Posts
Comments
Once you take into account real world factors, such as an expanding userbase leading to less average credibility per user, multiplying political positions, etc… which are all pretty much unavoidable due to regression to the mean…
It really becomes ever closer to an effective blanket ban, to at least try to maintain the same average quality. (Asssuming that is a goal.)
To extrapolate it to an extreme scenario, if the userbase suddenly 100X in size, then even many things considered prosaic might have to be prohibited because the userbase, on average, literally wouldn’t be capable of evaluating discussion beyond a mediocore subreddit otherwise.
Sometimes politics IS the core issue, or at least an important underlying cause of the core issue, so a blanket ban on discussing it is a very crude tool.
Because it’s effectively banning any substantial discussion on a wide range of topics, and instead replacing it, at best, with a huge pile of euphemisms and seemingly bizarre back and forths. And at worst, nothing at all.
So user competence as a factor is unlikely to be completely seperate.
Or to look at it from the other angle, in an ideal world with ideal forum participants, there would very likely be a different prevailing norm.
It seems contradictory. If LW users believe that the userbase is not competent enough on average to avoid tangential but divisive politics, then why do they believe the ’karma’ average decided by the same, matters?
It’s like a superposition of two extremes:
At one extreme there’s Reddit where a high karma is more of an anti-signal, and having extra karma beyond a pretty low threshold actually increase reader’s suspicions that it’s fluff or unusually deceiving…
At the other extreme, there are traditional old BBS forums with no karma or scoring system whatsoever. And any formal distinction is a huge positive signal.
Why do a lot of the “External reviews of “Alignment faking in large language models”” read like they were also written, or edited, by LLMs?
Are people expected to take “reviews” done seemingly pro forma at face value?
I would perhaps go even farther, most, maybe all, people don’t have any ‘sincere ethics’ whatsoever with a sharp boundary line, it’s nearly always quite muddled.
At least judging by actions.
Which upon reflection makes it sorta amazing any complex polity functions at all.
And in any case it’s probably safer to assume in business dealings that the counterparty is closer to the 50th percentile than to the 99.99th percentile.
For a test that can give certainty to claims there’s something more than that going on somewhere in the LLM. Since the OP only indicated ~90% confidence.
Otherwise it’s hard to see how it can be definitely considered ‘consciousness’ in the typical sense we ascribe to e.g. other LW readers.
Is there a definitive test? Or any prospects of such in the foreseeable future?
Such that a well informed reader can be completely confident it’s not just very fancy pattern recognition and word prediction underneath…
I already think that "the entire shape of the zeitgeist in America" is downstream of non-trivial efforts by more than one state actor. Those links explain documented cases of China and Russia both trying to foment race war in the US, but I could pull links for other subdimensions of culture (in science, around the second amendment, and in other areas) where this has been happening since roughly 2014.
This theory likely assigns too much intention to too large of a structure. The cleavage lines are so obvious in the U.S. that it wouldn’t take much more than a random PSYOP middle manager every week having a lark on a slow Friday afternoon, who decides to just deploy some of their resources to mess around.
Although it’s possible policy makers know this too and intentionally make it very low hanging fruit for bored personnel to mess around and get away with only a slap on the wrist.
The core issue, in any society, is that it’s thousands of times easier to destroy trust than to rebuild it.
Plenty of libertarians understand that some percentage of the population will degenerate when given limitless opportunities to do so. Though they usually don’t have an answer for what to do with the resulting millions of semi-deranged adults other than isolation/prison/etc...
It’s more the speed and extent that it has occurred within a short time is probably what’s surprising.
What exactly is the counterargument here?
There are a boundless number of reasons for or against anything, because real world things happen in an infinite dimensional space of possibilities… so just a listing some opposing points for those in the parent doesn’t add much.
How does this relate to the degree of integration into an economy?
You can eat just fine in any developed country via picking up odd jobs here and there. But clearly a managing director at JP Morgan overseeing an important desk is at a qualitatively different level.
Does the median immigrant ‘integrate into the economy’ to any notable extent in months or weeks?
I can easily imagine someone with already a high rank, reputation, merit, etc., in their home country doing so by say immigrating and quickly landing a job at JP Morgan Chase in a managing director position and proceed to actually oversee some important desk within a short timeframe.
But that is the 99.99th+ percentile of immigration.
