The LLM Critics Are Right. I Use LLMs Anyway

Amplification is not Augmentation. Just because you can yell louder doesn't mean what you have is worth yelling about. And just because a lot of people are yelling doesn't mean you should too.

There is a big difference between something that appears at face value to be high quality, and something that proves itself to be high quality. LLMs are good at replicating patterns that humans see as high quality. But there's no guarantee of this quality actually being real. Often it only appears to be so.

Just by listening to people or the talks or reading HN, I think I got a pretty good sense of why certain people refuse to use LLMs. And what makes it extremely weird is that I agree with almost all of their points!

Curiously, the author completely misses one of the biggest points; while they list concerns over copyright, ethics, slop, and the economy, they completely fail to notice that the output of Transformer-driven chatbot harnesses is bad writing. Indeed, the post only reads as coherent if it makes the assumption that chatbots are good writers. The author perceives this as almost-agreement, ignoring the difference between foundational assumptions and emergent obstacles, but it comes across as a dire misreading of a neighboring culture.