Users rejected the idea of coordinating multiple LBJs for success as a fundamental category error in thinking about AI systems.
Based on 2 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@JeffLadish @sebkrier @AdrienLE you happen to have picked someone about whom I’ve read thousands of pages, and so I’d actually say no, the notion of replicating LBJ doesn’t seem like it would help in terms of moving things through the world
@JeffLadish @sebkrier @AdrienLE “1000 LBJs” just feels like a complete category error in thinking about the world to me
I’m not sure what you mean by that. I chose him because I thought you’d be familiar! My knowledge is pretty limited to Master of the Senate, but if you think the info there is roughly correct we should be able to speak from some common ground. Re helping move things through the world, I wasn’t arguing it would help anything in a normative way. I was talking about skill! As in, if LBJ was alive today and super motivated to get high speed rail accomplished in California, he would be much more effective than most people in accomplishing that than the large majority of people in similar positions. Now would 10 LBJs be better than 1? I’m not confident about how would such people could and would work together towards common goals. Perhaps you could speak to that. I think thinking through this kind of thing matters because I think how AIs wield political power could matter a lot.
Maybe a question which would cut closer to the heart of it: do you think AIs will come to be better at politics than humans? I think you might quibble at the word politics. I mean the relationships, the coalition building, the speaking and letter writing, the getting specific laws passed and specific allies into specific positions, the commanding of loyalty, etc. Do you think a group of GPTs or Claudes or their successors will be able to do this thing better than the best humans in the future?
@JeffLadish @sebkrier @AdrienLE you happen to have picked someone about whom I’ve read thousands of pages, and so I’d actually say no, the notion of replicating LBJ doesn’t seem like it would help in terms of moving things through the world
@JeffLadish @sebkrier @AdrienLE “1000 LBJs” just feels like a complete category error in thinking about the world to me
I’m not sure what you mean by that. I chose him because I thought you’d be familiar! My knowledge is pretty limited to Master of the Senate, but if you think the info there is roughly correct we should be able to speak from some common ground. Re helping move things through the world, I wasn’t arguing it would help anything in a normative way. I was talking about skill! As in, if LBJ was alive today and super motivated to get high speed rail accomplished in California, he would be much more effective than most people in accomplishing that than the large majority of people in similar positions. Now would 10 LBJs be better than 1? I’m not confident about how would such people could and would work together towards common goals. Perhaps you could speak to that. I think thinking through this kind of thing matters because I think how AIs wield political power could matter a lot.
Maybe a question which would cut closer to the heart of it: do you think AIs will come to be better at politics than humans? I think you might quibble at the word politics. I mean the relationships, the coalition building, the speaking and letter writing, the getting specific laws passed and specific allies into specific positions, the commanding of loyalty, etc. Do you think a group of GPTs or Claudes or their successors will be able to do this thing better than the best humans in the future?
@deanwball @sebkrier @AdrienLE Why? You can imagine one LBJ right? Now imagine you copied him 1000 times. I get that this is very impossible with humans given our level of tech, but it’s not with AIs!
@deanwball @sebkrier @AdrienLE AIs are currently *way* worse at the kind of politics LBJ was good at. But seems likely they’ll be superhuman in that domain in a small number of years
@deanwball @sebkrier @AdrienLE Though it really depends on their ability to coordinate with each other!
@deanwball @sebkrier @AdrienLE No and no, but I’m not sure what you’re getting at.
Users rejected the idea of coordinating multiple LBJs for success as a fundamental category error in thinking about AI systems.
Based on 2 visible X reactions from 2 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@deanwball @sebkrier @AdrienLE AIs are currently *way* worse at the kind of politics LBJ was good at. But seems likely they’ll be superhuman in that domain in a small number of years
@deanwball @sebkrier @AdrienLE Though it really depends on their ability to coordinate with each other!