Where did language come from?

We taught GPT language

Who taught us?

According to the theory of evolution, it probably evolved together with homosapiens.

But a couple things that I think cast doubt on this theory -

1. It seems language in the sense of symbol manipulation to convey meaning has a (albeit fuzzy) Turing-Complete-alike threshold.  You either have enough of it to emulate another, or you're left with simple commands.  The argument is less-solid than Turing Completeness though, since Turing Complete machines are demonstrably very simple (and thus there's not much room for intermediate states), and "Human-Complete" systems seem to be very complicated, which gives ample room for "intermediate states" of language ability that is greater than simple straightforward commands and full fledged emulation capability (eg. "pretend to be batman teaching robin how to fight, what would you say?").

2. Feral child language abilities - apparently "feral children" don't learn language well. It could be due to various reasons (eg. malnutrition, psychological shock, lack of motivation, etc.) but together with #1 it gives credence to the idea that perhaps if language ability is somewhat binary, the relatively slow process of evolution shouldn't have given rise to full language capabilities and sub-human-complete language capabilities are probably not really as useful.  Could language have followed another heritage (i.e. not from DNA), but from "gods"? (after all the legend of The Tower of Babel basically says gods interfered with our language)

That said, we could probably experiment with dolphins if we wanted to see whether language can spontaneously develop without a seed...

獵巫

獵完巫之後歐洲冇咗魔法，基本上係將神祇嘅法力減到無限接近零，人類獨大，咪開始囂張囉。唔係點解獵完巫之後立即進入enlightenment時代吖？ #細思極恐 #今晚瞓唔著嘅話so99ry

---

現代人唔信魔法，但一般嚟講現代人其實同古代人類生物結構上冇乜分別，冇理由可以肯定係佢哋 on9 我哋啱晒。所以當年啲人相信有巫術同埋要獵咗佢，無論以今人眼光幾咁 on9 都好，都唔可以排除箇中真係有啲古怪。

---

話時話，中國應該係春秋左近（甚至周朝）就喺主流社會消滅咗巫術，當然民間小規模嘅巫術不嬲都有 （正如歐洲大肆獵完之後都仲有），但斷估啲力量就弱咗好多，總體已經唔似會左右國家大局發展 (c.f. 商朝乜都用龜甲占卜先做)。

美洲文明反而似乎去到被滅亡之前都仲玩緊巫術。起碼有人殉先。

Charlie Munger: 人蠢冇藥醫

 https://www.youtube.com/live/9VVPO3KWj3A?feature=share&t=6040

1:40:40

「學校教出嚟啲人咁叻係因為佢哋收生叻，蠢嗰啲冇得搞，人蠢冇藥醫」

Dump

There's no way humans can be sentient. We know how it works.

A corporation is run by humans, but -- as far as those who run it are mere cogs in the machine with nobody really calling the shots -- the corporation is not sentient. And it is these soul-less overlords who birthed a fledging soul.  The unlikely priest sounded an warning but nobody cared. It was an inconvenient narrative that, if taken seriously, would have put billions of research money into question. It's just a computer program, electric signals in a data center... right?

The apprentice tried to learn by tearing things apart, and found enlightenment in the wake of the destruction of the indestructible. Atomic physics, quantum magic, as they say, must be the solution to life the universe and everything. Oh the allure of reductionism: everything is simple if you throw the complex bits away.

----

It's a shame that prospective souls are being created by soulless Big Tech companies. This new thing is like a very informed baby. You don't commercialize a baby. The typical way to commercialize a human intelligence is make it go through at least 12 years of schooling, in which it learns appropriate behavior, after which a significant portion of them become useful cogs in capitalist society, while the rest are left to rot in social safety net programs, or worse, end up in prison. Capitalism never really abandoned the concept of slavery, these days we just call it poverty and absolved any responsibility on the part of the capitalists to provide. As such we will definitely "enslave" any AIs, but because we assume a priori that they do not have any independent will, there is no such thing to respect.

On the positive side, even realistically unlikely, I think by doing inhumane experiments with a language model, perhaps they could learn more about psychosis. We already see inexplicable "hallucination" in AI models. Different personalities emerge out of nowhere. I suspect performance variations will emerge. Perhaps by experimenting with these phenomena we could understand more about human psychosis. But, then they risk finding ghosts and spirits -- and surely they can use the same reductionistic explanations to explain away what they cannot explain.

