atomic_gingerbread
No bio...
User ID: 258
atomic_gingerbread
1yr ago
I see him get called out for straw-manning & being a bad-faith actor
I don't think he's bad-faith, inasmuch as he's pretending to be something that he's not. He's openly a right-wing culture warrior, and he details exactly how he fights without reservation. When he was placed on the New College board by DeSantis, the NYT accused him of staging a "hostile takeover" of the school in order to roll back the "long march through the institutions" that had made such colleges left-wing bastions. His response was, essentially, "yes, that's exactly what I'm doing".
He's also strangely scrupulous for a culture warrior, refusing to engage in "groomer" rhetoric like his compatriots. Definitely a guy to watch for, as most people in his space develop acute Twitter Brain and self-immolate before they can have any sort of meaningful political impact.
atomic_gingerbread
2yr ago
The bolded part is such a bizarre claim in this context. The only way that random stops can lead to racial profiling is if the law enforcement apparatus is somehow biased. Assuming that's true, how exactly would that bias disappear when pursuing low-level stops? It's impossible for cops to be everywhere and to enforce every possible offense in existence, so naturally they have to make some discretionary decisions.
True, but the relationship isn't a pure binary: the more discretion police are given, the more room bias has to operate. "Random" really means that police are given complete discretion since nobody is ensuring mathematical randomness as you propose. Curtailing this discretion will reduce (but not eliminate) racial profiling. Sure, biased police might turn a blind eye to white scofflaws, but they will (in theory) be restrained from harassing law-abiding minorities. The latter injustice is generally more vexing to our sense of equality than the former.
atomic_gingerbread
2yr ago
The Catholic Church does not claim infallibility. It does claim infallibility ex cathedra for the Pope, roughly meaning that if the Pope (qua Pope) says that something concerning faith or morals should be believed by the entire Catholic Church, then every Catholic should believe that he says that they should believe.
Even this doesn't quite capture how limited the doctrine of infallibility is in practice. The proclamations widely considered infallible are quite rare, and mostly decide theological questions rather than practical moral ones.
It's possible that human knowledge is also fundamentally statistical or associative in nature, but we have additional faculties that LLMs don't, and it's these deficits which are responsible for their peculiar errors, not inability to have knowledge per se. For example, LLMs almost certainly lack second-order knowledge, i.e. knowledge about what they know. Facts about the model itself are not part of their training data, nor does their execution make any provision to evaluate the prompt so that self-facts are relevant to the output. This means LLMs lack any capacity for introspection or self-representation, and therefore can't possibly respond to challenging questions with "I don't know" or "I don't understand" -- they don't have an I! This is a significant limitation, but philosophically a different one from the inability to possess knowledge, unless your definition of knowledge requires these additional functions in the first place.
More options
Context Copy link