Next-token-prediction cannot do calculations. That is fundamental.

It can produce outputs that resemble calculations.

It can prompt an agent to input some numbers into a separate program that will do calculations for it and then return them as a prompt.

Neither of these are calculations.

▲

gf000 3 hours ago | parent | next [-]

So you don't think 50T parameter neural networks can encode the logic for adding two n-bit integers for reasonably sized integers? That would be pretty sad.

▲

razorbeamz 3 hours ago | parent [-]

They do not. The fundamental technology behind LLMs does not allow that to be the case. You are hoping that an LLM can do something that it cannot do.

	▲	gf000 2 hours ago \| parent [-]
		https://arxiv.org/html/2502.16763v2 You are wrong. Especially that we are talking about models with 50T parameters. Can they do arbitrary computations for arbitrarily long numbers? Nope. But that's not remotely the same statement, and they can trivially call out to tools to do that in those cases.

▲

parasubvert 5 hours ago | parent | prev [-]

Humans can't do calculations either, by your definition. Only computers can.

	▲	datsci_est_2015 2 hours ago \| parent [-]
		Third things can exist. In other words, you’re implying a false dichotomy between “human computation” and “computer computation” and implying that LLMs must be one or the other. A pithy gotcha comment, no doubt. Edit: the implication comes from demanding that the OP’s definition must be rigorous enough to cover all models of “computation”, and by failing to do so, it means that LLMs must be more like humans than computers.