You wrote your comment one word at a time, with the next word depending on the previous words written.

You did not plan the entire thing, every word, ahead of time.

LLMs do the same thing, so... how is your intelligence any different?

A long time ago I noticed that I sometimes already had a complete thought before my inner monologue turned it into words. A few times I tried skipping the inner monologue because I'd clearly already thought the thought. Turns out the bit of my brain that creates the inner monologue from the thought, can generate a sense of annoyance that the rest of my brain can feel.

Not that it matters, there's evidence that while LLMs output one word at a time, they've got forward-planning going on, having an idea of the end of a sentence before they get there.

	▲	rcxdude 2 days ago \| parent [-]
		Indeed, and it seems like they would really struggle to output coherent text at all if there was not some kind of pre-planning involved (see how even humans struggle with it in games where you have to construct a sentance by having each person shout out one word at a time). Even GPT-2 likely had at least some kind of planning for the next few words in order to be as coherent as it was.

▲

lossyalgo 2 days ago | parent | prev [-]

Tell that to German-speakers, where the verb comes last, and the order of things in sentences is not anything like English, therefore requiring you to think of the entire sentence before you just spit it out. Even the numbers are backwards (twenty-two is two-and-twenty) which requires thinking.

Furthermore, when you ask an LLM to count how many r's are in the word strawberry, it will give you a random answer, "think" about it, and give you another random answer. And I guarantee you out of 3 attempts, including reasoning, it will flip-flop between right and wrong, but unlike a human, it will be random, because, unlike humans who, when asked "how many r's are in the word strawberry" will not be able to tell you the correct answer every. fucking. time.

edit: formatting

▲

pka 2 days ago | parent | next [-]

It seems models are pre-planning though:

> How does Claude write rhyming poetry? Consider this ditty:

> He saw a carrot and had to grab it,

> His hunger was like a starving rabbit

> To write the second line, the model had to satisfy two constraints at the same time: the need to rhyme (with "grab it"), and the need to make sense (why did he grab the carrot?). Our guess was that Claude was writing word-by-word without much forethought until the end of the line, where it would make sure to pick a word that rhymes. We therefore expected to see a circuit with parallel paths, one for ensuring the final word made sense, and one for ensuring it rhymes.

> Instead, we found that Claude plans ahead. Before starting the second line, it began "thinking" of potential on-topic words that would rhyme with "grab it". Then, with these plans in mind, it writes a line to end with the planned word.

[https://www.anthropic.com/research/tracing-thoughts-language...]

▲

nxor 2 days ago | parent | prev [-]

The part about strawberry is just not right. That problem was solved. And I do think it's a stretch to say German speakers think of the entire sentence before speaking it.

	▲	lossyalgo 2 days ago \| parent [-]
		LLMs were not designed to count letters[0] since they work with tokens, so whatever trick they are now doing behind the scenes to handle this case, can probably only handle this particular case. I wonder if it's now included in the system prompt. I asked ChatGPT and it said it's now using len(str) and some other python scripts to do the counting, but who knows what's actually happening behind the scenes. [0] https://arxiv.org/pdf/2502.16705