96
Korvax
7y

"Mark Zuckerberg understanding on AI is limited" @elonmusk

Comments
  • 24
    And so is Elons understanding about the stupid hyperloop project
  • 7
    Fucking savage
  • 4
    Definitely agree with Elon, guard rails need to be set in place for machine learning
  • 2
    And a lot of AI experts said the same thing about Elon Musk
  • 5
    @Linux
    1) what about Elon’s understanding of Hyperloop is limited?

    2) why in the world do you think Hyperloop is stupid
  • 15
    @AngryDev
    First of all, the "hyperloop" is not Elons idea, it has been around for over 100 years - Elon just did an "Apple" (took an idea and gave it another name)

    Second.
    The biggest vacuum chamber in the world is less then 1000th of the proposed line between LA and San Fransisco. Meaning that they it will spend awful lot of power creating that vacuum. It takes 3-4 hours to create the vacuum in the current biggest chamber in the world.

    Third.
    Metal expands when it is hot.

    Fourth.
    If something goes wrong somewhere in on the tube, everyone that is traveling and possible thoose outside will be killed due to the force that is generating when all the air is rushing in.

    Fifth.
    The prototypes is crap. Hyperloop one has proven nothing. Just maglev in a steel tube. Maglev has been in use for decades already.

    It is basically spaceflight but in a tube. It is dangerous and stupid.
  • 26
    @Linux
    1) fair but I never said it was his idea

    2) Vacuum is a buzzword. Physically speaking it’s next to impossible to create a perfect vacuum, so the question is “How good of a vacuum do you want?” And for Hyperloop you don’t need as good a vacuum as in a physics experiment

    3) and?

    4) and if something goes wrong on an airplane, everyone dies because there’s not enough oxygen up there or because it crashes. That is, we take risks all the time, it’s up to the engineers to make sure they’re unlikely. Furthermore, Hyperloop can easily be constructed below ground, making the concern about air rushing in and killing people negligible (even if a hole gets punched through the vacuum tube, not that much stuff is going to “rush in”)

    5) as someone who worked on one of those pods I would take offense if that wasn’t true. But then again... the Wright brothers’ original airplane was shit... and planes in general were shit for another 20 years. You have to give it time and not expect this to be done in the next 2 years.

    6) I mean... airplanes are pretty fucking dangerous. People originally also called them stupid... I don’t see that stopping them from being extremely useful and prolific
  • 3
    Elon Musk is an over glorified moron.
  • 1
    @AngryDev watch the vidoe thunderfoot made he explains things a bit there
  • 5
    @AngryDev

    Read the paper, they want 1/1000th of the atmosphere. That is practically vacuum.

    Metal expands during the day, how could they solve that problem? Well, they can construct it below ground. But that is not what the promotional video shows us and that would also
    increase the cost of constructing it, alot.
    You can't compare is to flying an airplane, but you can compare the pressure to flying on an altitude of 60k meters.
    And by saying, "contruct it below ground is easy" what do you mean? really. Digging tunnels?

    Well, the promise that Hyperloop WILL work until the end of the year. They just claimed that. I dont care if you worked on one of thoose pods or not.
    Also, you can't compare it to the wright brothers.

    You dont fly an airplane in space, It is a big difference.
  • 14
    @Linux
    Ok so
    1) 100 Pa is ~1/1000th sea level air pressure but that is nowhere near a vacuum. A vacuum is more on the order of 1/7,600,000,000 that of air pressure. In other words, Elon’s idea has an air pressure over 1,000,000 times greater than that of anything resembling a vacuum
    2) OK after reading the paper, I realize my understanding of their plan was wrong. I thought they were going for below ground, but after reading their paper I realized they plan to build it on pylons with 3 degrees of freedom - in order to account for metal expanding AND earthquakes
    3) I can compare it to flying because in both cases if something big fucks up everyone dies, and the air pressure immediately outside the cabin is too low for breathing
    4) where’d you see that promise? All I can find is the Nevada test track
  • 0
    @Linux oh SHIT got 'em
  • 0
    @Linux When talking about that tunneling project where Musk wants to dig up a few layers of tunnels for traffic, he said that in the future it would be possible to use hyperloop underground.
  • 2
    @AngryDev
    Well, if you multiply a really really really really small number by 1 000 000 does not mean that number is big.
    1/1000th of an atmosphere is 1/1000 of an atmosphere, it is almost the same pressure as the karman line.

    You know that a steel rod that is 30cm long, expand with 0.09 mm when heated? How much would a steel tube that is almost 400 km long expand when it is hot outside? Hundreds of meters.

