U.S. tells Google computers can qualify as drivers

http://www.reuters.com/article/us-alphabet-autos-selfdriving-exclusive-idUSKCN0VJ00H

by apo • 10 years ago

454 184 10 years ago

Flammy2• 10 years ago

Driver ID J039532 'Joe', a set of learning algorithms, has been licensed to drive multiple vehicles at once from both in vehicle computers and 'his' data-center based 'home'.

Out of the 1000s of instances of Joe running around, in the wild, a handful have gotten into situations where he has had points penalized against his license following existing state law. Joe exceeds his points limit and his license is suspended after 2 days.

... This could be interesting.

rashkov• 10 years ago

Is this the beginning of algorithmic personhood, then?

Flammy2• 10 years ago

Well if company can be people, and companies can copyright/patent/own/protect algorithms they've developed, I guess it makes sense an associative properties sort of way that algorithms can be people...

oh_teh_meows• 10 years ago

Not in the traditional individual sense, I think. All AI drivers across the whole world will seamlessly exchange and synchronize with each other their learning experiences, resulting in self-driving vehicles that improve at an exponential rate.

It's basically an AI collective, whose 'brain' scatters across data centers around the world.

jtolmar3• 10 years ago

Algorithmic personhood seems like a pretty good path forward for all the inevitable legal questions this will stir up. You could prosecute the individual instance of the algorithm when it breaks a law, and if that's successful charge its employer for something like negligent training or negligent hiring.

peter_wiggin• 10 years ago

Algorithm's won't be persons till they can anonymously fund super pacs. ^_-

raddad• 10 years ago

IBM's Watson for president 2016 http://watson2016.com/

nxzero3• 10 years ago

Algorithmic personhood already happened via corporations.

dragonwriter6• 10 years ago

Probably not. It's an interpretation that the word "driver" in one particular legal context does not necessarily mean a person, not a decision that computer programs are people.

This simply seems to be (without the text of the letter, just several news reports) an interpretation of law which is a basis for NHTSA to subsequently rewrite vehicle safety equipment requirements to support fully autonomous vehicles at the federal level.

Because of the way legal definitions are context-specific, it doesn't necessarily mean anything to anynof the other laws regarding drivers, much less anything outside of that domain.

yanose• 10 years ago

Would owning "Joe" make me a slave holder?

leeoniya• 10 years ago

wonder if i can swap out my driver mid-ride depending on neighborhood, weather conditions or my mood...

derefr2• 10 years ago

You already sort of can; 'sport mode' in most cars is effectively you, as the car's puppeteer, changing out the one persona for another of the AI "puppet" your commands execute through.

tzm• 10 years ago

And Joe is owned by a corporation, which is considered a person, my friend.

https://en.wikipedia.org/wiki/Corporate_personhood

ascorbic• 10 years ago

Can an algorithm have mens rea?

ConbrastanJorji• 10 years ago

One might choose to define mens rea as the difference between a bug and a feature.

pcurve2• 10 years ago

I'm curious to see if some vehicles AI develop road-rage over a course of period as it goes through self optimization.

nxzero3• 10 years ago

Going one step further, some AI will be designed to be aggressive; for example, law enforcement, military, security escorts, etc. -- I'm guessing that will it is very likely that you're right that some consumer AI will be aggressive, if it become too common, it'll be shut down.

davnicwil• 10 years ago

In situations which usually induce anger in people on the roads, it seems more likely that any resulting behaviours resembling road rage only disadvantage the driver, if the goal is safety.

So if anything, wouldn't any sort of sufficiently global optimisation only veer the ai away from behaviours resembling road rage over time?

