everyone save the most gullible tech-fetishists at this point know machine learning is fake — it's just so much gauze over statistical decision making processes. its true purpose is to absolve the people who make it from the decisions they sign off on: neural nets are so complex and intentionally mystified they can be used for plausible deniability even when they're obviously doing exactly what the people who designed them, fed them their biases, and supervised them intended
@esvrld that's mostly true, it's a lot of hype and social/marketing/policy uses are predicability awful. But some of the many things they call machine learning (including just big statistics) do have some uses in classifying and guiding in research computing.
@esvrld also uses up incredible amounts of time and computing power (huge carbon footprint) for mostly discarded attempts and junk results
