> So it seems I want the package of applause for insights that I think actually deserve applause.
Adam Smith in the Theory of Moral Sentiments: “Very few men can be satisfied with their own private consciousness that they have attained those qualities, or performed those actions, which they admire and think praise-worthy in other people; unless it is, at the same time, generally acknowledged that they possess the one, or have performed the other; or, in other words, unless they have actually obtained that praise which they think due both to the one and to the other.”
So you want genuine, high quality social validation (even if delayed). In other words, you want your contribution to a particular slice of human culture to be acknowledged in a way that you would find believable and satisfactory, which can only happen if it's about your chosen field, glorious insights.
Why a social creature would want social validation isn't an interesting question though.
What's interesting is why you chose glorious insights as your life's work?
I've also landed on the conclusion that "I want to be admired, but only for that which I think is admirable".
I think a lot of what makes me *me* is how specifically I define the latter, and how strictly I only care to be admired for that in particular.
And a core feature of what I think is admirable is the extent to which I hold myself accountable for my own beliefs (or something; not sure how to say that). So in the end, by recursing on what I think is admirable to admire, what I want ends up being quite distinct from what I will actually be admired for. Like a self-pollinating flower clade freed to explore deep into novel territory due to not having to worry about staying compatible with its conspecifics.
(Note: I struggled to put that into sentences, and much-focused on trying to make it sound pretty and profound, so fitness to actually-good-explanation was sacrificed a little... Sorry! ^^')
First, I think this question is at the heart of what separates AI from AGI. AIs, as we build them today, have very simple goal functions: Generally to reduce the error between an output and the desired output. An LLM is just iteratively predicting the next token.
Humans by contrast have remarkably varied goal functions. You listed yours but in truth you are leaving a lot of things out. If I put you in a room at 10 degrees Kelvin, or stranded you on a boat in the middle of the ocean, your goal function would rapidly change. In fact humans have so many goals and subgoals that it's a daunting prospect to even itemize them all.
Anyway, this stark difference between AI goals and human goals is I think the crux of why we can't make something convincingly human-like. We don't really know how to build something with such a varied set of goals as a person. All our methods of engineering rely on having fairly simple goals for the thing we're trying to build.
Second, I think this question of "what do I want" is probably the one for which my own consciousness - that submodule in my brain that excels at rationalizing and post-facto explanations - is probably the worst source of answers. Research shows that when you ask people what makes them happy, and then measure what makes them happy, the overlap is very low. Such questions are invitations for conscious self deception.
So my approach is to forget about what I think I want, and just observe what I do. And then try to infer from that some underlying decision process (i.e., what I actually want).
On that basis, what I want seems to vary depending on what's happening at the moment. Maslow's hierarchy is a crude approximation of the wants that can activate depending on circumstances. But I think at the highest level, when the base desires aren't at the forefront, I seem to be attracted to understanding the truth, especially the objective truth of the outside world, and trying to have ideas that have never been thought before.
I was also thinking of Maslow's Pyramid of needs and as far as I am concerned the top of the pyramid is mostly at the center of my deepest aims. That is why a lot of the discourses in western culture today fall too short for me. If you aim deep enough you begin to understand that your selfconcept and what your goals that are derived from this concept will never satisfy you. You begin to tap into the transpersonal. You begin to recognize that the phenomena which you nobody can really explain - like consciousness, love, dreams, intuitions, creativity etc. - are always with us but needed to develop and respected. From this understanding nearly everything is possible, if we want to.
I hear a meat robot frustrated with his own programming and desiring transcendence but realizing that that very desire is a part of his programming. Maybe I'm projecting? I'd like to live well and die and wake and find that the whole experience was just a VR-like homework assignment; to find that I was actually some kind of other superior being living in a universe more attuned to human desires.
I think everyone's "real" desires might be summarized as "granting more power to the truth." People want more power for themselves because they think they have the truth. They want more power for the political or religious or scientific causes or leaders that they think are based on the truth.
If a person had lots of power, but used it to act based on what they later recognize were falsehoods, then in hindsight that person would be ashamed of what they had done and would wish they had not had that power at that time. So, people don't want power for themselves unconditionally; they only want power to the extent that they would be able to use it to promote what is true.
I have a motto: "May the truth move you." This means: may it move you emotionally, and move you to action, and may you be free from falsehoods.