6stringmerc• 10 years ago

While I get the humor in your comment, I'd be curious if 'relaxing' some of the behavior parameters would result in some not-very-useful conclusions from testing:

"Tests indicate most efficient driving technique to include lots of honking, flashing high beams, and spraying winshield wiper fluid onto cars following too close on the highway."

anarchy8• 10 years ago

Wow, I'm surprised the legal world is catching up to this so quickly. You would think that the law would be a larger barrier than it has been so far.

wavefunction• 10 years ago

There's a lot of money involved, and that always greases the skids.

tyingq3• 10 years ago

Also worth mentioning that Hoover might be posthumously jealous regarding the amount of dirt Google might have on, well...everybody.

thrownaway2424• 10 years ago

This is progress, but not much. Driving, licensing of drivers, and regulation of auto insurance is almost wholly dictated at the state level. What the federal government considers to be "a driver" is nearly irrelevant.

ajdlinux• 10 years ago

It's relevant for the purposes of federally-mandated vehicle safety standards, is it not?

dclowd9901• 10 years ago

I'm sorry but you're quite wrong. We still have federal regulations that disallow smart directional dimming headlights because the regulation dictates high beams need be activated by a binary switch.

The biggest, slowest sticklers to this progress thing are at the federal level. This is a huge deal.

swalsh• 10 years ago

To be fair, that's probably a good thing. Every state has slightly different difficulties that the AI will probably have to learn to overcome. Though it might slow down technological progress, it's probably a safer way to roll out.

rdlecler1• 10 years ago

"It noted existing regulations requiring some auto safety equipment can not be waived immediately, including requirements for braking systems activated by foot control."

I guess you just need a little motorized foot that can control a little brake.

sleet• 10 years ago

Reminds me of how Thomas Peterffy's engineering team skirted the rule that all trading orders had to be entered through the keyboard. They built a robot with rubber fingers that typed entries into the keyboard.

And thus high-speed trading was born.

ghayes• 10 years ago

Couldn't he have built a keyboard with high quality macros or multi-letter keys, too?

thaumasiotes2• 10 years ago

No, just because the car is required to include foot-operable brakes doesn't mean you're required to brake by application of a foot. You can brake however you want as long as someone in the car could activate a brake with their foot.

seanp2k2• 10 years ago

This. Consider how currently-deployed active stability controls, traction control systems, or even ABS operate: by operating the brakes beyond the input the driver is directly providing.

That said, dear Google, please don't take away my foot brake. I'd really rather not have the algorithm decide that the lives of the occupants of my vehicle are worth less than %{UPCOMING_SITUATION} and kill everyone without the occupants having control.

ashwinl• 10 years ago

The specific letter from NHTSA to Google:

http://isearch.nhtsa.gov/files/Google%20--%20compiled%20resp...

... along with an abbreviation I have yet to see previously:

> self-driving vehicles (SDVs)

... and their definition of a Level 4 vehicle:

> Google describes its vehicles as having what NHTSA's May 2013 Preliminary Statement of Policy Concerning Automated Vehicles calls Level 4 Full Self-Driving Automation. According to that Statement, a Level 4 vehicle:
> is designed to perform all safety-critical driving functions and monitor roadway conditions for an entire trip. Such a design anticipates that the driver will provide destination or navigation input, but is not expected to be available for control at any time during the trip. This includes both occupied and unoccupied vehicles. By design, safe operation rests solely on the automated vehicle system.

taralx• 10 years ago

In case people are wondering:

"SDV" has been in use since at least 2011: https://scholar.google.com/scholar?cluster=10460039591121716...

The level classifications appear to date from 2013: http://www.nhtsa.gov/About+NHTSA/Press+Releases/U.S.+Departm...

Carrok2• 10 years ago

In case more people are wondering still:

"SDV" is not an acronym but an initialism. TIL that in today's thread about TIL: https://news.ycombinator.com/item?id=11069206

ocdtrekkie12• 10 years ago

Given that statistically, Google's self-driving cars are more dangerous than the average human, and if Google's computers were a driver, it's license would already be revoked, it's a little weird the government is willing to take this step.

Then again, they've contributed to about 200 members of Congress, several senior White House positions are filled by Googlers...

bnchrch• 10 years ago

I'm all for opinions in opposition of the norm but if you could provide references to your claims that would be much appreciated.

ocdtrekkie12• 10 years ago

http://fortune.com/2016/01/13/google-self-driving-car-accide... Ten of these would've been the Google car's fault. Given the mileage they've covered in 2015, and the number of accidents, they're terrible drivers statistically... to the point that a human is the safety device.

http://www.opensecrets.org/orgs/toprecips.php?id=D000022008&... Congress, 2014 term. Many of the top recipients actually wrote letters on Congress letterhead begging the EU not to rule against Google, which is a fairly unprecedented action.

http://blogs.wsj.com/digits/2014/09/04/two-googlers-take-top... http://www.govexec.com/management/2014/09/white-houses-roste... Googlers take up some of the top technology positions in the White House, including "Chief Technology Officer". Eric Schmidt and other high ranking Googlers also visit the White House nearly weekly.

