

[AaronW]

Additionally it has been shown that the phone is especially bad since it's a lot harder for your brain to process, especially over a cell phone due to the sound degradation due to all of the audio compression. I don't recall exactly where I heard this, likely on NPR Science Friday or one of the science magazines I subscribe to, but it makes sense. The brain has to do a lot more work to comprehend poor-quality speech than face-to-face speech, and the brain doesn't multitask all that well so it causes a much bigger distraction to driving.

Re:one more distraction while driving

[stephanruby]

This study is a piece of garbage though and falls into bad research, as the software wasn't used as intended in the car.

Before one can say the study is garbage, one has to ask if people really do use Siri (or vlingo) as those applications were designed to be used while driving. Personally, I can not say either way, I couldn't even use Siri (or vlingo, google voice search) while driving even if I wanted to. No application is capable of understanding my accent under any circumstance, let alone while driving.

Whatever happens, please do not force Siri to be used only when the car is stopped. I have a friend who's Prius built-in control panel only works for certain functions when the car is stopped, so she'll slow her car to less than 5 miles an hour on the freeway just so that *her* passenger can type in an address into the gps (thereby creating a dangerous situation where there shouldn't have been one to begin with).

And yes, a Prius is smart enough to detect if a passenger is in the passenger seat, because it does beep until such a passenger fastens his/her seat belt, but apparently, it's not smart enough to enable all the functions of the built-in computer panel while the car moving, even when such a passenger is present.

Re:one more distraction while driving

[fnj]

It doesn't matter how the software was intended to be used, only how the software will be used.

So all progress should cease because no one will use technology properly and any outlier who will use it properly should be penalized by not allowing its use. You don't believe it is possible to educate people to use technology properly and conduct their activities without putting innocent lives at risk. What a ray of sunshine you are.

Re:one more distraction while driving

[SkimTony]

I don't think the study was garbage, but I don't think it was all that helpful. They should've tested the recommended configuration and mode; if that turned out to be safer, then you could use the study to encourage people to change their behaviour. This study a) draws blanket conclusions it wasn't designed to test, and b) doesn't answer the most relevant question, i.e., "Is texting still bad if you don't have to look away from the road?"

Re:one more distraction while driving

[getmerexkramer]

Strawman level 5 complete!

Re: one more distraction while driving

[pchasco]

I call your call of bullshit. I have an iphone 5 with sprint, and I can say that cell phone audio quality is NOT ridiculously good. I do have difficulty understanding people sometimes, and it is certainly distracting. And hands free texting is even more distracting than hands on, in my experience. Because Siri is so often wrong, I must always visually confirm the text. And when it's wrong, it typically takes a great deal of tapping and dragging and typing to replace the words that were in error. That is I do the manual correction after about three attempts at getting Siri to work. Needless to say, I don't use Siri anymore. Google speech recognition was much better in my experience but still not good enough to talk and hit send without looking.

Re:one more distraction while driving

[aevan]

Don't forget the backseat driver: "You're too close! Slow down! Why are you only doing 30? You realise this lane ends right?".
That's a whole different sort of 'feedback' than wanted.

Distraction.

[MrL0G1C]

When you're driving you should be concentrating on driving, that's it, anything else can lead to an accident because your mind is not on the task at hand.

So, no, you shouldn't be pissing about sending texts, if you don't like it, get a bus/train where you can text to your hearts delight.

If you're so f**kin important that you need to text, then get a chauffeur.

Re:Distraction.

[Ironchew]

if you don't like it, get a bus/train where you can text to your hearts delight.

Among numerous other reasons, this is why we need a far more reliable public transportation system (The nearest bus stop where I live is almost 3 miles away and it only gets service once a day). If buses and trains were commonplace, law enforcement could penalize reckless/distracted driving far more harshly and the number of drivers texting while driving would quickly approach zero.

Re:Distraction.

[icebike]

People un-used to city traffic probably DO have to concentrate 100% on driving.

However this is not the norm for most people. You can drive down the freeway in light to moderate traffic and not have much of your conscious brain involved at all. You can arrive at your destination and not recall a single thing about the trip.