What about feelings - pleasure or happiness? Don't we want those, in addition to truth or power? We do want them, but we can understand happiness as the feeling we get when our will is fulfilled (and we don't want our will fulfilled unless we are acting based on the truth). Most would refuse to be placed in a happiness box where they were artificially forced to experience happiness while the world was destroyed around them. We only want happiness because it is an indicator of will fulfillment = truth having power.
> I think everyone's "real" desires might be summarized as "granting more power to the truth." People want more power for themselves because they think they have the truth.
I think that someday, you're going to get a harsh wake-up call to the stomach.
So you're saying, I think, that some people are nasty and don't care whether they're right or wrong as long as they have power.
What do you think is the distinction between what someone wants, and what they "really" want? I understand this distinction as: what you "really" want is what you would want if you were wiser/more informed. Someone may say "I want X" and someone else may say "no, you don't really want X, because X has bad consequences Y," and that second person may be right. What you "really" want may be different from what you think you want.
I think that if the nasty people were wiser, they would care more about the truth.
That is indeed the point we disagree on. I think the disagreement is actually hidden inside the word "wiser". To me, the word "wisdom" seems to be a word people only use to mean something like "like being smart, except for not actually being smart". I think "wisdom" is used to dishonestly legitimize conventional beliefs that can't be legitimized by appeal to reason.
Wisdom involves recognizing situations as similar to those you've seen before or heard about, and on that basis, predicting the likely outcome or understanding how best to deal with it.
Wisdom often deals with social and political situations. How is this person going to treat me? Will I be better off quitting this job? How should I speak to this person so they understand me? What's going to happen to my country? Is this a dangerous street to walk down? How can I make my wife happy?
Wisdom often deals with personal choices and virtues. Should I buy an insurance policy with this TV? Should I sue this person? Should I yell at this cashier? Do I just have to be patient here? Will this company treat its customers well or exploit them? Am I just deceiving myself?
Wisdom includes knowledge that is broadly applicable to many situations. For instance, "if something changes or you try something new, mistakes and accidents are more likely during the settling-in period." "When evaluating an extraordinary claim, beware of cherry picking bias." "If something is fully optimized for desirable criterion A, then it's probably not as optimized for desirable criterion B." "Initial cost estimates are usually too optimistic." "When a measure becomes a target, it ceases to be a useful measure."
Wisdom is distinguished from intelligence. Cognitive speed, short term memory, vocabulary, creativity - these traits may be useful for attaining wisdom, but they are not themselves wisdom. Wisdom is accumulated, crystallized knowledge, independent of how quickly or adroitly you think.
Wisdom is distinguished from mere factual information. Knowing all the state capitals is not wisdom. Wisdom can draw on factual information, but it necessarily involves a kind of "fuzzy" future prediction, not just factual certainty. Wisdom has to be generally useful for navigating life.
Those are all things I'd call intelligence. And I'd say it isn't distinct from "mere" factual information; it relies heavily on information! Experience is nothing but information about what happened in the past. You're cherry-picking an example (state capitals) of famously useless information that children are forced to memorize.
I would taboo the word "factual" in "factual information". Information is not factual; it's statistical. The mathematical definition of information is literally statistical. The word "factual" is rarely 100% accurate, and it draws in a host of philosophical critiques which are irrelevant to information in the more-general statistical sense.
> And I'd say it isn't distinct from "mere" factual information; it relies heavily on information!
As I said, it can rely on factual information, but wisdom must involve a kind of fuzzy future prediction, not just factual certainty.
You don't like state capitals as an example because state capitals are useless; very well, then I'll say that knowing the syntax of Python is not wisdom. That's useful knowledge, and wisdom can be built upon it, but the syntax of Python is still mere factual information and not itself wisdom. Wisdom that a person might build upon the syntax of Python would involve programming design principles like DRY. DRY is not strictly factual; there's an element of fuzzy foresight to it.
There are several ways to approach and dissect the term "information." The statistical, information-theoretic sense is just one such way. A competing major approach would be that of semiotics, which describes information in terms of signs. Most fundamentally, and informally, information is that which has the power to inform.
The term "factual information" means information conveying facts. For instance, if I tell you that the Moon has roughly 1/2 degree apparent diameter to an observer on Earth, this is factual information because by sending it to you I have conveyed a fact.
A fact is a slightly ambiguous term. I mean it in the ordinary, informal sense. Roughly, a fact is something that is true, and that we can say is true with high confidence, without making too many inferential or theoretical leaps from our direct observations. Wisdom, in contrast to knowledge of facts, involves uncertain theories and rules-of-thumb, and generally leads to predictions about the future, which we cannot directly observe.