Think they behaved themselves? Nope: http://techpresident.com/blog-entry/white-house-deputy-cto-a...

In case you're curious how Google's using their White House positions: It's to get money. http://thehill.com/policy/transportation/265932-obama-pledge...

And some more of our taxpayer dollars. While this article carefully avoids mentioning Chromebooks, they're now the leading purchased portable computers in schools, they're pictured in the header of this blog, and it's written by their "former" Googler, who is the CTO: https://www.whitehouse.gov/blog/2016/01/30/computer-science-...

That's a lot of strong government ties, but are they worried about regulation? Well... for the longest time they controlled the US's regulatory body too:

http://www.buzzfeed.com/williamalden/ftc-commissioner-appear... The FTC was basically run by a paid Google shill, Joshua Wright, for the last couple of years. (The FTC reopened their investigation against Google within a month of him stepping down last year.) His pre-FTC behavior wasn't much on the ethical manner either: He was paid as a university professor to write "academic studies" supporting Google's legal views.

One would wonder by the United States has been one of the slowest and most hesitant countries to regulate and pursue legal action against Google, until one does their research. Then it stops being much of a surprise at all.

bsder3• 10 years ago

Given that statistically, Google's self-driving cars are more dangerous than the average human,

Come to Southern California when it rains and lets see you say that.

The difference is that autonomous safety will continuously improve while human safety has actually been going backwards.

ocdtrekkie12• 10 years ago

I live in Chicago. Get on my level. ;)

Seriously though, you're also missing the fact that AFAIK, Google doesn't/can't drive these cars in rain. According to a Slate article in 2014, they also can't drive when it's too sunny, because it messes with the sensors. It goes without saying it's never seriously been tested in snow. From my understanding, self-driving cars are mostly tested during the day in fair weather, when humans are also at their safest as drivers. http://www.slate.com/articles/technology/technology/2014/10/...

Also, cars are getting safer, fatalities from cars is actually going down. I managed to find a version of this article that is less blatantly liberal propaganda, because I want to talk about the auto statistic: http://www.lowellsun.com/opinion/ci_29325302/cars-getting-sa...

cJ0th2• 10 years ago

Until some kid finds out how to hack and hijack these cars just to enjoy a game of real life GTA. People are that stupid, look at that guy who threw a live alligator through a restaurant’s drive-through window for a laugh.

https://www.washingtonpost.com/news/post-nation/wp/2016/02/0...

autoreleasepool• 10 years ago

This is refreshingly progressive news

shadytrees• 10 years ago

and to think it only took an inordinate amount of lobbying by corporate interests

Animats• 10 years ago

This isn't a big deal. The NHTSA defined levels of autonomous vehicles back in 2013 They are roughly:

Level 0: manual.

Level 1: some automation, maybe radar cruse control. (available now)

Level 2: smart automatic cruise control plus lane keeping (Available on several high-end cars now. Tesla is at this level, not Level 3)

Level 3: automatic driving good enough to handle ordinary driving tasks and route planning. (Google, Cadillac/CMU have this in test) but driver sometimes has to take over manually.

Level 4: full auto, all road conditions, driver not needed. (nobody really has this yet)

Now, the NHTSA is discussing modifying the federal safety standards so that, when Level 4 is achieved, vehicles which achieve will comply with federal motor vehicle standards.

Whether to allow autonomous vehicles on roads is a state matter. California DMV currently allows this in test, with manufacturer test drivers only, and requires reporting of incidents. Current DMV thinking is to stay with that for three years, see how things are going, then reevaluate. There's some whining about this from Google, but realistically, Google doesn't have the technology to go beyond that yet.

