"Eliza, why are you so against self-driving cars, when you yourself are a tech industry veteran?"
You see, it is *because* I am a tech industry veteran that I fear them so much.
And if you could understand how most things were coded, you would never trust anything a programmer touched ever again.
@Elizafox "but it's a feature"
@Elizafox a car than elon can't personally fuck with remotely? unacceptable
@CobaltVelvet "the passenger was killed to protect the car"
Something I am sure some AI will be programmed to do somewhere, what with leased cars and the fact that with a Tesla, you just own a *license* to use the car, you don't actually own it.
@Elizafox Working in auto, it's equally terrifying. I don't work with safety critical stuff, but my take on the overall engineering attitude is "this works well enough" and I don't think the sensor systems can be trusted in all weather conditions in the north
@Elizafox just pack a large collection of viruses together into the same ECU and make the cars share the viruses with eachother in real time.
@Elizafox cars should come with immune systems to fend off hackers
@Elizafox cars should not be programmed. they should be home-grown (or lab-grown, sometimes).
@Elizafox now I'm thinking of some code with the ability to have sexual intercourse with another code...
I keep telling anyone that will listen (IRL) that no, AI has not gotten that much better recently, companies have just gotten better at marketing it
@Elizafox *writes a shit python script that fucks up* it's not Us, it's the Advanced AI
@Elizafox i have found my use in society; illustrate good points with shitpost
@Elizafox the only valid AI is a large collection of viruses being runned at random and seeing what happens.
@Elizafox I will never in my life trust a car I can't fix.
I'm pretty bad at fixing cars so that means I'm pretty much limited to like, japanese trucks from the 90s
@Elizafox I think self-driving cars can be made to be safe. I agree it's dangerous and dishonest to claim AI is this amazing intelligent thing, because it's not. But with all due respect, I believe they're safe as long as we sufficiently test and regulate them. ultimately, automation can be safer than humans by removing human error from the equation. Consider surgeries like lasix which are completely computer-controlled but very safe.
@cheryllium @Elizafox there are many, many fewer variables in a surgical field than in the world where people drive cars, and surgeries also include extremely thoroughly trained and certified people who are ready to intervene
automation removes *some* human error from the equation by replacing it with machine error which can in theory be minimized, and in a world where perverse incentives don’t exist or are minimized, that can be fine. conversely: boeing and the faa.
@Elizafox all of the money that goes into making these shitty DRM cars really ought to be going to expanding public transportation
@Elizafox *fix this never
;p which is part of the problem...
@Elizafox "As long as everyone drives exactly like me, and has my exact life as a silicon valley tech bro, it'll be fine. And clearly they do! SHIP IT."
@Elizafox there's something even worse tho
@alexis @Elizafox It's already proven that self driving cars are way more reliable than human drivers, even if that still leads to deaths. And not only they can do that on the current environment, they can also plan ahead and cooperate together in ways humans would never be able to. This does lead to less likely, but also bigger ways to mess up in case of network-level exploits. Overall tho, it'll very likely be an improvement, even with very severe failures.
Yes, humans are THAT bad at driving
granted that humans suck at driving, but i’ve seen software engineers, and i do not trust them to produce robots that suck even equally at driving than humans do, to say nothing of sucking less.
@alexis @Elizafox Companies like Amazon and Uber are already using self driving trucks in the wild, and have been for a year or so. Waymo already provides a commercial, non-test pickup service in Phoenix. There's automated buses in China, which should be arriving to Japan as well very soon. Tesla's self-driving tech is already bundled in each one of their cars and properly tested in real roads, waiting to be legalized, but can be already seen in action with Autopilot. So on.
by “large scale” i mean, not demos, but production use at a scale that isn’t a rounding error in the number of road miles per year driven by humans
it’s going to be a minute before that happens, and i really hope my expectations are wildly pessimistic, but my point is we can’t know
@Elizafox But I am concerned about it from the data collection + privacy angle. Especially if the cars are internet-connected and broadcasting our location at every point.. .
@Elizafox Exactly this. The idea of a self-driving car, in and of itself, is not terrifying; it's desirable even. But, knowing what goes into making these things a reality today (and who or what is funding the R&D) gives immediate cause for concern. Especially in light of the Jeep that was remotely disabled from across the continent, by way of its ... wait for it ... /entertainment console/.
If remote infiltration isn't a high concern, on-prem exploits still are. https://www.wired.com/2016/08/jeep-hackers-return-high-speed-steering-acceleration-hacks/
@Elizafox Imagine an irate customer pissing off an already disgruntled mechanic with less than honorable connections. Kiss control over your brakes, accelerator, and steering goodbye.
@Elizafox A-fuckin'-men to everything you said.
A Fellow Industry Veteran Coder
This is a Mastodon instance run by the Interlinked Foundation, a 501(c)(3) non-profit devoted to eliminating discrimination. We are an instance that blocks authoritarian political violence, ultra-nationalism, fascism, the alt-right, Stalinism, and authoritarian ideology in general. It's intended to be a safe place for those tired of violent rhetoric as well as a place safe from discrimination.