In anything but rush hour traffic or high density traffic on a crowded freeway, driving simply isn't that difficult. If it was, we wouldn't hand out driving licenses to anyone with a pulse. Because an awfully large percentage of people just don't have 100% to devote to the task.

There are times when everyone has to pay attention. But the vast majority of my driving, and probably most people's driving, can be managed almost automatically, leaving plenty of time to listen to the radio, or the person on the next seat, or the person on the bluetooth.

Anyone who claims you have to devote 100% of your faculties to driving probably doesn't drive much.

Re:Distraction.

[shawn(at)fsu]

I'll agree that most of a trip you feel like you're on autopilot the problem arises when something unexpected happens. If you have your eyes off the road when that unexpected happens you're a lot worse off than had you been paying attention. So yeah if you get to your destination safely then you can look back and say man that didn't take any conscious effort at all but that's not why you need to pay attention. You need to be on your toes for when something novel or out of the ordinary happens.

Re:Distraction.

[dgatwood]

You are not supposed to be looking out at infinity when you are driving.

Um, yes you are. Infinity, as far as the human eye is concerned, is anything beyond about 20 feet, so except when you're looking in your mirrors (which you really don't need to do constantly, just occasionally), your eyes are almost always going to be at infinity while driving. Anything short of that means you're following too close behind the car in front of you.

The true "ludditism" is the refusal to recognize that the more that distracted driving is studied, the more it is clear that thinking about anything other than driving is a hazard. Distraction by radio is bad enough as it is.

First, that's the opposite of ludditism, which means rejection of technology, not rejection of scientific knowledge. Second, you can have my car radio when you pry it with a crowbar from my cold, dead dashboard. You can't eliminate all distractions, and it is stupid to try.

The goal of any driving-related safety improvements should be to minimize the distraction without being so invasive that people work around whatever solution you put in place. Passing laws against texting causes people to hide their phones while they text, resulting in them looking down even farther from the road, and thus driving even more dangerously. Driving modes that prevent visually reviewing your text messages don't help either, because most people don't want to send out text messages that make them look functionally illiterate, as is often the case with voice dictation under even the best conditions (which a noisy automobile ain't). Using HUDs, by contrast, can dramatically decrease the risk of sending a single text message, mitigating it so much that for the sane 99.9% of drivers who would use such things fairly infrequently, the additional risk of texting in that manner is likely to be lost in the statistical noise.

But of course, the best choice of all is to get the meatbag out from behind the wheel in the first place. That not only eliminates all risk from distractions, but also eliminates all risk from fatigue, illness, sudden cardiac arrest, seizures, sweat getting in your eyes, and probably hundreds of other risk factors.

Pros AND Cons

[ganjadude]

they both exist. I can only speak for myself and people I know who have used voice commands while driving, however EVERYONE, myself included, will speak to their phone for the text, HOWEVER we all double check the msg before hitting send. I think that is where the issue lies. we simply dont trust siri or google voice or other text to type things to be 100% yet. and until that can be true (if it can ever be) it will never be as safe as simply driving and not doing other things.

Re:Pros AND Cons

[The Good Reverend]

Siri reads back the text by default if you're in the eyes-free mode. But her text-to-speech isn't always easy to understand, so it's hard to tell sometimes if she got it right.

"Designed to be used" vs. "actually used"

[Anonymous Coward]

If the study tested Siri the way Siri is normally used, then how Siri was designed to be used is irrelevant.

Driving Performance

[CanHasDIY]

Of course your driving performance is going to be degraded if you're reading screens and pushing buttons,'

See, shit like this is why the Prophet Hicks was so adamant in his belief that advertising people should do the world a favor and kill themselves.

FYI, asshole, it's an issue because humans cannot multitask, and every second you pay attention to that goddamn toy is one more second you're not paying attention to the road.

Perhaps Mr. My-Sales-Figures-Are-More-Important-Than-Your-Safety should read the stacks upon stacks of other studies that prove any distraction from driving is dangerous. Even talking to the guy in the passenger seat.