The accident reports are on line.[1] Almost all of them are from Google, and most of them involve someone rear-ending a Google car which was driving cautiously. Except for one incident.

Cruise (YC W14) had a crash with a parked car last month, driving on 7th St. in San Francisco.[2] The report indicates that the vehicle swerved to the left under automatic control, then corrected to the right, then the driver took over manual control, and then hit a parked car at 20MPH. Both vehicles damaged, no injuries. The reported location [3] is across from the SFPD's Southern Station and a popular parking place for police vehicles. They hit a parked Toyota Prius.

I've written before about the "deadly valley": automatic driving good enough that the driver stops paying close attention, but not good enough to drive reliably. Cruise just demonstrated this. You cannot rely on the human driver to suddenly take over from the automatic driving system.

[1] https://www.dmv.ca.gov/portal/dmv/detail/vr/autonomous/auton... [2] https://www.dmv.ca.gov/portal/wcm/connect/bc21ef62-6e7c-4049... [3] https://goo.gl/maps/DoNdYSV8cgq

jayjay71• 10 years ago

Cadillac/CMU actually spun out into a company called Ottomatika which was acquired by Delphi.

andy_ppp3• 10 years ago

Is each car going to have to pass a driving test - how is the inspector meant to know if a computer is checking it's mirrors and indicating correctly!? ;-)

dragonwriter6• 10 years ago

This seems to have to do with federal law which underlies NHTSA's regulation on required safety equipment, not laws regarding licensing, etc. Without something like this, regardless of state laws on autonomous vehicles, all US street-legal vehicles, even if fully autonomous, and even if the state had no requirements for human control as an option or backup, would have to be equipped for human control.

Who state law governing things like driver responsibility would treat as the driver of a fully autonomous vehicle -- and whether they would allow such a thing at all -- is a separate issue.

Klasiaster• 10 years ago

So and if these devices are hacked to injury people nobody can prevent it or is even forced to take responsibility as it's the fault of the device and not those of the owners/creators?

jonknee2• 10 years ago

That can absolutely happen today and we're not freaking out about it. The bigger concern are the actual hundreds of thousands of people killed and maimed every year by the fault of other humans behind the wheel, not hypothetical evil geniuses.

kriro• 10 years ago

Issuing and revoking AI licenses (driver's licenses or others) seems like a very good use case for the block chain. You can predefine rules for removal etc. There's a bit of literature about automatically executed contracts on the block chain and this use case seems very similar.

jjangsangy• 10 years ago

Good to see progress being made on this front! I guess this means that makes the scores

US Gov: +1 Google: +1 Tesla: +1 Uber : -1 Robots: 0xffffffff

dtparr• 10 years ago

Is your robots' score signed or unsigned?

rawTruthHurts• 10 years ago

Looks like an open door for all kind of legal tricks for companies, the kind of trick corporate personhood achieved.

khgvljhkb• 10 years ago

We who browse HN like the idea of self-driving cars, but does anyone have a nuanced perspective on their usefulness? I don't know much about AI compared to many here. It has been said that things which humans learn at very young age are hard to automate (identifying things you see, walking, social things) while things we learn as adults are easy to automate (accounting, telegraphs, file cabinets).

Car driving definitely contains elements that we learn as kids (seeing things in motion, identifying what is a human and what is a rock) while some things are learned as adults (what does that sign mean, how to count KM/H, how transmission works).

Are we really close to autonomous vehicles on the streets?

thomasahle• 10 years ago

For one thing, the truck market seems incredibly ripe for automation. That should be a fairly easy, profit driven decision. Not many feelings involved.

Except for protests from truck drivers of course.

chrischen• 10 years ago

Pay off the current generation of truck drivers and tell the next generation to not pursue a career in truck driving. That ways we won't waste millions of dollars arguing with them when what they really want is to be paid (but they'll make up bullshit reasons for about how automated truck driving is bad).

omegaham• 10 years ago

Close as in widespread deployment, coming soon to a parking lot near you? No.

