30
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 01 Jul 2024
30 points (100.0% liked)
TechTakes
1435 readers
75 users here now
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
founded 1 year ago
MODERATORS
Dan Luu's "A discussion of discussions on AI bias", about techbros trying to gaslight the rest of the world into thinking ML models don't have problems
the entire post is very good, but my brain zeroed in on this as both a perfect example of why search was absolutely fucked even before LLMs (who in fuck deploys a language heuristic that doesn’t take the content of the page into account? who asked for this?) and of the engineering attitudes that feed into LLMs and generative AI having unevaluated biases and defenders that insist those biases can’t be real
I run into this exact problem so often with music: it’s extremely fucking annoying to convince search engines or other kinds of things that, yes, I do in fact want that turkish/finnish/russian/brazilian/thai/irish/….. result set from this english query, and there’s just almost zero affordance for it
they segment things into de facto silos and if you don’t already happen to have some way in which to pull on a string to lead the way there (or can input the desired query in the applicable language and spelling) you’ll have a bitch of a time. check a couple of those items from my entries on the music thread a while back for references to test with
god damn those recasting images
it will never not be hilarious how pathetic this technology/approach is