Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

thebullshitmachines.com

Post reply on HN
Print view View original

581–590 of 652 posts

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#581

Post by ninetyninenine »

wruza wrote: post #538
ninetyninenine wrote: post #504

Earlier quoted context omitted.

Reasoning is an abstract term. It doesn’t need to be similar to human reasoning. It just needs to be able to arrive at the answer through a process. Clearly we used the term reasoning for many varied techniques. The term doesn’t narrow to specifically one form of “human” like reasoning only.

Oh, that is true. "It" doesn't have to do human reasoning, at all. But we have to at least define "reasoning" for the given manifestation of "it". Otherwise it's just birdspeak. Because reasoning is "the action of thinking about something in a logical, sensible way", which has to happen somewhere if not finger-pointable, then at least somehow scannable or otherwise introspectable. Otherwise it's yet another omnidude…

Well no. If you create a machine that produces output indistinguishable from the output of things we "know" can "reason" aka "humans". Then I would call that reasoning.

If the output has a low probability of occuring by random chance then it must be reason.

>For example, if you prove that the reasoning is somehow embedded as a spatial in-network set of dimensions rather than in-time, wouldn't that be literally equivalent to "it just knows the patterns"? What would that term substitution actually achieve?

I mean, this is a method many humans use to reason themselves.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#582

Post by ninetyninenine »

daveguy wrote: post #572
ninetyninenine wrote: post #541

Earlier quoted context omitted.

We use Probability. Find a prompt that has a large range aka codomain. If it arrived at the correct answer then that the only possibility here is reasoning because the codomain is so large it cannot arrive there by random chance. Of course make sure the prompt is unique such that it's not in the data and it's not doing any sort of "pattern matching". So like all science we prove it via probability. Observations match…

Pardon my ignorance -- assuming that range and codomain are approximately equivalent in this context, how do you specify a prompt with a large codomain? Is there a canonical example of a prompt with a large codomain? It seems to me that, in natural language, the size of the codomain is related to the specificity of the prompt. For instance, if the prompt is "We are going to ..." then the codomain is enormous. But if…

You'll have to drop a bit of rigor here.

I ask the question, what is 2 * 2, which is an obviously loaded question that's pattern matched to death.

The LLM can answer "4" or "The answer is 4" of "looks like the answer is 4"

All valid answers but all the same. We count all 3 of those answers as just 4 out of the set of numbers. But we have to use our own language faculties to cut through the noise of the language itself.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#583

Post by daveguy »

ninetyninenine wrote: post #582
daveguy wrote: post #572

Earlier quoted context omitted.

Pardon my ignorance -- assuming that range and codomain are approximately equivalent in this context, how do you specify a prompt with a large codomain? Is there a canonical example of a prompt with a large codomain? It seems to me that, in natural language, the size of the codomain is related to the specificity of the prompt. For instance, if the prompt is "We are going to ..." then the codomain is enormous. But if…

You'll have to drop a bit of rigor here. I ask the question, what is 2 * 2, which is an obviously loaded question that's pattern matched to death. The LLM can answer "4" or "The answer is 4" of "looks like the answer is 4" All valid answers but all the same. We count all 3 of those answers as just 4 out of the set of numbers. But we have to use our own language faculties to cut through the noise of the language itsel…

> I ask the question, what is 2 * 2, which is an obviously loaded question that's pattern matched to death.

Yeah, that was my point. Small codomain -> easy to validate. Large codomain -> open to interpretation. You implied that to prove reasoning, pick a prompt with a large codomain and if the LLM answers with accurate precision, then viola, reasoning.

So my question was, can you give an example of a prompt with a high codomain that isn't subject to wide interpretation? It seems the wider the codomain the easier it is to say, "look! reasoning!"

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#584

Post by ssssvd »

Here's a bridging argument b/w the OPs and the commenters.

Anglo-Saxon thought (utilitarianism, behaviorism, pragmatism) treats truth as probability. If an LLM outputs the right tokens in the right order, that’s thinking. If it predicts true statements better than humans, that’s knowledge. The Turing Test? Behaviorist by design. Bayesian inference? A formalization of Anglo empiricism.

Continental philosophy rejects this. Heidegger: no Dasein, no being. Sartre: no self-awareness, no thought. Derrida: no deconstruction, no meaning. The German Idealists would outright laugh.

So in the Anglo tradition, LLMs are already "thinking." In the French/German view, they’re an epistemic trick — a probabilistic mirror, not a mind.

It’s not what LLMs are, it’s how your epistemic tradition defines “thinking.” And that’s probably why the EU is so "lagging behind" in the AI race — no amount of quacking makes an LLM a duck to a Continental. It’s still a parrot.

