Is there a better argument against driverless cars?

If a pedestrian pops out from behind a parked car?
Self driving cars today can already detect obstacles behind cars and objects by bouncing radar under them. It would be stoped before you even saw the pedestrians

If somebody walks in front of the car at an intersection and there's no space to swerve? The car would just run into the pedestrian. They already avoid swerving into hazards. A network of self driving cars could accommodate space for the car by moving over and letting the car swerve

computers are nowhere near that level of capability.

I figure the car should kill you if you're leasing it, and kill the pedestrians if you bought it, if it came to that.

*Right now

They already have traction control/ ABS

>Self driving cars today can already detect obstacles behind cars and objects by bouncing radar under them.
I genuinely didn't know that. that's actually really cool.

it's how teslas can currently see two cars ahead of you

oh no?

youtube.com/watch?v=--xITOqlBCM

But can they sense a white truck? XD

do you actually like traction control or abs? for me they are just annoying.

The amount of sensors and sheer computing power will make self driving cars too expensive

>do you actually like traction control or abs?
yes. sure if im gonna be hooning I disable them, but for day to day driving or driving in the rain they can be helpful.

The radar should pick it up yes. Of course modern self driving cars aren't perfect at all. I wouldn't trust a tesla on a tight mountain road.

>I have no argument at all, so I'll just keep moving the goalpost and talk out of my ass

>The amount of sensors and sheer computing power will make self driving cars too expensive
did you finish high school?
There is absolutely nothing complicated about that at this point. There's already university teams doing their own versions of self driving cars

It's not like the rest of the car is already regulated with the ecu

the ecu doesn't drive the fucking car

Sounds like you're an idiot then.

we're talking about autonomous driving on snow here

No but it communicates with ABS/Traction control. It also communicates with self driving systems.

>Says he likes traction control and ABS in the rain
>Literally when they're the least effective and potentially hazardous time to use them
Those are systems that only work on dry warm pavement with tons of grip.
The second peak grip is low, those systems are far too intrusive and cause longer stopping distances and stop you from climbing hill / rounding corners.

youre telling me that an ecu can already drive a car, and if not only needs minor modification?

shit's already standard in most new cars, cost is negligible

crash avoidance systems will soon become standard too, so the same hardware used by self driving cars will be in shitboxes too, once again cost being negligible

what makes you think a computer couldn't handle that
you're making it sound like we need a supercomputer to drive a car. that's completely wrong. it's routinely done with parts off the shelf and uni nerds doing the coding

there is nothing exotic about it. at this point, it is inevitable that self driving cars will outperform humans in the snow and most other conditions. the cost of all that technology gets cheaper every year, the algorithms get faster, and the computers get faster and smarter. Deep AI is getting insanely "smart" pretty much by the month, wait til it's ready to be put behind a steering wheel

Sure, cars can already change lanes on their own, turn at intersections, see hazards, and prevent lane drifting. I would trust a self driving car to drive me around town.

when do you think that will happen?

last year

it's just not ready for average Joe prime time yet

>A group of the world smartest engineers can design a car that can do a loopty loop all by its self
>SELF DRIVING CARS ARE GREAT IN THE SNOW!!!!

A computer will never be able to out logic a human in a random scenario, and a computer will never be able to infer and generate information about a situation when none is available.
The best AI we have can hardly answer a question, let alone compensate for rapidly changing scenarios.

Real question: how do I get my tesla to kill for me.

this

We already have AI that can learn. Put one behind a car on a track and let it fail a billion times before it gets good. Then export this AI to consumer vehichles. Yamaha has a self riding AI that taught itself to ride a motorcycle. They put training wheels on it first so it didn't have to worry about balance and eventually it could balance itself and lean in corners. Now they're trying to get it to beat Rossi. Look at the Atlas robot that taught itself to jump vertically like a human. Eventually it could do backflips.

wow I never thought of it that way

idk, i was alive when a computer was never going to beat a grandmaster at chess. give it a decade, this shit advances exponentially

and fwiw, i oppose automated cars because it will essentially get rid of one our last remaining freedoms

the low light performance is impressive

We already have AI that can beat any player in the world at GO, a game that has more possibilities than there are atoms in the universe. So it's not like it simulated all of the choices and then picked the best one, it chose the best option based on past experience, just like a human being.

put it into one of these hypothetical situations that Veeky Forumstists keep coming up with and greentext what they think will happen instead of what it would actually do in that situation

If it's not learning, it's not AI.

And that type of AI is entirely limited on experience, there's a reason humans are refeered to as "fuzzy logic computers" because we take all of our experience and apply it to every new situation in a split second, and act on it.
Sure, you can brute force an if/then flow by trial and error. But the code becomes massive and unmanageable, and it cannot adapt, it can only react.

can a human control brake, throttle and steering independently on all 4 wheels?

can a human run 24/7 without fatigue or getting distracted?