Re:Driving Performance

[Sique]

Actually, women don't perform any better than men at multitasking. Within both sexes, about 2-5% can really multitask, and everyone else basicly sucks at it. It's just that somehow upbringing and social roles allow women to still try multitasking and be content with the less-than-average productivity and quality.

Just pay attention already.

[Anonymous Coward]

It would appear speaking into Siri or other applications that do speech to text hasn't been studied enough to make a final decision, but I think it's going to end up OK. This study is a piece of garbage though and falls into bad research, as the software wasn't used as intended in the car.

The only valid study would evaluate the software being used as it is typically used, regardless of the manufacturers intent.

Re:Just pay attention already.

[fnj]

The only valid study would evaluate the software being used as it is typically used, regardless of the manufacturers intent.

No, that is complete bullshit. The only test whose results are of any interest at all would be a test which evaluates using the function PROPERLY, not in any onviouslly highly dangerous wrong manner. You could test the safety of some bird brain trying to drive his cars with his knees while woking on a rubik's cube, too, but that would be STUPID.

Texting while driving by typing manually and/or using a screen to verify results is obviously INHERENTLY highly dangerous. On the other hand, using voice exclusively to do the job is no different than talking to somebody inside the car or hands-free talking to a voice on a phone.

Now, the matter of getting some nut behind the wheel to use the proper and safe function in a proper and safe manner is another matter altogether. You could try to remove all driving distractions one by one, by taking out the builtin radio, outlawing the use of any radios or navigation devices or phones by the driver, putting soundproof walls between the driver and all occupants, trying to find some scientific method to suppress sexual thoughts when the driver sees someone attractive outside, or someone in the passenger seat is adjusting their bra, etc, etc - endlessly. I personally favor education instead. I know showing people how to live safely and relying on them to take safety seriously doesn't appeal to all mindsets.

I LIKE your subject line. I wish your message reflected it.

Re:Just pay attention already.

[SkimTony]

When using a study such as this, it's important to state the constraints of the study when explaining your conclusion. The headlines don't read "most commonly used mode of Siri/Vlingo not better than texting while driving." They read "Using Siri/Vlingo no better than texting while driving." If they didn't test the version of Siri designed to be used while driving, then this is an inappropriate conclusion; they didn't test Siri, they only tested a particular configuration (just like secure computing certifications only apply to the configuration tested, and can't be generalized to an entire platform).

I would assert that a more useful study would be to test the recommended modes, because the $64,000 question is whether there exists a mode of operation that is better than plain texting. If so, we can use that data to encourage people to stop using things the most common way, and start doing things in a slightly safer way. Is using Siri to text while driving a good idea? Probably not. Is it a less dangerous idea than looking away from the road to stare at a screen? That's the answer that would be useful, and it's not the answer this study was trying to find.

Re:Just pay attention already.

[zyzko]

It is not complete bullshit. If the software allows you to hand-correct the texts you dictate and vast majority of people do so while driving - that is then how the software is used. In software projects (at least in good ones) there is a testing period where the actual use is monitored, and people can be very creative in using the software in ways the designer did not mean to.

The only question is - what to do about it. In case of an business software used inefficiently the answer is often quite easy - make study why users are not using the features the way designers intended and make them better (better UI, better access to new features, training, etc.) - in the case of Siri and texting - a tougher one. Maybe the only way is to make the speech-to-text so good so you don't have to mind ending up on damnyouautocorrect or your business contact thinking that you are drunk.

Re:Just pay attention already.

[getmerexkramer]

Sorry but if the point of the study is to find out whether using Siri is dangerous, then the test should reflect typical usage. No one's interested in a study that shows whether or not Siri was designed to be dangerous, only whether it is. Your example is stupid because virtually no one drives a car with their knees, if many drove in this manner it would be a perfectly valid thing to test.

too bad studies have proven otherwise

[SuperBanana]

Merely having a conversation with someone impacts your driving; passengers tend to be aware of circumstances like intersections, onramps, cyclists, etc - but people on the other end of your call can't be. It's why Ray Lahood and NHTSA wanted cell phone calls by drivers to end, period. Then there's the issue of control of the car; regardless of whether or not you're "eyes free", if you're holding something in you hand, you're not able to control your vehicle as well as you can with two hands on the wheel. I attended a driving handling clinic (which was insanely fun) where they had you do a slalom course normally, and then do it holding a water bottle to the side of your head; the results speak for themselves.

