1328
Tr(rule)am
(midwest.social)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 08 Feb 2024
1328 points (100.0% liked)
196
16552 readers
2015 users here now
Be sure to follow the rule before you head out.
Rule: You must post before you leave.
founded 1 year ago
MODERATORS
Yeah, robo taxis are a dumb idea that should stop being proposed forever. https://youtube.com/watch?v=GcKUYbChE3A
I wanted to search for video that talks about megataxis, gigataxis and MetroVagonMash's gigataxies. You did it first. Thanks.
Yawn, obviously privately owned monopoly would be bad. I can imagine China doing this well as a public utility.
China is capitalism with beaslty grin.
The fact that it's a private monopoly only addresses, like, half of the problems. Why would China do this better? They have just as much incentive to prioritize the rich as Amazon does. Why would they do anything different?
China is ruled by a single party in an authoritarian regime. They have corruption and politicking but they still have remnants of a planned economy and can still make rational decisions for the benefit of their country. For example they have massive projects to build high speed rail and nuclear power.
The US can only make decisions for the benefit of profit maximization. That's overexaggerated of course but you get the gist.
Imagine a whole city converted to public transport, bicycles/quadricycles and robo-taxies to fill the gaps. They could be single seat the size of a velomobile (podbike is an interesting example) and only weigh 100kg and use like 250-500 watt to drive up to 50kmh. Or maybe two seats face to face so you have space to stretch your legs or put your groceries.
How would central planning solve problems like vandalism? And what benefit would this have over bikes and trains?
Autonomous vehicles seem to be literally an unsolvable problem, as covered in depth in the video. What magic would China bring that would make a problem even humans can't solve somehow solvable by AI?
They are not an unsolvable problem. What is your argument for this? And no I'm not watching the whole video lol
I work in computer security. It's just obvious if you have even the slightest awareness of the industry. Attacks on AI are Wiley Coyote shit like drawing circles around them. In an active environment they're even worse. With mountains of technology everyone who has ever tried it, the most advanced and well funded companies in the world, have all failed utterly and miserably. They've failed even though there's an emesne opportunity for profit. At a certain point, you have to start providing evidence that it's possible and there hasn't been any. It's a scam.
But here, I guess I have to do this for you:
https://gprivate.com/69dw4
Oh how funny, how clever! It's SELF EVIDENT they are impossible lol. Didn't think of that killer argument.
Obviously you're wrong because they already exist. They are just not yet good enough. I suspect you're some kind of religious nutjob who thinks there is something supernatural about human brains that computers can just never do lol.
I'm an atheist and I understand how LLMs work. I also helped threat model privacy for the NHTSA vehicle to vehicle communication program, so I have some familiarity with the field and challenges related to parallel technologies.
What I'm not is an AI cultists who can't distinguish between technology and magic. Anyone who's familiar with the field, with AI and how it works, and especially anyone who ever thinks at all about AI/ML security (which, I do, since I have both used ML in my work and reviewed projects that use ML models), recognizes the numerous inherent limitations in the technology.
An LLM replicates human errors by the nature of how they're trained. This is inherent to the technology. LLMs themselves were an incredible advancement that allows all kinds of new things, and yet they're just fundamentally incapable of doing the job in this case. So tell me, what technology do you believe would solve just this one problem inherent to LLMs, ignoring all other problems with sensors and computer vision?
What do you propose?
Or maybe just read something from an industry expert specifically in this field: https://spectrum.ieee.org/self-driving-cars-2662494269
Given the MASSIVE unsolved peoblems, massive amount of money and multiple years wasted already, and potentially infinite amount of money that could be spent solving these problems, what exact problems would be solved by robo taxis that wouldn't be solved, with much less investment, by trains and bikes?
That is basically what I'm thinking. In an ideal world we'd run a "Manhattan project" to create a massive and open research project for self driving AI with different approaches that doesn't rely on profit seeking and short sighted corporations. I still see no rational arguments for stopping to develop and improve self driving cars. There is no theoretical or practical hurdle for it to become near flawless with enough work. Potentially eliminating hundreds of thousands of deaths and injury and also saving a lot of energy and resources.
And yeah most transportation should be trains and bicycles and a better concept for velomobiles, something more like the podbike. But self driving small, cheap and low energy robo-taxis would have incredible value. Most of all it can be shared between people to a much higher degree because it can drive itself to the next pickup without requiring a full time driver or space for one. It doesn't need the range because it can park and recharge autonomously while another vehicle in the fleet replaces it. All large metropolis should ban all cars and only allow robo taxis or special vehicles for delivery etc. With regulation and while avoiding a monopoly. The potential is huge.
But I guess the pseudo-religious hive mind is turning on AI.
You keep saying there's value here, but you can't seem to say what it is. You say there's a solution, but you've just proposed dumping a ton of money in to research with no clear value over existing technologies.
I'm not against AI. I literally said I use AI. Before I used AI, I also believed in self-driving cars. Now that AI isn't magic to me, I understand why this is a fucking stupid idea. People are finally listening to experts who have been saying for years that AI isn't magic. People are turning against the grifters mean "magic" when they say "AI" without having any idea what the technology actually does or is capable of doing.
Here is an alternative Piped link(s):
https://piped.video/watch?v=GcKUYbChE3A
Piped is a privacy-respecting open-source alternative frontend to YouTube.
I'm open-source; check me out at GitHub.