diz
I love the “criti-hype”. AI peddlers absolutely love any concerns that imply that the AI is really good at something.
Safety concern that LLMs would go Skynet? Say no more, I hear you and I’ll bring it up in the congress!
Safety concern that terrorists might use it to make bombs? Say no more! I agree that the AI is so great for making bombs! We’ll restrict it to keep people safe!
Sexual roleplay? Yeah, good point, I love it. Our technology is better than sex itself! We’ll restrict it to keep mankind from falling into the sin of robosexuality and going extinct! I mean, of course, you can’t restrict something like that, but we’ll try, at least until we release a hornybot.
But any concern about language modeling being fundamentally not the right tool for some job (Do you want to cite a paper or do you want to sample from the underlying probability distribution?), hey hey hows about we talk about the skynet thing instead?
Also, my thought on this is that since an LLM has no internal state with which to represent the state of the problem, it can’t ever actually solve any variation of the river crossing. Not even those that it “solves” correctly.
If it outputs the correct sequence, inside your head the model of the problem will be in the solved state, but on the LLM’s side there’s just a sequence of steps that it wrote down, with those steps directly inhibiting production of another “Trip” token, until that crosses a threshold. There isn’t an inventory or even a count of items, there’s an unrelated number that weights for or against “Trip”.
If we are to anthropomorphize it (which we shouldn’t, but anyway), it’s bullshitting up an answer and it gradually gets a feeling that it has bullshitted enough, which can happen at the right moment, or not.
Other thing to add to this is that there’s just one or two people in the train providing service for hundreds of other people or millions of dollars worth of goods. Automating those people away is simply not economical, not even in terms of the headcount replaced vs headcount that has to be hired to maintain the automation software and hardware.
Unless you’re a techbro, who deeply resents labor, someone who would rather hire 10 software engineers than 1 train driver.
Both parties are buying into a premise we already know to be incorrect.
We may know it is incorrect, but LLM salesmen are claiming things like “90th percentile on LSAT”, high scores on a “college level reasoning benchmark” and so on and so forth.
They are claiming “yeah yeah there’s all the anekdotal reports of glue pizza, but objectively, our AI is more capable than your workers, so you can replace them with our AI”, and this is starting to actually impact the job market.
Yeah I think that’s why we need an Absolute Imbecile Level Reasoning Benchmark.
Here’s what the typical PR from AI hucksters looks like:
https://www.anthropic.com/news/claude-3-family
Fully half of their claims about performance are for “reasoning”, with names like “Graduate Level Reasoning”. OpenAI is even worse - recall theirs claiming to have gotten 90th percentile on LSAT?
On top of it, LLMs are fine tuned to convince some dumb ass CEO who “checks it out”. Even though you can pay for the subscription, you’re neither the customer nor the product, you’re just collateral eyeballs on the ad.
But if your response to the obvious misrepresentation that a chatbot is a person of ANY level of intelligence is to point out that it’s dumb you’ve already accepted the premise.
How am I accepting the premise, though? I do call it an Absolute Imbecile, but that’s more of a word play on the “AI” moniker.
What I do accept is an unfortunate fact that they did get their “AIs” to score very highly on various “reasoning” benchmarks (some of their own design), standardized tests, and so on and so forth. It works correctly across most simple variations, such as changing the numbers in a problem or the word order.
They really did a very good job at faking reasoning. I feel that even though LLMs are complete bullshit, the sheer strength of that bullshit is easy to underestimate.
I feel like letter counting and other letter manipulation problems kind of under-sell the underlying failure to count - LLMs work on tokens, not letters, so they are expected to have a difficulty with letters.
The inability to count is of course wholly general - in a river crossing puzzle an LLM can not keep track of what’s on either side of the river, for example, and sometimes misreports how many steps it output.
Well the problem is it not having any reasoning period.
Not clear what symbolic reasoning would entail, but puzzles generally require you to think through several approaches to solve them, too. That requires a world model, a search, etc. the kind of stuff that actual AIs, even a tik tac toe AI, have, but LLMs don’t.
On top of it this all works through machine learning, which produces the resulting network weights through very gradual improvement at next word prediction, tiny step by tiny step. Even if some sort of discrete model (like say the account of what’s on either side of the river) could help it predict the next token, there isn’t a tiny fraction of a discrete “model” that would help it, and so it simply does not go down that path at all.
I think you can make a slight improvement to Wolfram Alpha: using an LLM to translate natural language queries into queries WA can consume, then feeding them into WA. WA always reports exactly what it computed, so if it “misunderstands” you, it’s a lot easier to notice.
The problem here is that AI boys got themselves hyped up for it being actually intelligent, so none of them would ever settle for some modest application of LLMs. Google fired the authors of “stochastic parrot” paper, AFAIK.
simply pasting LLM output into CAS input and then the CAS output back into LLM input (which, let’s be honest, is the first thing tech bros will try as it doesn’t require much basic research improvement), will not help that much and will likely generate an entirely new breed of hilarious errors and bullshit (I like the term bullshit instead of hallucination, it captures the connotation errors are of a kind with the normal output).
Yeah I have examples of that as well. I asked GPT4 at work to calculate the volume of 10cm long, 0.1mm diameter wire. It seems to be doing correct arithmetic by some mysterious means which do not use scientific notation, and then the LLM can not actually count so it miscounts zeroes and outputs a result that is 1000x larger than the correct answer.