Re:It doesn't matter how Siri was designed

[icebike]

The research is still valid in the sense that most people probably have no idea about "car mode" and "no-eyes" mode.

Hmmm, seems a little shallow to claim the research is valid when it blames the device for ignorance of the operator.

The real problem is something like 60 or 70% of the people have given up on SIRI all together because it just doesn't work all that well.

Re:They're doing it wrong!

[skywire]

You're new 'round here, aintcha?

Re:They're doing it wrong!

[AaronLS]

" letting Texas A&M Transportation Institute do a study"

What do you mean "letting"?

Are you implying that our government should be in the business of banning universities from conducting experiments and studies?

What does the FCC post have to do with a transportation study?

That post is usually hand picked to be someone that will represent the elected president's agenda. For example, Bush picked Colin Powel's son as his FCC chairman, because of course they wanted hands off regulation, which is a bit ironic because that's what FCC does. Pretty much the Ron Swanson of FCC.

Stop trying so hard. If you squint your eyes hard enough you will see a conspiracy in anything.

Re:It doesn't matter how Siri was designed

[fnj]

Yes; GP's attitude seems to be a massively popular one, as well as an absurd and irrational one. But we already knew people were stupid, and always want to force other people who aren't stupid to live with the results of that stupidity.

Re:They're doing it wrong!

[AK Marc]

given most people don't pay have conversations requiring that level of concentration whlie driving

Why are you assuming that? Is it because you like texting while driving and you want it to be legal because you think you are doing it safely? You sound like the alcoholics in the 1980s. "I haven't crashed, so I must be safe".

Anything distracting reduces safety.

Re:Co-inventor? WTF?

[JasterBobaMereel]

Siri is a spin-out from the SRI International Artificial Intelligence Center, and is an offshoot of the DARPA-funded CALO project

Your government paid for Siri .... why is Apple able to monoplise it ?

Texting while driving is bad... period

[emaname]

Admittedly, I'm an "old guy" so maybe I'm way out of touch with the times, but I'm fairly tech-savvy, well-educated, so FWIW...

I've had 4 very unsettling experiences of near head-on collisions. Each time I saw the other driver look up and get a very astonished look on their face after which they (thankfully) swerved back into their lane.

Meanwhile I was slowing down while maneuvering for safety on the shoulder or sidewalk.

I can only hope that the person who claims texting while driving is NOT a distraction has the same experience, at some point.

As far as talking on a phone is concerned, I have my doubts about that, too.

Again, this is from my personal experience, so YMMV.

I deal with a wide variety of subjects. Some of them are design-oriented. While discussing a subject re the design of something, I find myself visualizing that which I'm attempting to describe. Those are the times I've found myself vulnerable to inattentive driving. For example, I've had some close calls rear-ending other vehicles or missed my turn-off. I DO make a point of getting over to the slow lane and dropping my speed, but I've been surprised by a semi or two that had changed into my lane further up the road in front of me. I missed it because I was... distracted. So I've been guilty, too. (Apparently, something is not happening between my visual cortex and other cognitive functions. Although, my friends from the 60's would probably say... well, never mind. That's for another post.)

Now I hand my phone to my wife and ask her to take the call or exit or pull way off on the shoulder (which isn't all that safe either now-a-days). And when I get a call from someone whose name/cell number I recognize, I ask if they're driving first. I don't want to be the person on the other end of a phone call that contributed to an accident. Besides, I still think most of our phone calls can wait.

Come to think of it, I've even had people walk into me or nearly walk into me in stores while talking/texting on their phones.

Anyhow, please be careful, folks.

Oh yeah... and get off my lawn, kid.

Re:Knees

[idontgno]

It's North Dakota. You could drive 30 random miles on any highway with no steering wheel and no brake pedal with no negative repercussions.

Northern tier highways are the reason people sometimes confuse "cruise control" with "autopilot".