These important limitations highlight why it’s still important to have humans involved in the analysis process here. The NYT notes that, after querying its LLMs to help identify “topics of interest” and “recurring themes,” its reporters “then manually reviewed each passage and used our own judgment to determine the meaning and relevance of each clip… Every quote and video clip from the meetings in this article was checked against the original recording to ensure it was accurate, correctly represented the speaker’s meaning and fairly represented the context in which it was said.”
It’s literally the paragraph right after.
They verify it.
Won’t the checking cost more time then to just write it themselves?
It’s 400 hours of audio, the transcripts ended up being 5 million words, and only snippets of it are useful.
It’s harder to create new content than to correct existing content.