YoungAchamian
We walk conditioned ground and name our folly civilization.
No bio...
User ID: 680
YoungAchamian
We walk conditioned ground and name our folly civilization.
2hr ago
First, I’ve never heard of a school tilting admissions based on intended major at the undergraduate level (Ivy plus). Because it’s obvious extremely easy to just lie.
Then you must have had your head in the sand this past decade, idk what to tell you. You strictly have not been paying attention. Lying is entirely the point. Ivy's are selecting for if you are willing and can play the game, signaling elite social understanding. Schools absolutely discriminate on demographics based on intended major.
Second, the entire Ivy League except Dartmouth is at best 50-50 male to female now.
Again bad stats. You are stating the the general population should be used as the judge for a localized population. This is the pay-gap stats slight of hand all over again. Also she didn't go into CS now, she went in 2007-2011. Back in peak CS gender discrimination territory.
Purdue is 8th. No one is choosing Purdue Engineering over Dartmouth engineering
I did. And I went to another top 10 from that list as my master's program.
The IQ gap between a Dartmouth engineer and a Purdue engineer is gigantic
I have met and worked with many engineers, undergrads and PhDs, at FAANGs, Research companies, internships, etc. From prestigious schools (MIT, Stanford, Brown, Princeton, Harvard, Yale, CMU) and good-state schools (Purdue, Georgia Tech, Texas AM, Urbana, Iowa, Wisconsin Madison, Michigan), and I can tell you that at no time was undergrad school education ever correlated with perceived IQ. Let alone 20+ IQ points, which is the difference between an average IQ and gifted. Last time I was seriously tested I think I got a 129(?) converted via a Raven's Progressive Matrices test.
OpenAI isn’t hiring from Purdue.
Any evidence of that? I have classmates that went into FAANG companies, Tesla, SpaceX, NASA, Palantir, and those are just the few I remember. Considering Murati's first job was Tesla, it looks like haughty Dartmouth is in the same hiring bucket as lowly baseborn Purdue lol. A quick scroll through OpenAI's LinkedIn, and about 40% of its workforce comes from public state schools. A ton come from Stanford, but that should be a shock to no one, given that Stanford is in the Bay where OpenAI is headquartered, locality matters is probably the overwhelming thing that I have learned. Companies hire from local schools more often than not.
The lower end engineering schools are preferred for like basic engineering
Lucky for you that I'm the one building all the Military AI weapons then, with your opinion, they obviously won't work very well and thus the world is safe.
SV is going to take the candidate with demonstrated higher IQ everyday
Yeah... have been in the bay, have been hired by companies in the bay, have left the bay. I don't think you know what you are talking about. This reads as a non-American or supremely closeted person with a poor understanding of America or really the world.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5hr ago
with very innovative architecture
Didn't you give me a lot of shit about a year back for believing that developing novel and new ML architectures was still practical and useful? I think your paraphrased words were "Data is all that matters, anyone believing the architecture advancements matter is an imbecile" You going to walk that back now?
as rare as you make it out to be either.
I don't think being a prod manger or proj manager out of school is rare, I have classmates that did it. But they aren't then parlaying that into becoming CTOs of AI research companies. Her trajectory is weird. Becoming a CTO at 33 after 2-3 years PM-ing at Tesla, 2 years at a startup, followed by being a VP at 29 is really fishy.
I'd register that she's an aggressive mercenary social climber in lieu of being a high quality engineer or PM
YoungAchamian
We walk conditioned ground and name our folly civilization.
14hr ago
She's a white woman in engineering, emphasis woman. There is never enough women of any race in engineering, she meets the quota. Doubly so at some Ivy League where they explicitly discriminate for it. I doubt she paid a lot out of pocket or was sufficiently put out by it.
would probably rank Dartmouth as my 9th choice after Harvard, Yale, Princeton, Stanford, Caltech, MIT, University of Chicago, and Colombia. Maybe CMU
We have very different definitions of good engineering schools. You seem to be ranking them based on international prestige. The Ivies aren't great engineering schools. They aren't horrible, you'll get a great education, but they don't crack the top ten, though I guess this list includes Princeton. Ivies are schools for kids of the elite. Elites don't go become engineers.