>So which of these things is the what we really want?
This seems like the wrong question. What matters isn't just what we want, it's what we're actually motivated to achieve, which involves taking into account both the value to us and the likely effort required. So, in a sense, what we "really" want is to continually find and take advantage of the best opportunities by the [value] / [effort] metric.
I just want to keep on living and feel entertained while doing so. As for anything deeper beyond that, I wish I knew. May many passions seem to just fall under the umbrella of keeping myself entertained: creating enough of a worthwhile struggle in my life that I constantly have to keep myself on my toes. I guess I am close to what Camus meant.
Realistically, an endless supply of low effort easy attractive women, good food, drugs and alcohol, and warm weather while not having to work. Not sure why people have to over complicate things. Oh and I'd like my kids to show up to my funeral and actually be sad that I passed.
Of course one needs good health, etc to realize that but that's a secondary affect.
I’m inclined to think the idea there’s something people ‘really want’ beneath all else is a metaphysical requirement rather than the basis for a discovery. Thinking of Wittgenstein’s similar idea here from the PI: 334. “So you really wanted to say . . .” – We use this phrase in order to lead someone
from one form of expression to another. One is tempted to use the following picture: what
he really ‘wanted to say’, what he ‘meant’, was already present in his mind even before
we articulated it. Various kinds of thing may persuade us to give up one expression and
Leaning into evolution, I want the maximum number of children (and dogs) with which I have great relationships, i.e. "family is wealth".
Large, society-level insights or contributions for a person in my position seem unlikely, though I'm glad people like you are doing what they can, especially for such huge problems as reproduction rates.
The most common way to inflate achievements is by falsifying your supposed disadvantaged beginnings. I have lost count of those who claim they grew up in a single parent household. The truth is more commonly that they grew up in a normal two-parent household but the mother never got round to legalising the marriage - technically remained "a single mom"
> So it seems I want the package of applause for insights that I think actually deserve applause.
Adam Smith in the Theory of Moral Sentiments: “Very few men can be satisfied with their own private consciousness that they have attained those qualities, or performed those actions, which they admire and think praise-worthy in other people; unless it is, at the same time, generally acknowledged that they possess the one, or have performed the other; or, in other words, unless they have actually obtained that praise which they think due both to the one and to the other.”
So you want genuine, high quality social validation (even if delayed). In other words, you want your contribution to a particular slice of human culture to be acknowledged in a way that you would find believable and satisfactory, which can only happen if it's about your chosen field, glorious insights.
Why a social creature would want social validation isn't an interesting question though.
What's interesting is why you chose glorious insights as your life's work?
I've also landed on the conclusion that "I want to be admired, but only for that which I think is admirable".
I think a lot of what makes me *me* is how specifically I define the latter, and how strictly I only care to be admired for that in particular.
And a core feature of what I think is admirable is the extent to which I hold myself accountable for my own beliefs (or something; not sure how to say that). So in the end, by recursing on what I think is admirable to admire, what I want ends up being quite distinct from what I will actually be admired for. Like a self-pollinating flower clade freed to explore deep into novel territory due to not having to worry about staying compatible with its conspecifics.
(Note: I struggled to put that into sentences, and much-focused on trying to make it sound pretty and profound, so fitness to actually-good-explanation was sacrificed a little... Sorry! ^^')
I have two thoughts on this topic.
First, I think this question is at the heart of what separates AI from AGI. AIs, as we build them today, have very simple goal functions: Generally to reduce the error between an output and the desired output. An LLM is just iteratively predicting the next token.
Humans by contrast have remarkably varied goal functions. You listed yours but in truth you are leaving a lot of things out. If I put you in a room at 10 degrees Kelvin, or stranded you on a boat in the middle of the ocean, your goal function would rapidly change. In fact humans have so many goals and subgoals that it's a daunting prospect to even itemize them all.
Anyway, this stark difference between AI goals and human goals is I think the crux of why we can't make something convincingly human-like. We don't really know how to build something with such a varied set of goals as a person. All our methods of engineering rely on having fairly simple goals for the thing we're trying to build.
Second, I think this question of "what do I want" is probably the one for which my own consciousness - that submodule in my brain that excels at rationalizing and post-facto explanations - is probably the worst source of answers. Research shows that when you ask people what makes them happy, and then measure what makes them happy, the overlap is very low. Such questions are invitations for conscious self deception.
