Adam Curtis on the dangers of self-expression (2017)

Re:Adam Curtis on the dangers of self-expression (2017) by Animats on Saturday May 11, 2024 @12:31PM (#40329097)

In the post-LLM era, that just isn't good enough. Now that we have well-written automated probabilistic meandering text generation, "thought provoking but utterly lacking in rigour" copy is everywhere. Usually followed by a clickbait link.

It's the end of a whole second-tier literary genre.

[ Parent | on HN ]

Re:Adam Curtis on the dangers of self-expression (2017) by andybak on Saturday May 11, 2024 @01:50PM (#40329582)

I disagree. AI will be able to do "rigour" before it can do "thought provoking".

Part of Adam's appeal is the fact that behind it is a human being struggling to understand the world and struggling to explain it.

LLMs are incredible but nobody cares what a hyper-evolved Markov chain pretends to say about culture.

[ Parent | on HN ]

Re:Adam Curtis on the dangers of self-expression (2017) by Animats on Tuesday May 14, 2024 @01:45AM (#40351997)

> AI will be able to do "rigour" before it can do "thought provoking".

I wish. Right now, thought provoking LLMs exist, but rigorously correct ones do not. Which is a big problem if you want to use them beyond blithering.

[ Parent | on HN ]

Re:Adam Curtis on the dangers of self-expression (2017) by xboxnolifes on Saturday May 11, 2024 @05:43PM (#40330824)

Llm output is not thought provoking.

[ Parent | on HN ]