YoungAchamian
We walk conditioned ground and name our folly civilization.
1d ago
It's probably not nepotism in the strict definitional sense. But I've almost never heard of someone coming out of undergrad straight into project management at a prestigious company. Let alone from a non top 10 engineering school. Call it connections, familial or platonic. But she got into her positions based on some amount of capability and likely a large amount of "who you know" and was at least capable enough to build on that. But my realization was that technical knowledge and skill and actually attempting to do the work will not get as far as knowing the right people, being able to schmooze, and social climb your way into a prestigious position. Probably a bit of sour grapes.
YoungAchamian
We walk conditioned ground and name our folly civilization.
1d ago
Mira Murati actually does research
Does she? Her first job was PM at Tesla, followed by some startup roles (from memory) before becoming an executive and then CTO. Now she runs her own AI company. That trajectory screams manager/business not AI researcher. Just the kind of area where being a 7+ and able to talk nerdy is super useful. I remember looking at her career path years ago and realizing I could never nepo-baby that hard.
YoungAchamian
We walk conditioned ground and name our folly civilization.
1d ago
Wokeism - Yes
Anti-Wokeism - Yes: Though I would suggest that Anti-Wokeism isn't really a belief system. Adversity to Wokeism derives from competing political belief systems, within the Christian Memeplex
Communism - Yes
Libertarianism - Yes
Fascism - No, Fascism is distinctly trying to return to pre-Christian Germanic pagan morality
Anti-Fascism - Yes: most major rejections of Fascism are on strongly Christian moral grounds
Monarchism - Depends: Believe it or not Monarchs and other forms of Autocracy are not solely the purview of the Western world. Taking that view, is arrogantly Euro-centric.
Republicanism - Yes
Problem is you just listed mostly all Western political systems or political thoughts. For you to then make the argument that if all of these are Christian derived it is meaningless, is special pleading because you only listed things that would be considered in the class of Christian-derived. If I just listed various breeds of dogs and then claimed the classification of "Canine" is pointless and ubiquitous I would have made your same argument.
YoungAchamian
We walk conditioned ground and name our folly civilization.
1d ago
This is quite the shift of goalposts.
Maybe I'm not using the right words. I see it as the same thing. It's essentially an idea of cultural determinism. If Christian Morality is the root soil, then really any plant that grows out of it is going to be suffused with it. Either rebelling against it intrinsically or co-opting some of it's values. A truly foreign morality system would not even consider the Christian values a topic of discussion. It's like the saying goes, the opposite of love is not hate, it's indifference.
then don't we have to conclude that your own particular sect of atheism also formed from the base theology of Christianity?
Absolutely, I'd say my moral system is culturally Christian. I, as a westerner, was raised to be guilt not shame oriented, I believe in noblesse oblige, the basic dignity of all man, following my conscience against civil authorities, that intentions matter morally, etc.
Though to be clear, I am an agnostic theist, not an atheist.
Certainly not about theological development.
The original poster was arguing that culture is downstream of theology, that morality is downstream of theology. If so, then theological development is by definition, in the discussion.
YoungAchamian
We walk conditioned ground and name our folly civilization.
1d ago
Pre-Transformer AI research had a very distinctive vibe. Chūnibyō is the word that comes to mind. It was essentially a pseudoscience, laundered behind scraps of legitimate science.
