346 sats \ 1 reply \ @Voldemort 29 Mar \ on: Elites, the curse of recursion, and the half-life of policy mostly_harmless
The parallels between rocks 1 and 2 are interesting. Elites pair with elites, creating further separation from reality. LLMs use data from LLMs, creating further separation from the raw data.
Though I have always qualified anyone who thinks they know better than the public as an elite. IE, people often argue we shouldn't have more school choice because "poor people don't choose the right schools." I find that an abhorrent stance, as it assumes their preferences are better than all others. I don't think this changes the elites pairing with elites changes, however.
I'm with you, which is one of the reasons I was wary about talking about "elites" at all, and probably too resistant to learning about them.
Every system of any kind, anywhere, has some smallish group that exerts outsize control over it. Sometimes -- perhaps often -- this is fine, and what you want. If I ever need neurosurgery I will be super glad to have HMS-trained surgeon who did residency at Cleveland Clinic or wherever. A lot of people who run things, and are sufficiently exposed to the consequences of reality, are both elite and incredibly competent. You want them on that wall.
"Elite" shouldn't be a condemnation, but sometimes, in practice, it is.
reply