Sure, let's break down each line of the post and explain its meaning:

"The corporations have done a great job confusing the fact that LLMs are really just giant balls of copyright infringement and essentially a giant archive of text that has been compressed using the lossy GPT algorithm."

This line criticizes corporations for obscuring the true nature of LLMs (Language Model Models). It suggests that LLMs are essentially massive collections of copyrighted content and are created by compressing text using a lossy algorithm called GPT (Generative Pre-trained Transformer). The use of "giant balls of copyright infringement" is a figurative expression to emphasize the claim that LLMs consist mainly of copyrighted material.

"Prompts are just lookup phrases to pull out specific text files and legible results / "emergent behavior" are decompression artifacts when an improper lookup phrase is entered."

Here, the post explains the role of prompts in LLMs. It states that prompts are simply phrases used to retrieve specific text files from the LLM's archives. The term "legible results" refers to the generated text output produced by the LLM in response to the given prompt. The post suggests that any interesting or unexpected output, often referred to as "emergent behavior," is actually a result of the decompression process when an incorrect or improper prompt is entered. In other words, the LLM may generate text that seems creative or novel, but it is essentially a side effect of the compression and decompression process, rather than the LLM's true understanding or intelligence.

Overall, the post expresses a critical view of LLMs, arguing that they are primarily composed of copyrighted material and that their output is not genuinely creative or intelligent but rather a result of the algorithms used for compression and decompression.