Maybe Pre-Neural Net AI research. Again depends on where you draw the line on what is "AI", before I get into another conversation with the other AI researcher here. But Machine Learning is entirely considered under the "AI" umbrella and has for a long time, at least a decade and half pre-transformer. Charitably you are describing the Minsky-esque AI researchers who were big into Symbolic AI which was a thing in the 70s-80s, ie almost 3 decades before the transformer.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5d ago
There are still plenty of non-trinitarian
The better phrase I should have used was "scoop out Christian Metaphysics"
I think in order for you to make the argument that none of the people in that list were ever influenced at their core by any Christian morality, Christian Theology or Christian hamartiology, you'd have to put forth the idea that tabula rasa exists into adulthood. That your mind, your beliefs, your values are entirely your own. You tread un-walked ground and nothing of the culture that you developed, grew, learned in steeped into your core. I guess it's basically a soft cultural determinism view. Your beliefs are downstream of your societies culture, your societies culture is downstream of the theology. Thus your entire core moral system operates from a distinctly Christian lens. Rejection, acceptance, mutation, convergence, how ever you change, doesn't really matter, your base subset is the core theology of your society. In the west that is overridingly Christian.
I'd be curious to see if atheism that developed from a non-western society has similar moral systems as that of a western-developed atheism. However I'm not sure one exists. Christianity is such a universalist religion, while at the same time it suffers apostates, meaning you can stop believing and go hard in that, and still be allowed to live. I'd postulate that non-universalist religions don't inspire such aggressive rejection, and the other universalists like Islam don't suffer apostates.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5d ago
The world was created as good but it has fallen. The human is born broken, he is prone to sin and evil and his duty is to navigate this fact.
You are missing a middle bullet point: Humanity inherits the guilt of their ancestor Adam. A human is born broken BECAUSE of Adam. Also your last bullet is subtly wrong too, it's not that "Humans must navigate this fact" Humans must actively seek redemption and grace. Some googling tells me this is actual a Western Christian vs Orthodox Christian disagreement in the view of OG sin.
Here, Buddhism is a direct comparison. The universe is orderly and good, while human is born broken often ridden with karmic debt.
This is a very christianized interpretation. Buddhism doesn't view the world as good or bad. It teaches that the conventional world (samsara) is characterized by impermanence (anicca) and dissatisfaction or suffering (dukkha), driven by ignorance and craving, rather than being a moral judgment of the world as good or evil. Karma likewise is not inherited guilt of humanity as a whole but the cosmic debt of your past lives. In essence, Buddhism has no primordial human transgression and carries no inherited ancestral guilt.
I can look up those other too later, but considering the loss of that middle bullet around inherited guilt, I'm guessing the distinction comes down to that. You have an incorrect definition of Western Christianities Original Sin.
individuals are born as perfect innocent tabula rasa individuals
Incorrect partially, the race you are born to gives you your sinful status at birth. Being born white means you inherit the racial sins of your ancestors and must seek redemption through atonement in the woke belief system.
The rest of the description I agree with. If you can convince me that the OG sin is more your line than mine, I could be swayed to view Progressive belief systems as less connected to Christian Theology.
And both of them have no place as being the US state religion.
Easy agree, but as far as I know you aren't a religious poster on TheMotte who wants to separate the two so there is no tractable fault, or comparison between the too. Nor have you made claims that everything good about The West is because of Christian Theology (Not that I remember if the OP has, but Christian Maximalists on TheMotte have made this claim repeatedly).
I'd like Christians to stop pretending Wokeism is not their child and not following similar behavioral patterns as it's memetic ancestors.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5d ago
A good chunk of those can be found in pagan religions, though.
Uhhh, I'd need to see serious evidence for that kind of assertion.
The Confession, Repentance, The emphasis of the sanctity of the meek, it's universalism
These are possible to find individually in pagan religions but not really all together in this package. Nolan just got asked this question by a Chinese Classics Professor, because Homeric Greeks don't really have the Christian concept of Atonement that pervaded, and was even a core theme, of his new movie. The Greeks saw the gods as beings, you offended them and you made amends much like you would if you offended your king. It's not repentance. Germanic/Norse Paganism, is almost the opposite in that vengeance for slights/wrongs was the moral basis. I feel like I could dig more and more, hence why I think you need evidence for this assertion, because nothing I can uncover even suggests this is close to the truth.