So my approach is to forget about what I think I want, and just observe what I do. And then try to infer from that some underlying decision process (i.e., what I actually want).
On that basis, what I want seems to vary depending on what's happening at the moment. Maslow's hierarchy is a crude approximation of the wants that can activate depending on circumstances. But I think at the highest level, when the base desires aren't at the forefront, I seem to be attracted to understanding the truth, especially the objective truth of the outside world, and trying to have ideas that have never been thought before.
I agree that in other contexts other priorities would come to the fore.
I was also thinking of Maslow's Pyramid of needs and as far as I am concerned the top of the pyramid is mostly at the center of my deepest aims. That is why a lot of the discourses in western culture today fall too short for me. If you aim deep enough you begin to understand that your selfconcept and what your goals that are derived from this concept will never satisfy you. You begin to tap into the transpersonal. You begin to recognize that the phenomena which you nobody can really explain - like consciousness, love, dreams, intuitions, creativity etc. - are always with us but needed to develop and respected. From this understanding nearly everything is possible, if we want to.
I hear a meat robot frustrated with his own programming and desiring transcendence but realizing that that very desire is a part of his programming. Maybe I'm projecting? I'd like to live well and die and wake and find that the whole experience was just a VR-like homework assignment; to find that I was actually some kind of other superior being living in a universe more attuned to human desires.
I think everyone's "real" desires might be summarized as "granting more power to the truth." People want more power for themselves because they think they have the truth. They want more power for the political or religious or scientific causes or leaders that they think are based on the truth.
If a person had lots of power, but used it to act based on what they later recognize were falsehoods, then in hindsight that person would be ashamed of what they had done and would wish they had not had that power at that time. So, people don't want power for themselves unconditionally; they only want power to the extent that they would be able to use it to promote what is true.
I have a motto: "May the truth move you." This means: may it move you emotionally, and move you to action, and may you be free from falsehoods.
What about feelings - pleasure or happiness? Don't we want those, in addition to truth or power? We do want them, but we can understand happiness as the feeling we get when our will is fulfilled (and we don't want our will fulfilled unless we are acting based on the truth). Most would refuse to be placed in a happiness box where they were artificially forced to experience happiness while the world was destroyed around them. We only want happiness because it is an indicator of will fulfillment = truth having power.
> I think everyone's "real" desires might be summarized as "granting more power to the truth." People want more power for themselves because they think they have the truth.
I think that someday, you're going to get a harsh wake-up call to the stomach.
So you're saying, I think, that some people are nasty and don't care whether they're right or wrong as long as they have power.
What do you think is the distinction between what someone wants, and what they "really" want? I understand this distinction as: what you "really" want is what you would want if you were wiser/more informed. Someone may say "I want X" and someone else may say "no, you don't really want X, because X has bad consequences Y," and that second person may be right. What you "really" want may be different from what you think you want.
I think that if the nasty people were wiser, they would care more about the truth.
That is indeed the point we disagree on. I think the disagreement is actually hidden inside the word "wiser". To me, the word "wisdom" seems to be a word people only use to mean something like "like being smart, except for not actually being smart". I think "wisdom" is used to dishonestly legitimize conventional beliefs that can't be legitimized by appeal to reason.
Wisdom involves recognizing situations as similar to those you've seen before or heard about, and on that basis, predicting the likely outcome or understanding how best to deal with it.
Wisdom often deals with social and political situations. How is this person going to treat me? Will I be better off quitting this job? How should I speak to this person so they understand me? What's going to happen to my country? Is this a dangerous street to walk down? How can I make my wife happy?
Wisdom often deals with personal choices and virtues. Should I buy an insurance policy with this TV? Should I sue this person? Should I yell at this cashier? Do I just have to be patient here? Will this company treat its customers well or exploit them? Am I just deceiving myself?
Wisdom includes knowledge that is broadly applicable to many situations. For instance, "if something changes or you try something new, mistakes and accidents are more likely during the settling-in period." "When evaluating an extraordinary claim, beware of cherry picking bias." "If something is fully optimized for desirable criterion A, then it's probably not as optimized for desirable criterion B." "Initial cost estimates are usually too optimistic." "When a measure becomes a target, it ceases to be a useful measure."
Wisdom is distinguished from intelligence. Cognitive speed, short term memory, vocabulary, creativity - these traits may be useful for attaining wisdom, but they are not themselves wisdom. Wisdom is accumulated, crystallized knowledge, independent of how quickly or adroitly you think.