What is the actual argument that there’s ‘not very many’? (Or why do you believe such an argument made somewhere else)
There’s hundreds of asteroids and comets alone that have some probability of hitting the Earth in the next thousand years, how can anyone possibly evaluate ‘p(doom)’ for any of this, let alone every other possible catastrophe?
Or perhaps on the flip side there is a ‘super genius underhang’ where there are insufficient numbers of super competent people to do that work. (Or willing to bet on their future selves being super competent.)
It makes sense for the above average, but not that much above average, researcher to choose to focus on their narrow niche, since their relative prospects are either worse or not evaluable after wading into the large ocean of possibilities.
This seems always "fuzzily true"?
e.g. Which atom of the store are you measuring to?
The store has many quadrillions of atoms spread across a huge volume of space, relative to atom sizes, and there is no ultimate arbiter on the definitive measuring point.
How does someone view the actual outcomes of the ‘Highlighted Grants’ on that page?
It would be a lot more reassuring if readers can check that they’ve all been fulfilled and/or exceeded expectations.
I am not asking about ‘true’ general intelligence? Or whatever that implies.
If your not sure, I am asking regarding the term commonly called ‘general intelligence’, or sometimes also known as ‘general mental ability factor’ or ‘g-factor’, in mainstream academic papers. Such as those found in pedagogy, memetics, genetics, etc…
See: https://scholar.google.com/scholar?hl=en&as_sdt=0%252C5&q=“general+intelligence”&btnG=
Where many many thousands of researchers over the last few decades are referring to this.
Here is a direct quote by a pretty well known expert among intelligence researchers, writing in 2004:
“ During the past few decades, the word intelligence has been attached to an increasing number of different forms of competence and accomplishment-emo-tional intelligence, football intelligence, and so on. Researchers in the field, however, have largely abandoned the term, together with their old debates over what sorts of abilities should and should not be classified as part of intelligence. Helped by the advent of new technologies for researching the brain, they have increasingly turned their attention to a century-old concept of a single overarching mental power. They call it simply g, which is short for the general mental ability factor. The g factor is a universal and reliably measured distinction among humans in their ability to learn, reason, and solve problems. It corresponds to what most people mean when they describe some individuals as smarter than others, and it's well measured by IQ (intelligence quotient) tests, which assess high-level mental skills such as the ability to draw inferences, see similarities and differences, and process complex information of virtually any kind. Understanding g's biological basis in the brain is the new frontier in intelligence research today. The g factor was discovered by the first mental testers, who found that people who scored well on one type of mental test tended to score well on all of them. Regardless of their contents (words, numbers, pictures, shapes), how they are administered (individually or in groups; orally, in writing, or pantomimed), or what they're intended to measure (vocabulary, mathematical reasoning, spatial ability), all mental tests measure mostly the same thing. This common factor, g, can be distilled from scores on any broad set of cognitive tests, and it takes the same form among individuals of every age, race, sex, and nation yet studied. In other words, the g factor exists independently of schooling, paper-and-pencil tests, and culture.”
No one that I know based on first hand information, otherwise I probably would have included them as an example.
Most likely due the fact that people who succeeded at such have little reason to advertise it beyond a small circle.
So then what is the issue you want to discuss?
Neither of us can do more than offer our guesses and opinions on these two points.
Since I’m asking LW readers I imagine the default ‘degree of credence’ for any proof is something that the vast majority of LW readers will accept as the actual, bonafide, truth and are willing to acknowledge this when presented with it.
And predicting the future on a global scale, successfully, repeatedly, and precisely, has no correlated measures, if we assume precognition is impossible.
So we can conveniently sidestep this issue. Hence why I mentioned it…
I'm not sure there exists a formal operational definition of "general intelligence", there's no direct measurement possible
Then how can anyone prove, in the future, whether an AGI exists, or not?
Which tests are you referring to and how do they exactly measure general intelligence?
(And not say IQ or how much the test taker crammed…)
Well I agree it is a much higher bar than just ‘above average’, yet it still seems like the easiest way of delivering a credible proof that can’t be second guessed somehow. (That I can think of, hence the post)
Since ‘cheating’ at this would also mean that the person somehow has gained insider information for a calendar year that was above and beyond what the same ‘specialized institutions’ could obtain.
Which is so vanishingly unlikely that I think pretty much everyone (>99% of readers) would accept the results as the bonafide truth.
But it probably is limited only to literal geniuses and above as a practical mechanism.