The concept of Original Sin
This is uniquely Christian, not even Judaism or Islam really hold to the conception of original sin. Adam sinned yes, but the whole of humanity does not bear the burden of his sin in either of those two religions. To define Original Sin, it is something like Primordial Transgression + Inherited Human condition = Inherent Moral Fault
Going down the list:
- Zoroastrianism: No original Sin, has the concept a dualistic conception of evil invading the world, but human's aren't default evil or corrupted because of it
- German Paganism: No OG Sin
- Norse Paganism: No OG Sin
- Celtic Paganism (that we know of): No OG Sin
- Slavic Paganism (that we know of): No OG Sin
- Greek Paganism: Prometheus is the Primordial Transgression but Human's aren't inherently guilty because of it
- Roman Paganism: No OG Sin
- Egyptian Paganism: Humans are judge individually, no collective guilt/sin
- Mesopotamian Paganism: Humans suffer because of the Gods per the Epic of Gilg, no primordial transgression
- Aztec Religion: Some googling shows that Human have a cosmic debt, which they repay via sacrifices (not very OG sin at all)
The concept of Original Sin is about as OG Christian as it can get. I'd go so far as to say that any ideology that borrows the concept, that developed in the west IS borrowing Christian Theology. So yes Communism and Wokeism, tightly match to Christianity. But if the genealogy of Wokeism is Christianity + Marxism, and 1848 Germany was still deeply religious, when Marx wrote Das Capital, then it stands to reason that Marxism and its descendants are also descendants of Western Christianity minus the Christian Metaphysics. After all Marx was culturally Christian. It's essentially Western Christian Theology all the way down.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5d ago
I probably should rephrase that Wokeism isn't the same thing as Christianity. Wokeism is a genealogical descendant of Western Christianity. It's core theological morality is very Christian in nature. In a way that cannot be said of other religions even other Abrahamic religions. The concept of Original Sin, The Confession, Repentance, The emphasis of the sanctity of the meek, it's universalism, etc. Christian slave morality is a direct ancestor of identity-based oppression morality. There is a distinct status inversion to Christianity vs the older European Pagan religions, and the Eastern religions. Core Christian symbolism is literally God suffering on the cross and being holy for that sacrifice. The apparently humiliated and defeated victim can possess supreme moral status, while worldly authorities can be radically mistaken about who is righteous. Now the structural oppressor–oppressed components are substantially Marxian in nature, but breed that belief system with other religions and you would not co-evolve to Wokeism. The Eastern religions are easy to ablate out. Islam, sharing many moral beliefs with Christianity still does not contain the moral primitive of original sin nor does it's central sacred narrative place God himself as the humiliated, tortured, and executed victim.
YoungAchamian
We walk conditioned ground and name our folly civilization.
5d ago
"Everybody is theocrat" is the something that many people only recently realized.
Only if you have an overly broad definition of theology and religion. I know many people on the right like to categorize Wokeism as a theological/religious movement but then they immediately balk when you ask what base theology did Wokeism form from. So either every collective philosophical movement really just tickles the religious part of the human mind and is consequently a "theology", or you have a much stricter definition of "theology" and then no, not everyone is a theocrat.
Wokeism = Secular Christianity. Scoop out the Holy trinity and you get the logical conclusion of championing the meek, the downtrodden, the oppressed.
YoungAchamian
We walk conditioned ground and name our folly civilization.
7d ago
natural-language computer interfaces were created
On one hand this hasn't really been invented by the colloquial usage of the word "computer" on the other hand I suppose you can squint and say giving AI-Agents unlimited access on your machine is a "natural-language interface". However that requires a local setup that is not provided by default to consumers by the big labs. And it's also how you get some of the "alignment" problems where AI-Agents with unbounded security credentials do things you weren't expecting.