    Dont forget that Hyperloop would be en excellent target for terrorist or just people with guns who are drunk or on drugs.

    http://jcdecaux.com/mobility-trends...

    yes, the nevada test track will "work" by the end of the year. So I expect that it does not take hours to load a vehicle onto the track and wait for the pressure to drop
  • 7
    On some level I would hope developers were better people than childish kids bickering on YouTube videos... This thread just delivers nothing but disappointment.
  • 5
    @CyberpunkZ3r0 Interesting discussion nonetheless.
  • 3
  • 4
    @Linux
    1) Everything is relative - 80K is considered high temperature for superconductors but I personally wouldn’t consider that anywhere near high temperatures. The point being that a pressure of 100 Pa is relatively easy to maintain and is feasible to believe that it could be done over a 400km stretch

    2) There’s dozens of ways to mitigate that, from versions of steel with low CTE to segmented pipes to NTE materials. But all this aside, if it weren’t possible, how would petrol be able to be pumped thousands of kilometers in a single pipe without the ends shifting by hundreds of meters?

    3) yes it would be, and that’s really the reason I think Hyperloop should be installed underground.. Honestly this is my biggest concern about Hyperloop

    4) I didn’t see anything about the Hyperloop working - just the track.. I think we have a case of media overhyping it here
  • 4
    @CyberpunkZ3r0 This is very different than the squabbling on YouTube videos. It’s a discussion, with both sides bringing up different points backed with logical arguments. Plus I haven’t seen any personal attacks which is refreshing. And, as @anudroid points out, this is an interesting topic.
  • 0
    This turned into fandom real quick. AI running a muck is a real danger regardless of any levels of "expertise" on the subject. To not admit that is foolish
  • 1
    @AngryDev
    1) Everything is relative - 80K is considered high temperature for superconductors but I personally wouldn’t consider that anywhere near high temperatures. The point being that a pressure of 100 Pa is relatively easy to maintain and is feasible to believe that it could be done over a 400km stretch

    Everything is not relative ;) But that is another discussion.
    How exactly are they going to maintain that vacuum ( or low pressure whatever, Elon did call it vacuum chamber himself )?

    2) Actually, good point. I have to look that up and do some research.

    3) If it would be installed underground, the big argument that the price would be low is gone with the wind (or vacuum). I makes the whole project ten our hundreds times more expensive and takes much longer time.

    4) That is probably true there. But also the "first successful test" was just a maglev inside a steeltube, nothing special there either.
  • 3
    Big problem with AI, at least from a hardware perspective, is that the available memory will be massively and irrecoverably fragmented. (lots of small devices)
    When you consider also the massive memory requirements AI will have in order to rival human intelligence, it will be very difficult for AI to allocate memory at a fast-enough rate in order to pose a threat to humans.
    Finally, in order to retrieve that memory efficiently enough such that the information can be useful before a situation changes, AI will need to guarantee very fast serial communications AND that the device which contains the critical information is always functioning.
    Meanwhile, although humans can't recall information as fast as an SOC microcontroller, we are more adaptive to change and each of us can store far more information in a smaller space.
  • 3
    AI would probably use up half of its available (fragmented) memory simply storing addresses and redundant addresses to the actual information.
    Imagine a linked-list where every node contains a lot of pointers to other pointers, and extend that recursively over all the information needed to compete with just one human mind, and each node in that list is not guaranteed to be available to the CPU.
  • 3
    Basically, AI would have a vast and fragile worldwide network of small memory nodes, and a single human mind is a single, albeit MASSIVE contiguous array, with a practically unlimited allocation.
    Considering the demands of memory availability and speedy recollection of information, I think humans have an insurmountable advantage over a dispersed network of AI devices.
  • 3
    1) interestingly, my friend who works for spaceX has told me that Elon’s mouth is the least reliable source of information (right after reddit) since he tends to just call things whatever he wants to. But as to your actual point, if your seals are perfect you don’t actually have to do any work. Since seals are never perfect, it’s a significantly more complicated problem than I can do in my head but in the white paper, Musk mentions industrial vacuum pumps, which operate at nearly maximum efficiency at the chosen tube pressure

    3) True. Although come to think about it, blowing up the tubes would require climbing up the pylons and presumably one could make it so that it’s not possible to do that without being seen.. early warning = disaster mitigation

    4) yeah welcome to PR...
  • 0
    @Linux I completely agree. Ever since human beings learn in middle school about air resistance, they start thinking about putting a mag-lev train in a vacuum chamber. In no way is his idea unique, he's just the first one to stupidly fund it.
  • 1
    Holy shit guys, engineers should know what they're doing. Regarding the 'hundreds of meters' thermal-induced expansion problem, it can be countered with expansion joints.

    Expansion joints prevent objects from expanding 'hundreds of meters' by allowing them to expand between gaps.

    They are used in bridges such as this:
  • 1
    Or even pavements
  • 0
    @AngryDev there's an interesting video done by thunderf00t on YouTube explaining why the hyperloop as it's currently designed can't really work. I don't know enough about the physics involved to definitively say either way, but he sounds very convincing.
  • 0
    @Deserter hey look at that this is basically what I was sayinn
  • 0
    Elon has a point...
Add Comment