>implying self driven cars need ai when driving on public roads is mostly constantly repeated simple actions
>implying most drivers on the road care about anything but getting from point a to point b
>implying most human drivers can even grasp the thought of the threshold of grip on their tires

>>A group of the world smartest engineers can design a car that can do a loopty loop all by its self
>ad hominems and lies: the post

Like I said, self driving cars, both in real world and in simulations, are literally in the realm of university students at this point. and here you are HURR DURRR U MUST BE ONE OF DEM DARN GENIUSES TO DO IT

>and a computer will never be able to infer and generate information about a situation when none is available.

and what makes you think a human would be any better at it? and are you implying that a human is going to be able to outperform a computer that is handling multiple sensors at once?

>The best AI we have can hardly answer a question
See: Watson.

>let alone compensate for rapidly changing scenarios.

LMFAO. A computer's response is in the nanoseconds. A human takes literally an eternity to react to rapidly changing scenarions

Is this board full of high school dropouts?

AI of a fixed game like that is using a combination of programmed strats, trial and error, and is playing several dozen moves in advance.
It's a situation a computer is more apt to than a human.
That's very different from a real life situation.

Humans learn with trial and error aswell. It's almost like babies are actively trying to get hurt. It takes a million NO's by the parents before a toddler can build itself some common sense. The only things that we do have built into us is instinct. Nobody is born with mechanical skills.

>he still thinks computer AI is limited to "trial and error" and number crunching

please wiki "Deep learning" and stop pretending to know what you're talking about

also see

ITT: Trump supporters pretending they know about computers

>I tells ya, dey computers be too slow and can you imagine how heavy dey sensors be? no way we could make a car with that many sensors and one of them Cee Pee Usss able to handle them

LMFAO

>Thinks that the students at MIT aren't the top engineers in the world
Okay m8.

>and what makes you think a human would be any better at it? and are you implying that a human is going to be able to outperform a computer that is handling multiple sensors at once?
Because humans are better at it, show me one single prediction program that works better than a human.

>See: Watson.
Absolute garbage and literally takes more storage and computing power than an entire country.

>let alone compensate for rapidly changing scenarios.

>LMFAO. A computer's response is in the nanoseconds. A human takes literally an eternity to react to rapidly changing scenarios
I'm not talking about response times, I'm talking about predictive and new situations.

They have AI that has taught itself to ride motorcycles and walk by itself without any human help aswell.

Deep learning is a meme, it's not real.
It's just a techie 'future' agenda. It's not what you think it is.

theguardian.com/technology/2017/sep/07/new-artificial-intelligence-can-tell-whether-youre-gay-or-straight-from-a-photograph

People have been able to do that for centuries.

Hell people can tell if you're gay within your first syllable.

University of Waterloo
U of T
University of Michigan

just a quick google search btw. it's not only MIT

>Because humans are better at it, show me one single prediction program that works better than a human.

See the Tesla video posted above. The sheer amount of sensors makes them far better at predicting an accident before it happens. Yet here you are pretending that a pair of eyes are superior to radar.

>Absolute garbage and literally takes more storage and computing power than an entire country.

>beats a human bean
>a prototype
>already outperforming doctor at diagnoses"
>"absolute garbage"

>I'm not talking about response times, I'm talking about predictive and new situations.

The simple fact that a computer can do million of calculations per second, only that fact makes them far superior at calculating the best scenario compared to a scared human

ITT: Pic related

Yes that's why AI can learn things without any help

It says literally right in the article humans performed worse in a comparison you fuckstick.

AI can learn, it cannot infer.
People attempt to push AI technology as something it's not because there has been literally no advancement in the last 50 years.

cool, now i'll get to fly off the road and crash because it detected a squirrel

Humans aren't born with the ability to infer either. They learn to infer

It won't drive the car off the road to avoid obstacles

TC and ABS are way, way more primitive than what is describing.

>Absolute garbage and literally takes more storage and computing power than an entire country.
Have you ever heard of ARPANET or Internet? Like you know, sending data via internet, from a server to client stuff...

Yes I understand he's talking about an AI that could react to slippery conditions like a human can. Put one in a car with slippery conditions and let it learn

Good idea! Lets have our cars connected to the internet and require the internet to drive!

The DRM in UBI soft games has proven that a constant internet connection to a remote server to simply play a game isn't intrusive and game breaking at all!

Wrong, babys can infer, it's an innate human ability. Inference makes use of knowledge, it's not learned.

