Back to News
Advertisement
Advertisement

⚡ Community Insights

Discussion Sentiment

50% Positive

Analyzed from 79 words in the discussion.

Trending Topics

#human#models#https#wapo#obvious#reading#produced#phrases#treasury#announcement

Discussion (3 Comments)Read Original on HackerNews

bookofjoeabout 6 hours ago
paulpauperabout 6 hours ago
It's obvious reading it that it was produced by AI. Phrases like, "Treasury's announcement gave the game away," which is also a sentence fragment, so the AI fails at even basic grammar (and shame on the WSJ editor for not noticing this too).

The irony is that although LLMs are trained on human writings, the output is conspicuously un-human. It's like after 4+ years and billions of dollars invested on SOTA models, AI is still incapable of producing natural-sounding writing. This should not be that hard, yet it is.

13yearsabout 5 hours ago
It is due to the RL done on the models to make them "more pleasing" to work with.