Okay, let's start a new game where we try to guess the prompt that produced this text.

My guess would be "Write a 2000 words blog post based on the analogy between the difficulty of understanding the laws of physics and minified source code."

It's a cool analogy, no doubt, but reading the 2000 words of the article, I felt there was no substance or flesh around it. Okay there were some examples vaguely related to the analogy, and writing was smoothish, but what was the message in the end? What were the new ideas (beyond the analogy)?

Every time I read a LLM generated long form text I am reminded of the communication engineering: communication channels can only decrease the information content. For example, in the communication setup:

   Alice sends msg  --encode--> codeword --channel--> codeword' --decode--> msg' Bob received

The best we can hope for is that the message was transmitted faithfully, so msg' = msg, since the channel noise can only degrade the information content, specifically, the mutual information is a non-increasing function:

   I(msg;codeword') ≤ I(msg;codeword),

due to the monotonicity of information (a.k.a. data processing inequality).

Similarly, if Alice uses LLM to communicate to Bob, we have:

   Alice's idea --encode--> prompt --LLM--> generated text --decode--> idea'

Monotonicity of information tells us

   I(idea;generated text) ≤ I(idea;prompt).

So I really don't get it, why do people believe that adding LLM generation step could improve the communication of their ideas? It seems directly analogous to passing a message through an additional noisy channel. Just show us the prompt.

▲ wsowens 2 hours ago | parent | next [-]

Agree the essay has a suspiciously Claude/ChatGPT-like organization. This comment perfectly articulates my distaste for AI-generated text with words I've been struggling to find.

▲ 0gs 2 hours ago | parent | prev [-]

sometimes i feel like the reason nobody ever posts their raw prompts (or at least extremely rarely) is because even though we all know our prompts are not golden geese, they still feel like that in our hands. i'd bet everybody's prompt writing style is like a fingerprint and we all assume they look like our own.

▲

andrewfromx 2 hours ago | parent [-]

totally. https://dontpastetheai.com/ all we want is to stop seeing obvious copy paste. Give me plausible deniability that it's human authored even though we all know nothing is human authored any more.

▲

mannanj 2 hours ago | parent [-]

plenty of content is human authored. the low status ones are not, in my argument.

You could keep pretending that everyones using AI to make low quality or "AI high quality" output but either way, when its obvious its human made, and many humans still write, it will be higher status.

edit: isn't it ironic that that link looks AI generated asking to not share AI generation output

	▲	andrewfromx 2 hours ago \| parent [-]
		^ this one is perfect! Sounds almost 100% human authored.