Comment on Anthropic Researcher Says Pretty Much All Humans Will Become “Permanently Unemployable” After Its IPO: "It will be able to defeat you, personally, very soon."
You are viewing one comment and its replies. View all comments · View the comment above
Comments
What always gets me is that, like, even if these machines were exactly as intelligent as advertised, most jobs require some degree of physical action unrelated to interacting with a computer. Even an infinitely intelligent computer can’t so much as lift a box off the floor unless it is given some sort of robot or similar device to control. A fleet of robots tooled to do every job humans do in sufficient numbers to replace them if only the software was up to controlling them, does not currently exist. Building them all would take significant time, and likely would require increasing the supply of certain components given the robotics industry isn’t alreadly manufacturing them at the needed scale, and all those new factories and resource extraction take time to set up too. Therefore, there are blatantly obvious bottlenecks to humans all having their jobs automated within just a few months, which have nothing to do with the present state of AI and therefore aren’t changed by an improvement in that state.
This is true, neither Clause or ChatGPT or any other of the AI services they claim will disrupt everything can do anything practical.
But of course work is being done in that area too, so they can (sort of), the most famous being Boston Dynamics, but other companies are working on it too. https://www.youtube.com/watch?v=gRwyIiLWT8Q&t=21s
But those are not really threatening any real jobs, at least not yet.
never mind the fact that building a robot that can do everything a human body can do is a stupid waste of time and effort, and currently impossible besides.
Have you seen Boston Dynamics robot dogs? You think an LLM couldn’t be thrown in one and it just makes decisions on its own? Might not be exactly human but could be close enough to accomplish many tasks poorly for a while until one day it does them well and then we’ll really be living in a different world.
What exactly makes you say, LLMs are taking decisions? From what I understand it plays a probability game of predicting next set of words.
From those words chosen there is ReAct(Reason and Act) framework which selects the best possible next executable task from the choices provided to it.
For all I know, decisions are thought out actions using a logical flow and not predicting the next best token based on the previous token. And for every decision we are held accountable brute forcing our way to make decision is not viable IMO. Would like to hear other perspectives or would like to know if I missed something.
https://www.ibm.com/think/topics/react-agent
This makes me wonder, and I want to see a story of how a more grounded less fantastical AI and robotic uprising story would be like. Like the only impossible science fiction is the omniscient and omnipotent AI that’s trapped in servers or supercomputers, and somehow wants to destroy or enslave humans for whatever reason. Everything else such as the day to day life technology and warfare should be somewhat 5 - 10 years in the future of our current time.
No terminators, no nanobot grey goo situation or anything like that.
Maybe I should try to write this. It’s been 23 years since I last wrote a fictional essay, back in highschool lol. I’m already having some ideas; I should write it down for fun. In a physical notebook, just in case lol
Relevant almost-xkcd
https://what-if.xkcd.com/imgs/a/5/robot_apocalypse_comparison.png
Sign in to comment. Sign in