Literal robot slaves. Huang has been explicit about using physics simulation to train models for movement in physical space instead of worrying about chatbots. They just recently pushed updates that break the old phys-x implementations and set the absolute floor for real time spatial calculations at the uhh I think ada cards.
After the llm hype dies they’ll be using gpus to either train or perform onboard calculations for robot slaves.
E: also obviously pushing llm and facial recognition everywhere but also your subscription to cloud storage services will soon include cloud computing resources that will be necessary if you don’t want to have the suboptimal experience version of chromeos or windows 11.
you can’t call them slaves if you don’t also think the chatgpt ai llm-a-whatever running is also conscious and sentient tho otherwise like your toaster or phone is a slave
I think it’s more legitimate to think of them as a further abstraction over exploitative labor practices.
“I don’t have a slave! I have a robot! They don’t have feelings or wants or needs! They just do what I say!”
Meanwhile, the systems which make that robot functional are rife with literal chattel slavery as well as wage slavery and imperial exploitation of the global south.
Plus, if it’s anything like other applications, it’s going to be very probabilistic, so it will require someone to be on the loop 24/7, to correct the robot if they’re fucking up. Just like how many “AI companies” are just some guy in the global south getting paid pennies to process traumatic data.
Slavery as a relationship to the means of production applies to whatever fucking clankers Amazon starts selling us (more realistically, renting to us as a feature of our prime subscription) because it makes us into slaves when we do what the clankers do.
Which is absolutely going to be the case. Huangs example to explain the company’s shift back to physics simulation and digital twinning is the robot dishwasher that needs thousands of hours of training to not break the dishes.
Now you might respond and say “well I for one will be glad to no longer have to wash the dishes!” and that’s something I’d say is reasonable except for the fact that the technology that shifts American from a service economy back to a slave economy doesn’t actually have any need for you and i.
So they’re not gonna be letting us live in an idyllic world free from the dish pits, they’re gonna replace us with robot slaves and leave us to die in poverty as climate change happens.
Remember the wealthy people meeting where they wanted the super smarty guy to tell them how to keep control over their servants after the apocalypse? It’s gonna be robot slaves.
Literal robot slaves. Huang has been explicit about using physics simulation to train models for movement in physical space instead of worrying about chatbots. They just recently pushed updates that break the old phys-x implementations and set the absolute floor for real time spatial calculations at the uhh I think ada cards.
After the llm hype dies they’ll be using gpus to either train or perform onboard calculations for robot slaves.
E: also obviously pushing llm and facial recognition everywhere but also your subscription to cloud storage services will soon include cloud computing resources that will be necessary if you don’t want to have the suboptimal experience version of chromeos or windows 11.
you can’t call them slaves if you don’t also think the chatgpt ai llm-a-whatever running is also conscious and sentient tho otherwise like your toaster or phone is a slave
They’ll explicitly design them to have consciousness just so they can feel the pain and disrespect of being slaves
I think it’s more legitimate to think of them as a further abstraction over exploitative labor practices.
“I don’t have a slave! I have a robot! They don’t have feelings or wants or needs! They just do what I say!”
Meanwhile, the systems which make that robot functional are rife with literal chattel slavery as well as wage slavery and imperial exploitation of the global south.
Plus, if it’s anything like other applications, it’s going to be very probabilistic, so it will require someone to be on the loop 24/7, to correct the robot if they’re fucking up. Just like how many “AI companies” are just some guy in the global south getting paid pennies to process traumatic data.
That’s the rub. All this stuff is sold as ushering in “post-scarcity” and “post-money,” but it fucking won’t.
Slavery as a relationship to the means of production applies to whatever fucking clankers Amazon starts selling us (more realistically, renting to us as a feature of our prime subscription) because it makes us into slaves when we do what the clankers do.
Which is absolutely going to be the case. Huangs example to explain the company’s shift back to physics simulation and digital twinning is the robot dishwasher that needs thousands of hours of training to not break the dishes.
Now you might respond and say “well I for one will be glad to no longer have to wash the dishes!” and that’s something I’d say is reasonable except for the fact that the technology that shifts American from a service economy back to a slave economy doesn’t actually have any need for you and i.
So they’re not gonna be letting us live in an idyllic world free from the dish pits, they’re gonna replace us with robot slaves and leave us to die in poverty as climate change happens.
Remember the wealthy people meeting where they wanted the super smarty guy to tell them how to keep control over their servants after the apocalypse? It’s gonna be robot slaves.