Where you land in this debate is easy to test: Are you comfortable with the statement, "Truth is just what’s most probable given what we already know"?

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#585

Post by ssssvd »

ssssvd wrote: post #584

Here's a bridging argument b/w the OPs and the commenters. Anglo-Saxon thought (utilitarianism, behaviorism, pragmatism) treats truth as probability. If an LLM outputs the right tokens in the right order, that’s thinking. If it predicts true statements better than humans, that’s knowledge. The Turing Test? Behaviorist by design. Bayesian inference? A formalization of Anglo empiricism. Continental philosophy rejects t…

The hilarious outcome? Americans eventually build something they consider "smarter" than themselves — French philosophers agree, but only because it lets them place themselves one step higher.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#586

Post by anukulrm »

Bookmarked and will be sharing this with my friends.

Separately, glad to see you, @ctbergstrom and (now Prof) Jevin West still collaborating :). You gave me the opportunity to spend a few days as an undergraduate in your lab 15yrs ago when Jevin was a grad student. It was during the H5N1 scare and I remember you being called up. I was just star struck then and honestly overwhelmed. I think I helped fix some perl code for the eigenfactor project, iirc. Will never forget you and Prof Ben Kerr. Anyways, just want to say thanks after all these years.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#587

Post by hnfong »

cristiancavalli wrote: post #526
semiquaver wrote: post #524

Earlier quoted context omitted.

What would definitive proof look like? Can you definitively prove that your brain is capable of reasoning and not a convincing simulation of it?

I can’t and that’s pretty cool to think about! Of course if we’re going that far down the chain of assumption we’re not quite ready to talk about LLMs imo (then again maybe it would be the perfect place to talk about them as contrast/comparison; certainly exciting ideas in that light). From my own perspective: if we’re gonna say these things reason and we’re using the definition of reasoning we apply to humans, then…

IMHO, any argument against LLM intelligence should be validated by first applying them to humans.

And then you'd realize that a lot of naïve arguments against LLMs would imply that a significant portion of homo-sapiens can't reason, are unable to really think, and are no more than stochastic parrots.

It's actually a rather dangerous line of reasoning.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#588

Post by wruza »

ninetyninenine wrote: post #581
wruza wrote: post #538

Earlier quoted context omitted.

Oh, that is true. "It" doesn't have to do human reasoning, at all. But we have to at least define "reasoning" for the given manifestation of "it". Otherwise it's just birdspeak. Because reasoning is "the action of thinking about something in a logical, sensible way", which has to happen somewhere if not finger-pointable, then at least somehow scannable or otherwise introspectable. Otherwise it's yet another omnidude…

Well no. If you create a machine that produces output indistinguishable from the output of things we "know" can "reason" aka "humans". Then I would call that reasoning. If the output has a low probability of occuring by random chance then it must be reason. >For example, if you prove that the reasoning is somehow embedded as a spatial in-network set of dimensions rather than in-time, wouldn't that be literally equiva…

A side effect of this is that a zip.exe that unzips a zip into a book that contains text indistiguishable from the output of a human must reason too.

From what I can see, you’re only massaging semantics. That is uninteresting.

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#589

Post by crackalamoo »

JPLeRouzic wrote: post #550
Mistletoe wrote: post #478

Earlier quoted context omitted.

I’d take the Wikipedia answer any day. Millions of eyes on each article vs. a black box with no eyes on the outputs.

> " Millions of eyes on each article " Only a minority of users contribute regularly (126,301 have edited in the last 30 days): https://en.wikipedia.org/wiki/Wikipedia:Wikipedians#Number_o... And there are 6,952,556 articles in the English Wikipedia, so an average article is corrected every 55 months (more than 4 years). It's hardly "Millions of eyes on each article"

But of those 126,301 people who have edited in the last 30 days, some of them have edited more than one article. In fact, some have made up to millions of edits (lifetime), which disproportionately increases the total. At least 5000 people have edited more than 24,000 times.

https://en.wikipedia.org/wiki/Wikipedia:List_of_Wikipedians_...

(And also: each editor has (approximately) 2 eyes :) )

Re: Modern-Day Oracles or Bullshit Machines? How to thrive in a ChatGPT world

#590

Post by llm_trw »

habinero wrote: post #570
llm_trw wrote: post #354

Earlier quoted context omitted.

I don't get offended when people call my work a stochastic parrot. I just put them in the same bucket of intelligence as an 8b model and weight their inputs accordingly.

Right, DARVO.

[flagged]
Post reply on HN