Is it a possibillity that certain aspects of logic or human communication have a secret sauce that is not reliably replicable? I wonder if its expected to basically be able to perfectly interface with humans or logicical structures that flow from said imperfect beings, who are themselves deeply imperfect and irrational
Is the reversal curse in LLMs real?
181–190 of 211 posts
Re: Is the reversal curse in LLMs real?
#182Earlier quoted context omitted.
There is no "Acting Chancellor", that is a poor translation of what the job is. There is only a Chancellor, not an "Acting Chancellor" - you have to trust that the Germans and their institutions kind of know who was Chancellor and who was not. The machine is a complete moron for not being able to get basic data from basic sources right.
> The machine is a complete moron for not being able to get basic data from basic sources right. I'm just going to point out that you're calling an AI moronic for not getting this right when a bunch of humans are also disagreeing with you. Frankly, this really undermines your case that this is an AI failure.
What is wrong with me wanting my machines to stick to the actual definitions?
Re: Is the reversal curse in LLMs real?
#183Earlier quoted context omitted.
> but that the category of things we call apple also belongs to the category of things that are red No, in English this would be “apples are red”, not “the apple is red”.
I don't disagree though perhaps it's worth mentioning that sentences such as "the great spotted woodpecker is a medium-sized woodpecker" are used in English with the meaning "great spotted woodpeckers are medium-sized woodpeckers" so it seems to me that "apples are red" is grammatically a possible meaning of "the apple is red" even if it would be stylistically and pragmatically so weird that nobody would ever do that…
Except that apples can also be green, whereas great spotted woodpeckers are always a medium-sized woodpecker species. So 'the apple is red" implies at least a red variety of apples, never all apples, so "the" doesn't cover the whole group. But "the great spotted woodpecker is a medium-sized woodpecker" does cover the whole group (species). "The" can be either a definite or indefinite article, depending on context. That in turh changes "is" between equivalence and implication.
"An apple is red" does not imply that "Red is an apple". Use of the indefinite article makes the non-equivalence clearer.
Re: Is the reversal curse in LLMs real?
#184Earlier quoted context omitted.
Not a technicality. It's like inventing a proper head of government that wasn't there. Also, knowing German chancellors was always easy, so that fact that a machine falls behind any normal dictionary or the German Chancellor's website is poor form. It's like a calculator getting basic addition wrong. I don't want a future full of poorly performing machines.
He was acting Chancellor. It's like saying a red apple is not an apple. It's a technicality. The fact that we have a completely reasonable explanation to why GPT-4 includes him in the list and you guys still make it sound like the AI is a complete moron is just hilarious to me.
A "red apple" is an apple that is red. An "acting officeholder" is not an officeholder who is acting, but a person who is acting as if they have the office (... with institutional support - Norton was not acting Emperor).
It's a technicality, but a valid technically and it's a technical question. I agree that it's wrong in a relatively small and surprisingly human way, but it's wrong in a way that including a red apple amongst apples is not.
Re: Is the reversal curse in LLMs real?
#185Earlier quoted context omitted.
It is exactly what “acting X” means in English.
You can clearly see it is confusing the heck out of people here by seeing "acting X" as a version of X when it is not in this case here.
Yeah, but those people are just wrong. To be fair, it's not something most people have much direct experience with.
Re: Is the reversal curse in LLMs real?
#186Earlier quoted context omitted.
So the English Wikipedia author is a moron, or is there ambiguity when describing the role in English?
No. I think the problem starts with that the role Scheel had isn't given a name in the article of the basic law that creates it. So either need to be fully descriptive (e.g., something like fulfilling the functions of the Chancellor, while not ever having the office) or it will be also open to being misunderstood. The issue here really is that the German succession doesn't ever transfer the office, but only the funct…
I see, so when you said the machine is a moron for getting it wrong, the English editor isn’t a moron for getting it wrong due to ambiguity in the law?
I think there’s also a functionally useful way to describe someone’s role as what it functionally is even if it’s not legally that. You’re point is super well taken, if the machine is intended to be a fact oracle, it’s awfully loose and adds a lot of interpretation in areas of ambiguity.
I would say IMO that’s specifically the power of these machines. An awful lot of human endeavor doesn’t require literalism but semantic approximation and interpretation that machines were literally incapable of. Its weird language is enough to achieve that, but I think it’s overly restrictive to assert a broad lack of utility in critical systems. An awful lot of critical systems actually need more “probabilistic” interpretation than literal fact oracling.
Re: Is the reversal curse in LLMs real?
#187Earlier quoted context omitted.
What distinction are you making between 'acting X' and 'caretaker X'? I'm British, not American, so I'm not assuming something like the vice president automatically becomes president, we don't have that either. If anything it's even less than Germany since (at least in theory and history, modern media etc. makes it a bit different in practice) there's nothing special about the PM, it's just the governing party's lead…
The real issue here is that people stick to the word "Chancellor", not the "acting" as such. There is no named role of "Acting Chancellor" created in the basic law, instead someone is just tasked with performing certain duties (and people might sometimes call it "acting Chancellor" or whatever). However, the position "Chancellor" is a well defined & special role (unlike perhaps PM) and whatever that other "acting" ro…
Charles III was for years described as a 'King in waiting', nobody was confused that perhaps he was already king just because the phrase used that word.
I don't think in English usage there is any meaning attached to 'acting X' which 'caretaker X' (as you're happy to call it) doesn't also carry. Both are used interchangeably, the former you'd put on your CV, the latter might be used by the media when your employer put out the less release announcing it, but same thing: for some reason there is not currently an X, but you are fulfilling some necessary duties that that person would do in the meantime.
Re: Is the reversal curse in LLMs real?
#188Earlier quoted context omitted.
The real issue here is that people stick to the word "Chancellor", not the "acting" as such. There is no named role of "Acting Chancellor" created in the basic law, instead someone is just tasked with performing certain duties (and people might sometimes call it "acting Chancellor" or whatever). However, the position "Chancellor" is a well defined & special role (unlike perhaps PM) and whatever that other "acting" ro…
Charles III was for years described as a 'King in waiting', nobody was confused that perhaps he was already king just because the phrase used that word. I don't think in English usage there is any meaning attached to 'acting X' which 'caretaker X' (as you're happy to call it) doesn't also carry. Both are used interchangeably, the former you'd put on your CV, the latter might be used by the media when your employer pu…
When used in a narrow sense "acting X" can capture it, but then sometimes people start to use it somewhat interchangeably with "interim X" etc. and it depends on the specifics what happens in successions etc.
Re: Is the reversal curse in LLMs real?
#189I couldn't duplicate it in GPT4 here. It answered correctly, unless I posed it wrong: https://chat.openai.com/share/75d46a03-a223-4f3a-987d-f8fec3...
The paper is not talking about information provided in the context, only training and fine-tuning. > In fairness, it’s also worth pointing out here that they’re making the claim that the reversal curse only applies to training and fine-tuning and not in-context – i.e., putting all your information inside a prompt. They point out in a footnote that you can put A to B data in a prompt and GPT-4 will make B to A connect…
ah, I see. OK.
Re: Is the reversal curse in LLMs real?
#190Earlier quoted context omitted.
No. I think the problem starts with that the role Scheel had isn't given a name in the article of the basic law that creates it. So either need to be fully descriptive (e.g., something like fulfilling the functions of the Chancellor, while not ever having the office) or it will be also open to being misunderstood. The issue here really is that the German succession doesn't ever transfer the office, but only the funct…
I see, so when you said the machine is a moron for getting it wrong, the English editor isn’t a moron for getting it wrong due to ambiguity in the law? I think there’s also a functionally useful way to describe someone’s role as what it functionally is even if it’s not legally that. You’re point is super well taken, if the machine is intended to be a fact oracle, it’s awfully loose and adds a lot of interpretation in…
Maybe when read in a strictly narrow sense it is even a good translation. Practically, it is more of a caretaker role (especially in the case of Scheel here), though, and interpreting into the phrasing can be dangerous. Naming things that aren't named is always tough, and even tougher when needing to go from one language to another.
I'd agree that often things human are more loose but when there are narrow definitions ignoring them is dangerous.