Wisdom is distinguished from mere factual information. Knowing all the state capitals is not wisdom. Wisdom can draw on factual information, but it necessarily involves a kind of "fuzzy" future prediction, not just factual certainty. Wisdom has to be generally useful for navigating life.
Those are all things I'd call intelligence. And I'd say it isn't distinct from "mere" factual information; it relies heavily on information! Experience is nothing but information about what happened in the past. You're cherry-picking an example (state capitals) of famously useless information that children are forced to memorize.
I would taboo the word "factual" in "factual information". Information is not factual; it's statistical. The mathematical definition of information is literally statistical. The word "factual" is rarely 100% accurate, and it draws in a host of philosophical critiques which are irrelevant to information in the more-general statistical sense.
> And I'd say it isn't distinct from "mere" factual information; it relies heavily on information!
As I said, it can rely on factual information, but wisdom must involve a kind of fuzzy future prediction, not just factual certainty.
You don't like state capitals as an example because state capitals are useless; very well, then I'll say that knowing the syntax of Python is not wisdom. That's useful knowledge, and wisdom can be built upon it, but the syntax of Python is still mere factual information and not itself wisdom. Wisdom that a person might build upon the syntax of Python would involve programming design principles like DRY. DRY is not strictly factual; there's an element of fuzzy foresight to it.
There are several ways to approach and dissect the term "information." The statistical, information-theoretic sense is just one such way. A competing major approach would be that of semiotics, which describes information in terms of signs. Most fundamentally, and informally, information is that which has the power to inform.
The term "factual information" means information conveying facts. For instance, if I tell you that the Moon has roughly 1/2 degree apparent diameter to an observer on Earth, this is factual information because by sending it to you I have conveyed a fact.
A fact is a slightly ambiguous term. I mean it in the ordinary, informal sense. Roughly, a fact is something that is true, and that we can say is true with high confidence, without making too many inferential or theoretical leaps from our direct observations. Wisdom, in contrast to knowledge of facts, involves uncertain theories and rules-of-thumb, and generally leads to predictions about the future, which we cannot directly observe.
Wisdom is predictive ability?
>So which of these things is the what we really want?
This seems like the wrong question. What matters isn't just what we want, it's what we're actually motivated to achieve, which involves taking into account both the value to us and the likely effort required. So, in a sense, what we "really" want is to continually find and take advantage of the best opportunities by the [value] / [effort] metric.
I just want to keep on living and feel entertained while doing so. As for anything deeper beyond that, I wish I knew. May many passions seem to just fall under the umbrella of keeping myself entertained: creating enough of a worthwhile struggle in my life that I constantly have to keep myself on my toes. I guess I am close to what Camus meant.
To be honest become immortal and rule the universe
Immortality is a possibility, but not one I can do much about. That other one is not much of a possibility.
I would say even the first is rather unlikely as even eternal health doesn’t guarantee immortality.
https://www.paulgraham.com/want.html
Status. Some prestige status, some dominance status. But all want status.
Realistically, an endless supply of low effort easy attractive women, good food, drugs and alcohol, and warm weather while not having to work. Not sure why people have to over complicate things. Oh and I'd like my kids to show up to my funeral and actually be sad that I passed.
Of course one needs good health, etc to realize that but that's a secondary affect.
I’m inclined to think the idea there’s something people ‘really want’ beneath all else is a metaphysical requirement rather than the basis for a discovery. Thinking of Wittgenstein’s similar idea here from the PI: 334. “So you really wanted to say . . .” – We use this phrase in order to lead someone
from one form of expression to another. One is tempted to use the following picture: what
he really ‘wanted to say’, what he ‘meant’, was already present in his mind even before
we articulated it. Various kinds of thing may persuade us to give up one expression and
to adopt another in its place.
Do you remember Mr. Morden, from the mid 90's TV show "Babylon 5"?
For myself, I answer: I want a world where my kids have the -choice- to have a good life. Everything else is a downstream implication from that.
Leaning into evolution, I want the maximum number of children (and dogs) with which I have great relationships, i.e. "family is wealth".
Large, society-level insights or contributions for a person in my position seem unlikely, though I'm glad people like you are doing what they can, especially for such huge problems as reproduction rates.
The most common way to inflate achievements is by falsifying your supposed disadvantaged beginnings. I have lost count of those who claim they grew up in a single parent household. The truth is more commonly that they grew up in a normal two-parent household but the mother never got round to legalising the marriage - technically remained "a single mom"
Yes, a common tactic.