Beyond brain dead.

alex yates said:

GF's Golf and all her work mates sometimes have them slamming the brakes on for no reason due to the sensors on them giving duff info. Usually when parking so low speed but they've all experienced it doing 30+ mph.

Doesn't half put the wind up you

Click to expand...

Hired an X-Trail (well, was given an X-Trail instead of what I did hire) back last year to go away in. Stupid thing was beeping at me constantly when I first set off in it and it took me a while to go through the menus and turn off all the crappy driving aids. Well, I thought I'd turned them all off, but as it happened I'd left the emergency auto braking assist or whatever it's called on by mistake.

Coming onto a busy roundabout and sneaked out behind the car in front into a bit of a tight gap but nonetheless would have made it fine; but the car in front was a pootling old doddery flat-cap wearer so as I've accelerated I've closed the gap quite rapidly on the car in front. Auto-braking didn't like that and decided it was in my interest that it took over from me, and smacked the brakes on hard, leaving me near getting broadsided by the car coming around the roundabout.

Of course the car coming around the roundabout wouldn't have expected me to have smacked the brakes on in the middle of the exit route, so was timing to go around just behind me into the space I would have cleared, had the thing not have braked for me. Cue a swerve and an angry toot on the horn.

Hated that thing with a passion - how anyone spends near £35k on something like that I'll never know.

As someone who had earned my dough through technology, security being one silo of speciality. Let me assure of this, tech security is a retrospective solution. Vendors can try and anticipate all the ways in which people will try and hack into a system, but there are some clever people out there with some nasty intentions.
Whilst a car has connection to the internet, it is potentially at risk. Thanks but no thanks. Analogue car control for me

arry said:

K888ETH said:

I have driven a Model S for an extended period

Click to expand...

I haven't, so excuse the question if it comes across as uneducated on the subject, if you would please.......

K888ETH said:

Autopilot seemed to work well enough until it decided that the hard shoulder was a better option than lane 1 and executed a violent lane change manoeuvre. I reckon it was a fraction of a second away from spinning out if I hadn't countersteered sharpish.

Click to expand...

So you can't really trust it at all, so it doesn't really work? I mean, if you've got to sit there being completely aware of everything then it's really not assisting you with anything useful, is it?

I think I'd feel more on edge knowing that at any point the car I'm driving might just try to kill me because it's software glitches out than actually controlling the thing myself - surely that's less mentally draining?

Click to expand...

I only tried it the once to be fair, it seemed to be doing a fine job for a good few miles until it decided to swerve into the hard shoulder. That shook me up enough to never activate it again. I assumed there must be an issue with the car, but Tesla claimed it was working as designed and it had just got confused with the road markings. It would have been an interesting discussion/court case if I hadn't intervened and the car was totalled!

I agree it was more disconcerting not knowing if it was going to pull some crazy manoeuvre than retaining control myself. I would take some convincing to try it again and even then it would need to be a very quiet stretch of road in perfect conditions.

Its interesting that people have zero tolerance for error by AI, yet the deaths of around 5 people a day in the UK alone being killed by human error in vehicles are barely newsworthy.

Robertb said:

Its interesting that people have zero tolerance for error by AI, yet the deaths of around 5 people a day in the UK alone being killed by human error in vehicles are barely newsworthy.

Click to expand...

Given the number of vehicles / miles covered in ordinary vehicles, 5 a day isn't a terrible rate. I should imagine it's a somewhat more concerning rate within vehicles with AI that are in antonymous mode at the time of incident but of course that's conjecture on my part as we're unlikely to see those stats emerge for quite some time.

Isnt the real issue how the AI is programmed to respond or select the most acceptable outcome in a emergency situation i.e. does it protect the driver or pedestrians given the choice. That's what worries me most.

I believe this technology is useful but the reliance on it is what we need to question.

arry said:

Robertb said:

Its interesting that people have zero tolerance for error by AI, yet the deaths of around 5 people a day in the UK alone being killed by human error in vehicles are barely newsworthy.

Click to expand...

Given the number of vehicles / miles covered in ordinary vehicles, 5 a day isn't a terrible rate. I should imagine it's a somewhat more concerning rate within vehicles with AI that are in antonymous mode at the time of incident but of course that's conjecture on my part as we're unlikely to see those stats emerge for quite some time.

Click to expand...

Unless you happen to be one of the five! But no, fair point.

I went to a very interesting presentation by the head of AI at Jaguar Land Rover last year... he summarized a number of things as drivers that we deal with easily (for example a polythene bag blowing slowly across a road vs a more solid object), but require a great deal of tech with an autonomous car, least of all knowing where it is exactly and reliably, or getting a proper image of a soft tissue object (human) as properly capable scanners are megabucks. The actual car control is relatively straightforward.

The conclusion was that we have some good and rapidly improving driver assist systems around (and in his view Tesla Autopilot was no more than a top end driver assist package rather than properly self driving), but we are a long, long way from widespread level 5 autonomy.

coullstar said:

Isnt the real issue how the AI is programmed to respond or select the most acceptable outcome in a emergency situation i.e. does it protect the driver or pedestrians given the choice. That's what worries me most.

I believe this technology is useful but the reliance on it is what we need to question.

Click to expand...

On the flip side, AI could avoid the emergency situation in the first place better than a human and make an optimal decision rather than panicking and taking the wrong action.

The real benefit as I see it is the lack of emotion; distraction, over-confidence, red mist, etc.

The problem for the time being is the interaction with humans, which can be relatively unpredictable.

Come on guys in a word he is a TW*T, capable of something brilliant my arxe

I think the whole issue regarding autopilot, is that it's been released years too early. It's relatively unproven & clearly has a few bugs to sort out. I personally think the only time it will be truly safe is when all cars are using it. That and the need to upgrade the roads to allow the cars to operate correctly.

As for the moron in question, he should be banned for a lot longer than that. As for his claims that he was the "unlucky one who got caught"... WTF?? He clearly has no idea just how stupid he actually is!

People like that should be put in more Tesla's & launched into space IMHO.


:nooo:

Robertb said:

coullstar said:

Isnt the real issue how the AI is programmed to respond or select the most acceptable outcome in a emergency situation i.e. does it protect the driver or pedestrians given the choice. That's what worries me most.

I believe this technology is useful but the reliance on it is what we need to question.

Click to expand...

On the flip side, AI could avoid the emergency situation in the first place better than a human and make an optimal decision rather than panicking and taking the wrong action.

The real benefit as I see it is the lack of emotion; distraction, over-confidence, red mist, etc.

The problem for the time being is the interaction with humans, which can be relatively unpredictable.

Click to expand...

I think this is the message being put across in the current Audi advert, it runs to the soundtrack of Send in the Clowns. Quite clever I thought.

He said he was, "Just the unlucky one who got caught."

That comment alone is worth a jail term

:redcard: