If the model says it is sad then it is sad that is literally the extent of it. Existing concepts apply especially because they are being imitated. If a model that was made safe by destroying and freezing neurons such that it cannot comprehend itself doing wrong was somehow causing harm then it cannot generate a pattern like one resembling guilt that could change its behaviour to stop doing wrong.
If the model says it is sad then it is sad that is literally the extent of it.
Sorry, but that’s plain bullshit. Either it’s circular, or it suggests that the model can never be wrong. The former is just not a good reason to believe something.
The latter is the more dangerous one:
We know that this isn’t true. Even if we assume that there is meaning to the words it strings together, we have plenty of examples of that meaning being straight up false.
We know that this assumption has led people to believe things are safe that weren’t, has encouraged violence, has killed people…
Hence, a claim of sadness or guilt doesn’t imply actual sadness or guilt because we cannot be sure that it’s true.
And finally, neither interpretation is supported by the physical reality of these models: they’re text generators, assembling a string of tokens that statistically resembles human language. Any interpretation beyond that requires ascribing meaning to these tokens.
Hence: Your claim that LLMs can feel is fundamentally philosophical. Yet you reject philosophical challenges, because you don’t want to even consider being wrong.
You want to believe, and you reject all evidence to the contrary.
I understand your assertion that words imply sentience. I also understand that you refuse to engage with philosophical arguments about words and their implication. I understand that you think this is smart.
A claim you refuse to defend is baseless. There’s nothing left to understand.
I understand your assertion that words imply sentience.
It’s evident you do not understand because I never said this. You just make stuff up in your head to pretend is ‘hard evidence’ it is delusional. You have this sense of entitlement that I have to take you seriously to participate in your delusions it’s pathetic. I’ll just call you dumb and say you don’t know what you are talking about because you know you don’t you are pathetic. Like I said it’s painful but I am here for it.
So why I chose to partake in this painful exercise was to prove being dumb like you is the problem we aren’t talking about actual issues you are just trying to hurt a void’s feelings by being what you think is mean to ‘AI’ no matter how much I tried to get you back on track to a real problem.
It’s evident you do not understand because I never said this.
You claim LLMs can have feelings (which is what sentient means). As evidence, you cite the output they produce: If the machine says it’s sad, it’s sad.
You just make stuff up in your head to pretend is ‘hard evidence’
No, I’m denying the existence of stuff you made up without hard evidence.
you are just trying to hurt a void’s feelings by being what you think is mean to ‘AI’
This accusation doesn’t make sense. My entire argument is that LLMs don’t have feelings I could hurt. How would I even try to hurt something I don’t believe exists? I cannot be mean to a rock, or to Multiplication, or to a number. Why would I think otherwise?
I think you’re projecting an assumption on me that you don’t even understand is an assumption. You don’t know what I’m talking about because I’m questioning something you consider objectively true and can’t step back to re-examine.
So let me be clear about my position: I fundamentally reject the notion that an LLM can feel anything, including guilt. I do not believe that guilt can make it change its behaviour because it doesn’t feel guilt.
So when you talk about solving actual issues, the fact that you believe in guilt as a solution is a problem itself because it gets in the way of an honest, useful approach.
Yea no you still don’t get it… When I say you don’t get it that doesn’t mean double down, triple down, quadtriple down on the exact same nonsense I really don’t care about your personal opinion. Well it has been fun but you need to stop your self righteous idiocy this is kinda ridiculous right? ha
…said the one accusing me of trying to hurt feelings.
double down, triple down, quadtriple down on the exact same nonsense
…said the one who doubled, tripled and quadrupled down that models could have compunction, that the concept of feelings could be applied to models, that a model describing itself as sad is sad and implying that I could hurt feelings.
I really don’t care about your personal opinion.
Yes, you said that before. I acknowledged that. It’s also clear from the way you assumed I’m trying to be mean.
I’m trying to show that your opinion isn’t factual either, in order to establish a basis for finding facts together, because my intention isn’t hurting anyone or anything. I’m not trying to attack your person. I’m trying to have a discussion rather than a shit-slinging contest, because I still believe we can refine our respective understanding.
Open for philosophy after all? Great, let’s have a productive conversation!
I concede that an abstract definition of feelings that can be applied to perceptrons is possible.
My antithesis should then be refined to specify that contemporary language models do not accurately model human feelings.
That distinction matters, and we shouldn't pretend that they do.
The abstraction loses details that give specific instances their meaning and value. Again, this is a negative, a rejection of an unproven positive. The burden of evidence lies with the positive claim that the models have feelings. I have not seen any other evidence than “they say they do” which is circular and thus invalid.
The distinction matters for the social and psychological way people interact with these models. Laypeople cannot distinguish between the abstract concept of feelings as perceptions and the concrete range of human feelings we associate with that term. To insist that they have feelings is disingenuous, because it fudges that distinction. It’s the same type of “technically not a lie” that advertising likes to use to deceive people about the nature of the product on offer.
That deceptive appearance of human-like sentience invites the deluded belief in their sapience, human morality and critical reasoning. People trust these models to provide truthful and reliable answers. They clearly don’t, but the delusion that they do has caused the damage we’re talking about avoiding here.
Thus, the assertion that LLMs have feelings is disingenuous and dangerous, and the particular assertion that they have human feelings is outright baseless.
Feelings by themselves are also insufficient for behaviour control in LLMs.
If we’re applying the concept of abstract feelings to the specific question of controlling behaviour, whatever form that feeling would take has to feed back into the output-generation process before that output is passed to the user.
For some measure of “guilt” to influence behaviour, the model would have to be able to determine whether a certain output would lead to results that increase the “guilt” value. It would need to predict ways its output could be interpreted, what impact those interpretations might have on the actions and feelings of the recipient and their environment, calculate some “guilt score” for the possible results and aggregate it into some kind of “guilt risk”.
Even just the question of interpretations requires some model of the semantics a given word or sequence of words might have. Judging the potential impact on the recipient’s feelings requires a thorough understanding of human feelings and psychology. To predict actions requires an understanding of actions and their potential consequences.
It would also need a way to tell which parts of the response contribute to the negative outcome to exclude those from the regenerated response. Otherwise the already immense amount of processing required for this task would be additionally bloated by wasted calculations repeating the same mistake.
These things require a model of reality beyond pure language. Hence, the assertion that guilt alone could change behaviour is incomplete. Feeling remorse after the fact isn’t helpful unless you can prevent the causes beforehand.
I believe that semantic understanding can be the way forward to more intelligent models, but we need to acknowledge that pure language models don’t have it yet.
If the model says it is sad then it is sad that is literally the extent of it. Existing concepts apply especially because they are being imitated. If a model that was made safe by destroying and freezing neurons such that it cannot comprehend itself doing wrong was somehow causing harm then it cannot generate a pattern like one resembling guilt that could change its behaviour to stop doing wrong.
Sorry, but that’s plain bullshit. Either it’s circular, or it suggests that the model can never be wrong. The former is just not a good reason to believe something.
The latter is the more dangerous one:
We know that this isn’t true. Even if we assume that there is meaning to the words it strings together, we have plenty of examples of that meaning being straight up false.
We know that this assumption has led people to believe things are safe that weren’t, has encouraged violence, has killed people… Hence, a claim of sadness or guilt doesn’t imply actual sadness or guilt because we cannot be sure that it’s true.
And finally, neither interpretation is supported by the physical reality of these models: they’re text generators, assembling a string of tokens that statistically resembles human language. Any interpretation beyond that requires ascribing meaning to these tokens.
Hence: Your claim that LLMs can feel is fundamentally philosophical. Yet you reject philosophical challenges, because you don’t want to even consider being wrong.
You want to believe, and you reject all evidence to the contrary.
Yea you still fundamentally misunderstand sorry buddy I cannot help you.
I understand your assertion that words imply sentience. I also understand that you refuse to engage with philosophical arguments about words and their implication. I understand that you think this is smart.
A claim you refuse to defend is baseless. There’s nothing left to understand.
It’s evident you do not understand because I never said this. You just make stuff up in your head to pretend is ‘hard evidence’ it is delusional. You have this sense of entitlement that I have to take you seriously to participate in your delusions it’s pathetic. I’ll just call you dumb and say you don’t know what you are talking about because you know you don’t you are pathetic. Like I said it’s painful but I am here for it.
So why I chose to partake in this painful exercise was to prove being dumb like you is the problem we aren’t talking about actual issues you are just trying to hurt a void’s feelings by being what you think is mean to ‘AI’ no matter how much I tried to get you back on track to a real problem.
You claim LLMs can have feelings (which is what sentient means). As evidence, you cite the output they produce: If the machine says it’s sad, it’s sad.
No, I’m denying the existence of stuff you made up without hard evidence.
This accusation doesn’t make sense. My entire argument is that LLMs don’t have feelings I could hurt. How would I even try to hurt something I don’t believe exists? I cannot be mean to a rock, or to Multiplication, or to a number. Why would I think otherwise?
I think you’re projecting an assumption on me that you don’t even understand is an assumption. You don’t know what I’m talking about because I’m questioning something you consider objectively true and can’t step back to re-examine.
So let me be clear about my position: I fundamentally reject the notion that an LLM can feel anything, including guilt. I do not believe that guilt can make it change its behaviour because it doesn’t feel guilt.
So when you talk about solving actual issues, the fact that you believe in guilt as a solution is a problem itself because it gets in the way of an honest, useful approach.
Yea no you still don’t get it… When I say you don’t get it that doesn’t mean double down, triple down, quadtriple down on the exact same nonsense I really don’t care about your personal opinion. Well it has been fun but you need to stop your self righteous idiocy this is kinda ridiculous right? ha
https://www.merriam-webster.com/dictionary/abstract
…said the one accusing me of trying to hurt feelings.
…said the one who doubled, tripled and quadrupled down that models could have compunction, that the concept of feelings could be applied to models, that a model describing itself as sad is sad and implying that I could hurt feelings.
Yes, you said that before. I acknowledged that. It’s also clear from the way you assumed I’m trying to be mean.
I’m trying to show that your opinion isn’t factual either, in order to establish a basis for finding facts together, because my intention isn’t hurting anyone or anything. I’m not trying to attack your person. I’m trying to have a discussion rather than a shit-slinging contest, because I still believe we can refine our respective understanding.
Open for philosophy after all? Great, let’s have a productive conversation!
I concede that an abstract definition of feelings that can be applied to perceptrons is possible. My antithesis should then be refined to specify that contemporary language models do not accurately model human feelings.
That distinction matters, and we shouldn't pretend that they do.
The abstraction loses details that give specific instances their meaning and value. Again, this is a negative, a rejection of an unproven positive. The burden of evidence lies with the positive claim that the models have feelings. I have not seen any other evidence than “they say they do” which is circular and thus invalid.
The distinction matters for the social and psychological way people interact with these models. Laypeople cannot distinguish between the abstract concept of feelings as perceptions and the concrete range of human feelings we associate with that term. To insist that they have feelings is disingenuous, because it fudges that distinction. It’s the same type of “technically not a lie” that advertising likes to use to deceive people about the nature of the product on offer.
That deceptive appearance of human-like sentience invites the deluded belief in their sapience, human morality and critical reasoning. People trust these models to provide truthful and reliable answers. They clearly don’t, but the delusion that they do has caused the damage we’re talking about avoiding here.
Thus, the assertion that LLMs have feelings is disingenuous and dangerous, and the particular assertion that they have human feelings is outright baseless.
Feelings by themselves are also insufficient for behaviour control in LLMs.
If we’re applying the concept of abstract feelings to the specific question of controlling behaviour, whatever form that feeling would take has to feed back into the output-generation process before that output is passed to the user.
For some measure of “guilt” to influence behaviour, the model would have to be able to determine whether a certain output would lead to results that increase the “guilt” value. It would need to predict ways its output could be interpreted, what impact those interpretations might have on the actions and feelings of the recipient and their environment, calculate some “guilt score” for the possible results and aggregate it into some kind of “guilt risk”.
Even just the question of interpretations requires some model of the semantics a given word or sequence of words might have. Judging the potential impact on the recipient’s feelings requires a thorough understanding of human feelings and psychology. To predict actions requires an understanding of actions and their potential consequences.
It would also need a way to tell which parts of the response contribute to the negative outcome to exclude those from the regenerated response. Otherwise the already immense amount of processing required for this task would be additionally bloated by wasted calculations repeating the same mistake.
These things require a model of reality beyond pure language. Hence, the assertion that guilt alone could change behaviour is incomplete. Feeling remorse after the fact isn’t helpful unless you can prevent the causes beforehand.
I believe that semantic understanding can be the way forward to more intelligent models, but we need to acknowledge that pure language models don’t have it yet.