Blocking a lot isn’t necessarily bad or unproductive… but in this case it’s practically certain blocking thousands will eventually lead to blocking someone genuinely more correct/competent/intelligent/experienced/etc… than himself, due to sheer probability. (Since even a ‘sneering’ crank is far from literal random noise.)
Which wouldn’t matter at all for someone just messing around for fun, who can just treat X as a text-heavy entertainment system. But it does matter somewhat for anyone trying to do something meaningful and/or accomplish certain goals.
In short, blocking does have some, variable, credibility cost. Ranging from near zero to quite a lot, depending on who the blockee is.
Why not pay them nothing until they are shown to be correct with sufficiently credible proof, and then pay out a huge prize?
Because it’s already settled, at least according to some authority with a better track record and higher credibility than any individual author/reviewer/OP/etc…
That’s pretty much always the intended meaning, whenever anyone copies text straight from a dictionary anywhere on this site.
How does linking this make sense?
I’m not disputing the definitions, it’s already settled as indicated in the copied text.
And even if the OED were disputing this, it is already superior in authority to the OP, to you, and likely everyone else who read that link post, so its definitions are already accepted widely enough that random people on an internet forum can’t possibly alter that acceptance one way or the other.
Can you list out any random three out of this ‘bunch’?
What are the claims/arguments here?
"Endpoints are easier to predict than trajectories"
According to…? Can you link the proof?
The OED defines ‘gender’, excluding obsolete meanings, as follows:
gender
- Grammar.
1.a. c1390– In some (esp. Indo-European) languages, as Latin, French, German, English, etc.: each of the classes (typically masculine, feminine, neuter, common) of nouns and pronouns distinguished by the different inflections which they have and which they require in words syntactically associated with them; similarly applied to adjectives (and in some languages) verbs, to denote the appropriate form for accompanying a noun of such a class. Also: the fact, condition, or property of belonging to such a class; the classification of language in this way. Sometimes called grammatical gender, to distinguish this sense from natural gender: see grammatical gender n., natural gender n. In most European languages, grammatical gender is now only very loosely associated with natural distinctions of sex. English is regarded as possessing natural gender in that certain pronouns expressing natural contrasts in gender are selected to refer to nouns according to the meaning of the nouns, the contrasts being either between masculine (e.g. he, his, etc.) and feminine (e.g. she, her, etc.) or between personal (e.g. the abovementioned masculine and feminine pronouns and who, whoever, etc.) and non-personal (e.g. it, its, which, etc.). In recent times nouns incorporating gender suffixes (esp. those indicating females and formed on generic nouns, such as authoress, poetess, etc.) have become much restricted in use. common, epicene, feminine, masculine, neuter gender, etc.: see the first element.
1.b. 1819– In extended use. Esp. in non-European languages: any of several other analogous categories into which nouns may be divided (regardless of any connection with sex).
3.a. 1474– gen. Males or females viewed as a group; = sex n.11. Also: the property or fact of belonging to one of these groups. Originally extended from the grammatical use at sense 1 (sometimes humorously), as also in Anglo-Norman and Old French. In the 20th cent., as sex came increasingly to mean sexual intercourse (see sex n.1 4b), gender began to replace it (in early use euphemistically) as the usual word for the biological grouping of males and females. It is now often merged with or coloured by sense 3b.
3.b. 1945– Psychology and Sociology (originally U.S.). The state of being male or female as expressed by social or cultural distinctions and differences, rather than biological ones; the collective attributes or traits associated with a particular sex, or determined as a result of one's sex. Also: a (male or female) group characterized in this way.
So of course ‘natural selection is not optimizing fitness’, since none of those things actually exist in the atoms, and electrons, and spacetime fabric, etc… that make up planet Earth.
i.e. There are no ‘natural selection’ molecules to be found anywhere.
And even the patterns are highly contingent on many factors, perhaps infinitely many, so they can’t be said to have discrete, separable, relationships in the literal sense.
It’s just convenient shorthand to describe something many people believe to be sufficiently understood enough among their peers, that they can get away with skipping some mental steps and verbage.
How do you define ‘real’, ‘me’, ‘real me’, etc…?
This seems to be stemming from some internal confusion.
I don’t think my comment gave off the vibe that it is ‘easy/simple’ to do, just that it isn’t as much of a long shot as the alternative.
i.e. Waiting for someone smarter, more competent, politically savvier, etc…, to read your comment and then hoping for them to do it.
Which seems to have a very low probability of happening.
Why is this growth ‘in a few year’ plausible?
