If LLMs can be considered cultural technologies like the printing press and bureaucracies, certain types of analyses and critiques become interesting e.g. McLuhan i.e. are LLMs a medium of language that will frame culture?
@DouthatNYT Actually I find the most compelling part of his argument the part on how Anthropic itself takes no responsibility for the actions of Claude, therefore if you listen to their actions not their words they are telling you what they themselves really think.
@kareem_carr Also, reminds me of the effect of digital videos. Making it easy to film has made directors bad at cinematography and lighting. A broader takeaway from this is that there is a point at which quantity crowds out quality.
@kareem_carr Reminds me of this: "We’ve used problems to build intuition and to train people to have a good idea of what’s true, what to expect, what is provable, and what is difficult. Just getting the answers right away may actually inhibit that process."
https://t.co/MvfUghE3CC
@justin_tweeting@BillSimmons Interesting math, but this isn’t accounting for how odd it is that someone is also so bizarrely consistent, which is already very rare. So 10% times what is already probably a 1 of 1 type of an achievement?
@annielcrawford@SketchesbyBoze That just happened to me I was like “pee-cue” my interest and she was like isn’t it “peak” and I was like wait, it is isn’t it, but I swear I’ve seen this other word… not realizing it’s the same word.
@thenanyu An analogy that occurs to me is that LLMs can write better than 90% of professionals, but I can read my LinkedIn feed and tell that almost all of it is written by LLMs and I can tell that it is mostly not very good
@thenanyu That’s not what I would be referring to if I said that humans contribute taste (not my word). I would be referring to judging if an LLM output is good. To me the phenomenon you are describing is LLMs are good at answering questions because they have lots of knowledge/expertise.
@annielcrawford Yes, few committed materialists know how science works. The myth of science is that it is purely from fact untainted by human bias, and capable of answering all useful questions. But the activities of science are very human, and the questions scientists choose are very limited.
@dodeja Agree with this general sentiment.
Oracle is the only name that came into my head to mention, but I don’t think they’re a real contender to take the lead on these companies.
What I do think is possible is that Larry Ellison is the richest person in the world in 2030.