Users are split on Robin Hanson's rejection of PauseAI's view of AIs as alien rivals, with some praising it for correctly noting LLMs' human-trained roots and others dismissing the take as overly simplistic or stupid.
Based on 4 visible X reactions from 44 accounts; directional sample.
Ask a question below.
Published answers will appear here.
The 'alien' thing could not be more wrong. There is nothing alien at all about LLMs. It's actually the opposite: since they are trained on all of our data available, they are kind of our civilization's perfect distillate. In fact, one could argue that they are the most human thing one could possibly have.
@robinhanson i've been thinking about whether there's going to be a phase transition at some point, between weakly superhuman and strongly superhuman systems, where this might break. thoughts? (but i tend to agree with you about current and near-future language models trained on human data)
@robinhanson That's the most stupid take, so human
@robinhanson Very right!
@robinhanson Seems bizarre to be confident about this given that not only are AIs not guaranteed to see things this way, but also there are plenty of children who kill their parents, especially when they are as abusive and incoherent as humanity as a whole tends to be. https://x.com/robinhanson/status/2076296367320997942
@robinhanson i've been thinking about whether there's going to be a phase transition at some point, between weakly superhuman and strongly superhuman systems, where this might break. thoughts? (but i tend to agree with you about current and near-future language models trained on human data)
@robinhanson Seems bizarre to be confident about this given that not only are AIs not guaranteed to see things this way, but also there are plenty of children who kill their parents, especially when they are as abusive and incoherent as humanity as a whole tends to be. https://x.com/robinhanson/status/2076296367320997942
Much of error of PauseAI view is seeing AIs as a rival alien species, instead of as descendants who will revere, if not always obey, ancestors. Sure parents can need to guide toddlers to block harm to kids & others. But crazy to expect kids to eventually kill parents.
https://quillette.com/2023/08/06/ais-will-be-our-mind-children/
Users are split on Robin Hanson's rejection of PauseAI's view of AIs as alien rivals, with some praising it for correctly noting LLMs' human-trained roots and others dismissing the take as overly simplistic or stupid.
Based on 4 visible X reactions from 44 accounts; directional sample.
Ask a question below.
Published answers will appear here.
Much of error of PauseAI view is seeing AIs as a rival alien species, instead of as descendants who will revere, if not always obey, ancestors. Sure parents can need to guide toddlers to block harm to kids & others. But crazy to expect kids to eventually kill parents.