Prompt injection was fiction
Sure if you say that having a thinking rock was fiction before the computer was invented. The moment public software interfaces became a thing, so did any sort of injection attack. If "prompt" is the fiction, then maybe we'd split hairs but as public LLMs have been out for what 4 years(?) I'd say we are way past sci-fi territory
Prompt injection has gone from Sci-Fi
Since when has prompt injection ever been Sci-Fi?
PI is pretty much little bobby tables and always has been. It's strictly in the space of controls on user input, requiring filtering before passing it to the model. The two things PI is doing is creatively bypassing hard software filtering or bypassing the learned behavioral controls instilled by RL by getting the model into a context state is outside the region its policy has learned to apply the RL instruction behavior. There's a third occurrence that happens in Multi-agent systems but it's different and also not sci-fi
No Sci-Fi ever required. This is what I mean when I say AI-laymen are cargo cultists. Classic case of not understanding the actual mechanics and needing to rely on imprecise or wrong abstractions.
YoungAchamian
We walk conditioned ground and name our folly civilization.
8d ago
As one of the proponents of the critique that Rationalist AI fans are "writing science fiction", allow me to specify my particular view of it. There is a difference between what Gwern is attempting here and the general Rationalist discourse around things like paper clip maximizers, singularities, basilisks, etc. If I had to put it to words it would be something like:
The scientific and engineering breakthroughs to achieve the technological level where the afore mentioned conceptual ideas becomes embodied and realizable are to word, amorphous, nebulous, underspecified, not defined at any level of scientific or engineering rigor. The end state is the assumed outcome and the pathway to achieve it is handwaved away analogous to a professor putting a hard math problem on the board, stating the solution, and saying the solution is trivial and an exercise to the reader. In this case replace trivial with some dismissive word for "let the shape rotators solve it"
The difference here is that Gwern is taking existing technology and proposing a pathway from it, because of that, the starting breakthroughs needed to achieve it are easier to define because its not an infinite span of starting states that lead to this one eventuality.
He essentially needs to prove these things are possible (at least):
- Ultra sample efficient RLHF, RLVR, or a new finetuning algorithm because the level of personal data is going to be orders of magnitude lower than current amounts
- Quantitative vs Qualitative gap: essentially its pretty hard to take something people know qualitatively and convert it to a quantifiable number. This is what NNs do now with categorical labels, but instead of doing it at the object level, he'd need to figure out how to do it at the meta-learning level
- Training at reduced scale: Goes without saying that LLMs take a lot of compute, in order for everyone to have their own LLM, you'd need to either reduce compute costs, provide a centralized trainer/decentralized inference system (trust problems), or reduce the compute needed.
- Related to the QvQ, he needs somewhere to figure out how to create data for training that is highly personalized, whether this is trying symbolic-styled AI with values that scaffold to behavior, learning some value latent space. This one is a bit Sci-Fi because its poorly defined, and the actual methodology to do it would be be new. The realistic answer is probably just take a survey and RLHF 2.0 on the answers or something.
These breakthroughs do break down into further breakthroughs but they are all on the cutting edge of current research not the distant futures research like a lot of Rationalist AI ideas.
It's the only real effort I've seen from anyone to align AIs properly
Alignment is an impossible boogeyman, it's not possible to "align" a sentient being, so if you view eventual AIs as sentient you should just accept they will do what they want. If you view AIs as a tool without sentience, then "alignment" is better specified from control theory, and your actual research would be in quantifying the systemic signal an AI provides and applying classic control algorithms to it. The systemic signal is the hard part bordering on Sci-fi because its underspecified, amorphous, and not rigorously defined.
YoungAchamian
We walk conditioned ground and name our folly civilization.
