So, boston dynamics has put out another video showcasing the Atlas robots, this time having one manipulate a plank of wood to bridge a gap, go retrieve a bag of tools, carry them up some stairs and across the gap, tossing them up onto a platform to a human "construction worker," turning around. knowcking down large box, jumping down onto that box, doing a "sick flip" from that box to the ground, and then turning around giving a thumbs-up.
Everyone should take a look at the BtS video for this one (https://youtu.be/XPVC4IyRTG8); not because it's particularly bias-bracketed or anything— it's still Boston Dynamics trying sell you their "awesome tech"— but rather because their VERY CAREFUL word choices are quite revealing. …These boston dynamics engineers and programmers are all talking AROUND the idea of whether this system is truly autonomous by using words like "we" did such and such, and "wanted to show," and "future research," and terms of art like "predictive programming." All of this provides a kind of obfuscatory cover so people are wowed by Boston Dynamics' capabilities while still letting BD say that they never really "misled" people as to what they're doing in the original video.
So to be clear: This video is NOT the Atlas system autonomously responding to a completely novel situation with no prompting. It IS the result of a lot of hard work, and that work involves a lot of pre-programming and modelling of EXTREMELY similar "likely" situations, with a lot of fuck-ups in the interim, until they get a whole run right enough, and then that's the video they use.
Boston Dynamics is doing a LOT of research into these areas, but the things they've achieved and are planning to work on are not what most people in the public think they are.
Now, that being said, lots of people who thinking about this in terms of what it's going to do to the value of human labour, and I think that overarching question is a very important one. In a better world, what would come to pass is that the jobs dismantled by automation wouldn't matter because we'd all be getting UBI from the taxes levied against the companies which revenues and profits were increased by, again, dismantling humans' jobs. However, as has been noted, the forces of automation are currently controlled by those who want to both a) not have to pay people to work, let alone to just live, AND b) have those same people somehow still continue to pay into consumer capitalism.
We are, as I've said, looking at a post-WORKER economy, not a post-work one.
And this is without getting into the fact that we're not even ACTUALLY looking at a "post-worker" economy! Most "automated" algorithmic tools are still maintained and supported by humans— just humans paid pennies and exploited for their crucial labor; cf., most recently, ChatGPT and Kenyan workers: https://time.com/6247678/openai-chatgpt-kenya-workers/.
At the end of the day, there are still lots of humans involved in the programming, maintenance, and support of Atlas and other Boston Dynamics stuff, but their labour is often intentionally occulted for a bunch of reasons— chief among them, the prospect of selling more units while paying those humans less.
@Wolven
Certainly the robot was not moving autonomously, but what boss wouldn't want an employee that would do whatever you told it to just like you told them to do it is my big takeaway from that latest video.
@GreenFire Yep, and a lot of humans will get exploited and screwed over in the deal 🤷🏿♂️
@Wolven @GreenFire if the money is right and it satisfies their geeky kink, they will surely contribute to our demise. This is not going to be some sci fi fantasy where someone has the foresight to build a back door to save us from what they intend on wielding against us.We are fools, look through history. Haven't they destroyed the worlds they knew before in pursuit of their perverse desire to rule? Capable and WILLING to destroy on a scale never imagined. We never end it before it starts
Isn't there a Ray Bradbury short story about a AI house making pancakes and emptying trash long after the homeowners are gone extinct?
@Wolven My favorite portions of the video were the robot recklessly tossing the toolbox in an upward parabolic path without seeing the landing zone and pushing a crate off scaffolding to the floor. OHSA would not approve.
With the table saw present, I honestly thought the robot was going to cut the board to spec and deliver to the workman, or perhaps mange the workman a sandwich.
@christopherbrown The board cutting was my first thought too, honestly.
@Wolven If the bag of tools was a squidgy lump of jelly and the plank made of spongecake, it'd be fucked.
All this stuff is done without disclosure to the government or people involved, and no human impact study is ever done.
"...ChatGPT and other generative models are not magic – they rely on massive supply chains of human labor and scraped data, much of which is unattributed and used without consent"
@allanb
That's totally magic if you've read Foucault's Pendulum. The machine learning is the magic. The "feed me" part is also the magic of Wikipedia and Google Search.
@Wolven @LibertyForward1
@Wolven
Getting enough labeled data tends to be the bottleneck of those big neural networks: you need a whole lot of human eyes to feed the algorithm. And everyone is farming that out to Asia or Africa.
The only positive thing is that it creates work, but it's boring badly-paid temporary work.
So even American Mechanical Turkers cost too much these days?
https://www.npr.org/sections/money/2015/01/30/382657657/episode-600-the-people-inside-your-machine
@Wolven I work in a very technical field, and to me, it's very obvious that a lot of human work is needed for robotics, AI, etc. An undergrad trying to implement image analysis on a Raspberry Pi camera will quickly understand all of the difficulties inherent in trying to make a computer do what our brains are good at intrinsically.
But the general public doesn't understand these things at all.
@Wolven “That works really well when the environment is something like what we’re expecting.”
Ah. Teaching to the test, ML style.
@Wolven A friendly reminder how to deal with unfriendly dogbots in case of Black Mirror IRL: https://yiff.life/@Gulfie/109671520435254317
@Wolven Their videos remind me of slapstick silent movies, with carefully choreographed Rube Goldberg-esque scenes.
@Wolven I really wish people would start asking the harder questions.What could the ultimate goal actually be? In ultra simplistic terms, destroy jobs in a greater degree than creating replacement jobs, means destroying consumers too. In a world stressed by population, where for 1% the pursuit of wealth (power over others) has grabbed the pendulum. There will be no swing in the other direction. Come on now, we know they are intended to enforce without conscience, to serve the few, not us.
Actually, this deserves its own point:
ChatGPT used Kenyan workers for content moderation and paid them at a rate of TWO DOLLARS AN HOUR:
https://time.com/6247678/openai-chatgpt-kenya-workers
Like… we all know the mental health harms of content moderation processes, right? So to put that level of harm onto undervalued already-exploited workers is just… morally disgusting.