Close as in "doing proofs of concept and being adopted for experimental tasks?" Yes.

lazaroclapp2• 10 years ago

When Asimov wrote about robots, he imagined them being fully sentient way before getting any legal rights (see e.g. "Evidence"). I don't think before this I would have ever bet on the process going the other way around...

andy_ppp3• 10 years ago

Companies have more rights than human beings now and have done for a long time; I'd suggest that robots also will be considered more important for the whole of society and will therefore get considerably more rights than you might at first expect.

Adam Curtis' documentary becomes ever more relevant:

https://en.wikipedia.org/wiki/All_Watched_Over_by_Machines_o...

true_religion• 10 years ago

What rights do companies have that human beings do not?

rdxm2• 10 years ago

if you spend any real amount of time on a bike interacting with traffic you quickly come to realize that 99.5% of the humans behind the wheel shouldn't be driving, so it's hard to see where there would be any noticeable difference in quality of operations between human and autonomous.......

dexterdog• 10 years ago

You get close to the same number by spending any amount of time behind the wheel as well.

ams61102• 10 years ago

By the same token many many cyclists should get their own heads out of their asses and obey the traffic laws as they are supposed to do.

marincounty• 10 years ago

Out of all my years driving, I've never been really concerned over a bicyclist, even when they go through stop signs, or hang on my lumber rack. Yes, a lot don't obey traffic laws to a tee, but it's a bike. They do not have the mass, nor size of a automobile. They just haven't been a concern for myself.

I honestly don't get the outrage towards bicyclists. I think it's more psychological? Personally, I have found the most out of shape individuals complain about cyclists the most.

rdxm2• 10 years ago

sigh....sure there's a "10%" on the bike side. but the scale of that "10%" is infinitesimally small in comparison to the level of incompetence in motor vehicles, and moreover a bike isn't an (n) thousand pound kinetic energy weapon.

this is a tired argument that lacks any basis in fact. what is supported by fact is that we annually kill tens of thousands of people each year in this country (USA) as a result of driver incompetence, i.e. distracted, drunk, etc, etc.......

mikeash6• 10 years ago

I've never had a cyclist blow a red light and nearly run me down in a crosswalk. I've never had a cyclist cross a double yellow and nearly hit me head on because they can't stay within the lines. I've never had a cyclist merge into me and nearly flatten me because they apparently think yielding to traffic already on the road is optional.

Sure, cyclists violate the law, but let's get our priorities straight.

jonknee2• 10 years ago

it's hard to see where there would be any noticeable difference in quality of operations between human and autonomous.......

It will be very noticeable, the autonomous vehicles will be at least an order of magnitude better!

mikeash6• 10 years ago

It's interesting to read the accident reports for autonomous vehicle testing. Most of them go something like, "The autonomous vehicle stopped when legally required, and the human-driven car behind it didn't notice and rammed the autonomous vehicle."

People come up with all sorts of crazy failure modes for autonomous vehicles. What if the sensors fail open at the same time another car is speeding and the sun is at a 37 degree angle and a family of ducks is crossing the road.... Meanwhile the current gold standard can't be counted on to stop when the vehicle in front stops.

jhbadger• 10 years ago

"It own itself?" "Swiss citizen, but T-A own the basic software and the mainframe." "That's a good one," the construct said. "Like, I own your brain and what you know, but your thoughts have Swiss citizenship. Sure. Lotsa luck, AI." -- From William Gibson's "Neuromancer"

mrharrison• 10 years ago

I've been following this closely and think we are headed to self driving cars a lot sooner than we think. Obama has been making sure the regulators have been getting the money, GM is working with Lyft to have self driving cars in Austin. http://popsnip.com/topic/882/Selfdriving-cars-race-towards-o...

conductr• 10 years ago

I agree. I commonly have the debate with friends who don't follow it as much. It's widely known that Google, Uber, etc want this to happen. However, the general perception that I hear is we are still 10-20 years away. My argument is ~2020 is more realistic.

Most people I talk to don't realize Google has been working on this for a decade (or more) already and the tech is actually really good. Also, safety is a common point many people mention. When I mention how unsafe human drivers are, it's interesting to see how many people never consider that angle.

I'm hoping my current car is the last one that I own. That would require the whole self-driving-Uber thing to come to fruition within next 5 years, but I'm hopeful.

U.S. tells Google computers can qualify as drivers

Related Stories