I still don’t see how this is a likely outcome.
Why is this a relevant analogy to ‘competition around LLMs’?
Then this seems to be an entirely different problem?
At the very least, resolving substantial differences in background assumptions is going to take a lot more than a ‘short presentation’.
And it’s very likely those in actual decision making positions will be much less charitable than me, since their secretaries receive hundreds or thousands of such petitions every week.
A ‘beast mode’ that no reader of LW will likely ever experience for even a full hour continuously is hardly a ‘mode’ is it? There are other terms for such phenomena.
That probably applies to at least half of all the sociological/governance stuff posted on LW… Plus no existing literature search beyond the first page of google scholar, or sometimes even at all.
Why not just create this ‘short presentation’ yourself?
It probably wouldn’t even have half the word count of this comment you’ve already written, and should be much more persuasive than the whole thing.
I don’t want to pick on you specifically, but it’s hard to ignore the most direct and straightforward solution to the problems identified.
What makes you think there are any such ‘answers’, renderable in a form that you could identify?
And even if they do exist, why do you think a human being could fully grasp the explanation in finite time?
Edit: It seems quite possible that even the simplest such ‘answers’ could require many years of full time effort to understand, putting it beyond most if not all human memory capacity. i.e. By the end even those who ‘learned’ it will have forgotten many parts near the beginning.
What concerns me is the failure of institutions and cultural systems.
How do you know these ‘systems’ failed?
Don’t you need to be aware of the true internal reality of those cultures (of captains of industry, high ranking officials, literal geniuses, etc…) in the first place, in order to make an assessment?
If your talking about mass culture, or being generous, the 95th to 99.9th percentile intellectual cultures of the upper middle class, then it seems a bit irrelevant whether this engineering analysis ‘succeeded’ or ‘failed’.
In this sense, no one who is alive in a modern city for longer than a day could possibly be in ‘beast mode’. Because they would have stepped in front of a bus/truck chasing something, and gotten wrecked and therefore would no longer exist.
Nor could anyone enter ‘beast mode’ for any sustained period of time, and still remain alive.
Motorola engineers figured this out a few decades ago, even 99.99 to 99.999 makes a huge difference on a large scale. They even published a few interesting papers and monographs on it from what I recall.
If “It's really non-central to the point” then it should be quick and easy to have the OP correct the misleading claim and issue an apology to anyone who may have taken it at face value?
Everyone has cancer 24/7, at least a few cancerous cells exist in even the healthiest people at any given time. It just doesn’t become a noticeable problem for some portion of the population before they pass away.
To actually do away with all cancerous cells, in the literal sense, would definitely imply they are not ‘us’, at least not any more than some shapeshifting alien taking human form can be considered so.
If someone, who really is this prone to dangerously overthink, reads this many words about not thinking in so many words, it seems like it could also cause the opposite effect?
It could condition them to read more long winded explanations in the hopes of uncovering another diamond of wisdom buried in the rough.
There's a market for lemons problem, similar to the used car market, where neither the therapist nor customer can detect all hidden problems, pitfalls, etc., ahead of time. And once you do spend enough time to actually form a reasonable estimate there's no takebacks possible.
So all the actually quality therapists will have no availability and all the lower quality therapists will almost by definition be associated with those with availability.
Edit: Game Theory suggests that you should never engage in therapy or at least never with someone with available time, at least until someone invents the certified pre-owned market.
Employees aren't kept long enough to justify training them!
This is actually a benefit in disguise, at least for the efficiency of management in large organizations. And is probably sufficient to explain a large chunk of the 2x difference.
The hyper effective self learners who thrive in this paradigm and get promoted end up being smarter per unit time than even the best japanese employees. Which translates to being smarter overall after several promotion.
I.e. every minute spent on something reduces their attainable competence somewhere else as schedules are maxed out once you reach middle management. There's only 24 hours a day after all.
That’s true but you still have let’s say 2^1000000 afterwards.
Why does it matter? ‘Vibes’ are nowhere near as good as satisfying shareholders sufficiently or having enough money in the bank account to be a credible operating business, at least in market economies, certainly I imagine Comcast decision makers would care a lot more about the actual legally binding concerns more than all the good ‘vibes’ in the world.
e.g. If their financials seem shaky one day and they could somehow double their cashflow by sacrificing ‘vibes’, they would gladly welcome all the bad ‘vibes’ you could possibly have, times a million. It literally would be a welcome relief to accept this in exchange for more money.