LLM also makes it way too easy to “cheat”. It’s really like bilbo and the ring of power: “this is pretty hard to explain in a coherent way, maybe, just maybe, I can just this one time use AI? Nobody will know!”. And little by little you end up arguing with yourself that actually generating the whole article isn’t really that bad, while slowly morphing into a human harness

Your professor is right. There’s nothing wrong to use AI to aid the process _of thinking_. That’s actually really helpful, at least to me.

You’re right about firewalling it in your mind. The moment it starts steering you into ideas that aren’t yours, the piece is screwed.

> I had a professor who said that writing is thinking.

Socrates would have laughed at that.

Maybe in the future we'll believe that prompting is thinking.

Did you mean to write

> If you start writing today, there's no way to know if you can think without writing.

Yes, silly title both because it isn't in the article and because it is not true. Of course you can write without "AI", al it takes is a person and a writing implement of choice ranging from a clay tablet through a slate, papyrus, paper, typewriter or a free software document processor like LibreOffice which does not contain "AI" tools. Once so equipped all the person has to do is to start writing, write whatever is on his mind and finish the job. Voila, "AI"-free writing, started "today".

My joke around this is

"HN, detecting 200 of the last 10 AI written articles"

Tools like Panagram will simply fall into uselessness as more people read AI and write like AI. Language converges. Much like people started saying "that's cool" to mean something is neat rather than cold, "If not X then Y" will become a pattern humans use far more commonly because they read it in AI all the time.