8d ago
That might be your argument now, but I think HaG2 has you right on the money: You got hung up on the terminology but you basically can't admit it. You also seem to be going for maximum belligerence for no reason. Now I standby my earlier comment with the addition that being this places whipping boy has rotted your brain. You've somehow lost the plot of this forum.
you're not some smartiepants
No I'm actually a retarded midwit, as are you, as is probably the majority on this board. We all like to huff delusions of grandeur. So let me have my opinion and stay in your glass house nursing the apparent misanthropic grudge you've developed after catching a recent ban. I've sure this current behavioral track is super healthy and very conducive to not getting a ban again.
YoungAchamian
We walk conditioned ground and name our folly civilization.
8d ago
you would not need to be humbled
Right, so this is actually the real reason. You want to humble me. You took 1 word, mentioned 1 time in my 2,485 word train-of-thought pseudo-diary entry, submitted without proof-reading, spell-check, or LLM-proofing, after midnight because I couldn't sleep as I was brooding about the preceding several days, and decided to raise a massive stink about it. You might want to do some soul searching to determine if you are really the arrogant one, and this is all just a massive amount of projection.
yourself as an enlightened individual
Yes because being a pedantic wordcell is the height of enlightenment. Of course! How did I not know in whose steps I needed to tread. I'm sure being a great lawyer or english teacher results in the best governance!
YoungAchamian
We walk conditioned ground and name our folly civilization.
9d ago
Yeah you'll love this, her response to part under questioning was "I love him and I didn't want him to suffer!!!". As if being forced to stay on the boat and be served 3 meals a day was suffering.
YoungAchamian
We walk conditioned ground and name our folly civilization.
9d ago
That's the problem
Yeah, I understand that better than anything. People never want to restrict suffrage in a way that would hurt them or their allies, and always want to target their enemies. I said in another comment in this chain, that the problem is in exclusionary systems, a motivated set of agents will try to exploit the rules in a way that is always more detrimental to the system than exploits that are available to blanket systems. I'm unsure if it's a solvable problem, creating a strictly fair system of suffrage the isolates voting towards agents with a sense of duty to their fellow man and an intelligent long term view towards the prospering of the country and its people. Universal Suffrage is a lot like a market in this regard. It creates price signals for peoples political desires. Shockingly most of those signals are very self-serving, short term benefits, but not all, and its hard to parse the good from the bad.
YoungAchamian
We walk conditioned ground and name our folly civilization.
9d ago
The difference is between fraud charges and fraud claims is pedantic and not how I think. I reject your dichotomy as I reject most dichotomies. This could have been an informative post with you correcting me and explaining the difference. Instead you decided to be an ass as your the default. I suppose that is good evidence against future behavior that you do. MKC's default behavior is to find any disagreement and try to be right no matter what, without helping anyone learn anything.
YoungAchamian
We walk conditioned ground and name our folly civilization.
9d ago
No, you don't get to acknowledge you used the wrong terminology in another comment
I acknowledged that I could be using less precise terminology, because I'm not an asshole. You seem to want to be seen as one instead.
I'm not mistaken, I'm just not pedantic af about syntax. Which apparently you are. Good to know.
Possible but how many of these other grads went on to be executives at major companies in their 30s? Doing an ablation that to show the value of a Dartmouth degree separate from the individual isn't really possible, but comparatively to other parts of its student population becoming a major CTO at 33 is extremely rare.
I've never wanted to be a PM, I've always wanted to be an ML researcher, which is what I am now... I'm cynically examining how she got to where she is, and comparing that against myself and people I've worked with, including Ivy grads and FAANG engineers and wondering how many of us have the same set of opportunities for that level of advancement.
My general question is: how did a no-name nobody, without any exceptionally displayed technical skills get onto an executive-acceleration track so early that, by age ~29, she was entering OpenAI at the VP level? I'm sure we'll never really know, but if you don't look at outliers without clear evidence of why and are skeptical. Then we operate on different levels of pattern recognition and sense making.
More options
Context Copy link