You know those captchas you have to solve to post?
They're building databases for use with AI with it. They are literally too dumb to even look at a picture and know for a fact that it sees a building.

you seem to be stuck 50 years in the past. Go browse through some siggraph papers. Also, self driving tech is 2x further along than what is currently shown to the public, generally. The first company to get it Right will make billions

Babies cannot infer until they are taught to. Inference uses evidence to guess the outcome. AI uses the same thing. Much like a baby, AI needs to be taught what a building looks like before it can determine that what it's looking at is a building. AI can use what it knows to determine the outcome. Of course it's not as fast/ efficient as a human being yet but the fundamentals are all there. We haven't even built a computer that works faster than the human brain yet. This is subject to change though as technology progresses.

A human can see one building from one side and identify any building from any angle after that.

Inference is not taught, and AI cannot truely infer, it can appear to infer because it's learned an inference path, but it cannot organically infer.

I disagree. You think a toddler understands what a building is after seeing one for the first time? It might take few tries

Because they don’t work?

Does a human understand the alphabet after one song? Surely a human should be able to spell something correctly for the rest of it's life after doing it once.

AI can recognize the face on a person even if they have never seen that person before.

TC And ABS won't do shit in snow or Ice, not enough to aid the computer in controlling the car.

>"y-yeah I can totally outperform a computer in unexpected situations"
and keep in mind Tesla's autopilot is just a kinda of "alpha"/test version. if you keep telling yourself this isn't the future then you're literally just as pathetic as the boomers this board makes so much fun of

youtube.com/watch?v=DuIrjRAzNPQ

They do better than your average woman driver

and that's with only 30% of the Tesla sensor suite activated, with less than half rez images being processed

Those cars weren't teslas.

just how dumb are you? That triple beep is the sound Tesla's make when AP detects a hazard

Oh you mean the beeps that were added into the video and not from the car itself? you must be stupid gullible.

Not true. A human can foresee upcoming hazards and can draw on experience with the particular conditions, location and vehicle. Which makes the human able to predict a hazard and the comp to only be able to react to input as it happens. So your "theory" isn't with a shit.

Hell just look at the last clip and listen to the beeps real closely. You can tell the beeps were taken from a different video, it's recycled audio.

this

Some serious denial going on here. It's actually pretty funny

>its not fake if it on the internet

I love what tesla is doing but that last clip is definitely fake. One of the other clips has one too few beeps for it to be real

t. Tesla apologist.

I would give you sources for each of those videos in the compilation, but it's really too much work for someone who won't even change his mind when it comes to Tesla. Some people are just too far down the anti-tesla rabbit hole for anything to have an effect. Plus like a 1/3 are from TMC, 1/3 from YT, and 1/3 from reddit/vk etc. Would take a while

have fun with your ICE shitboxes while they last, duuuude

lol never mind that's a different compilation than the other more popular one. Guess I should have watched it.

yeah, some of those are fake

Or maybe the audio is clearly copy and pasted? You can tell. I like tesla btw

yeah definitely. The cars aren't even all teslas.

sorrrry

I'm still not getting why people are sauing that human drivers HAVE to be banned, as if it's inscribed in stone tablets. Can't the tech evolve to react to unpredictable situations? Also, if more normies adopt the tech, the aggregate skill level increases. I don't see why the two cannot coexist; same goes for ICE and EVs. We still have grandfather laws as well as a strong romanticization of classic cars, especially in America.

I can't see this happening until we have a generation where the majority hasn't driven.

*the aggregate skill level increases of the remaining human drivers

As in, the accidents caused by them will diminish to the point where banning them will likely be too much effort for too little gain

Have fun paying super high premiums for not getting a self driving car

>human drivers HAVE to be banned
no one in this thread has suggested this, what was established is normies hate/suck at driving and that a computer is faster and more efficient at repetitive tasks

driving will not be taken away during your lifetime so long as you live the US, we have the grandfather clause and it's the reason why big scary muscle cars and pre-war cars are allowed on our public streets without cops impounding them for having no catalytic converters/bias ply tires/drum brakes/no speedometers, etc

self driving cars will remove shitty drivers from the possible list of shit on/around the road to fuck up your fun shitbox

exactly! besides the possibility of insurance rates going up, self driving cars will only bring positives for car enthusiasts.

>Driving in the snow is mostly a reactionary task for humans (ie "my tail is skidding, time to adjust it by steering)

Not 100% true. Most driving is proactive, you make a decision before something happens as you can actually predict what would happen.

>Not true. A human can foresee upcoming hazards and can draw on experience with the particular conditions, location and vehicle. Which makes the human able to predict a hazard and the comp to only be able to react to input as it happens. So your "theory" isn't with a shit.
Yep that how it works for more a less skilled driver.

>>"y-yeah I can totally outperform a computer in unexpected situations"

You do, actually computers are still sucks when it comes to real-world situations, like that one

youtube.com/watch?v=lWcpKBHJ3As

Also there are not so much self-driving motorcycles. How come?

no market for them. at all. like at all.

are you 14? serious question