As Marshall McLuhan said "The medium is the message.". You cannot separate the cargo from the ship in this case because the functioning of the ship/AI unavoidably contaminates the cargo/message.

So, what does the medium of LLMs tell us about the message passed through it? What are the sub-textual aspects of messages with obvious AI tells? None of them are positive: risk of factual errors from hallucination, risk of it being a low effort message, risk of the message being subtly and unintentionally biased by the biases of the AI, the list goes on.

Just change the analogy from cargo/ship to item/package. If the packaging is damaged I wouldn't expect the item inside to be in good condition.

https://xcancel.com/i/article/2089796750576066852