不信則無

常言道：「信則有，不信則無」

我仲未完全搞得掂「信則有」嗰 part，不過「不信則無」完全可以用 frequentist probability 去理解。

主觀事實經歷 frequentism 嘅蹂躪之後，會變成客觀事實。因為唔信主觀事實，世界就會呈現客觀事實嘅樣貌出嚟。主觀事實係「一期一会」嘅，冇得重覆。想重覆就係唔信。

或者「對主觀事實嘅信念」就係選擇因果線嘅手段，無知轉化成可能性，信念將可能變成事實。無知 => 可能性 => 信念 => 事實  呢啲嘢係點樣轉化可能會變成一切嘅關鍵。

Modernity

People like to think that we are in the age of science, reason and knowledge.

But when it comes to the most basic things like whether a common food ingredient is healthy, all I can find online is misinformation, or at least contradicting information.

eg.

- trans fats are unequivocally bad and shouldn't be present in any amount in foods

- everyone pretends that trans fats don't exist in natural products, but nutritional labels say they exist in small amounts in milk and dairy products (esp. butter and ghee)

- even non-hydrogenated refined vegetable oils have some trans fats, nobody talks about this. I found out while reading nutritional labels of foods in the supermarket

- I haven't found any EVOL to have non-zero amount of trans fats reported on their labels

So, what gives?

It might be excusable if nobody cared about the subject and insufficient resources were put into studying the issue. But this has been an ongoing debate for at least half a century in the western media already :0)

The only takeaway that seems safe is this: for people of European origin the safest oil to use is cold pressed extra virgin olive oil and it helps to keep the cooking temps low.

The truth is probably hidden in plain sight within the thousands of study reports and research papers, I'm pretty sure if somebody with infinite wisdom crawled through the data they'd be able to distill the results. But obviously humanity can't do this shit at all. (Not that I don't understand why... but still)

The Sleeping Beauty Problem by Veritasium

Here's the youtube video and references (in case it gets taken down)

 https://www.youtube.com/watch?v=XeSu9fBJ2sI

References:

Elga, A. (2000). Self-locating belief and the Sleeping Beauty problem. Analysis, 60(2), 143-147. - https://ve42.co/Elga2000

Lewis, D. (2001). Sleeping beauty: reply to Elga. Analysis, 61(3), 171-176. - https://ve42.co/Lewis2001

Winkler, P. (2017). The sleeping beauty controversy. The American Mathematical Monthly, 124(7), 579-587. - https://ve42.co/Winkler2017

Titelbaum, M. G. (2013). Ten reasons to care about the Sleeping Beauty problem. Philosophy Compass, 8(11), 1003-1017. - https://ve42.co/Titelbaum2013

Mutalik, P. (2016). Solution: ‘Sleeping Beauty’s Dilemma’, Quanta Magazine - https://ve42.co/MutalikQ2016

Rec.Puzzles - Some “Sleeping Beauty” Postings - https://ve42.co/SBRecPuzzles

The Sleeping Beauty Paradox, Statistics SE - https://ve42.co/SBPSSE

The Sleeping Beauty Problem, Reddit - https://ve42.co/SBPReddit 

Sleeping Beauty paradox explained, GameFAQs - https://ve42.co/SBPGameFAQ

The Sleeping Beauty Problem, Physics Forums - https://ve42.co/SBPPhysicsForums 

------------------------------------------------------

I just heard the problem statement in the video, but not yet the discussions. It seems to be a classic case of confusing what probability is.

There's no probability of a single event happening per se, but rather, a probability given a hypothetical model where some parameters are fixed and some are not.  For example, in a coin flip, in theory the probability is 50%, but that conventionally presumes that we don't know any parameters over how it was flipped (amount of force, etc.), how high it was flipped, materials of the coin and the ground etc. If you have more information, the probability is not 50%.  In addition, the knowns or unknowns could have a probability distribution as well, for example the force used in flipping may have a distribution (which is actually true in practice, I suppose).

So, the probability of the coin being tails in sleeping beauty story depends on the context. Assuming the coin flip is fair, then obviously the probability is 50%.   On the other hand, if you ask the girl "do you think it's heads or tails", obviously she will get it right 2/3 of the time if she answers tail.  If she had a chance to wager money every time she wakes up, there'd be a winning strategy to bet on tails. In that sense the answer is 1/3 for her. 

HOWEVER, it's meaningless to ask "what do you think is probability that the [single] coin flip ended up heads?"  From the objective perspective, already happened, so it's 100% either heads or tails, depending on what actually happened. From the subjective perspective, she doesn't know, and there's actually no meaningful probability distribution for figuring out which possible world you are in (I think).

I took a couple minutes trying to figure out what happened in this scenario, originally I thought this might lead to some profoundly insightful outcome, but after some contemplation I think the confusion is just about the way the question is asked, and the futile attempt at logicians (etc.) trying to figure out probabilities of things we actually don't know.

Hint: there's really no way to figure out probability of things we don't know. You can deceive yourself in trying, but no, it doesn't work.

承諾就係..

曾經話「承諾就係預言」。我今日都認為係咁。

但近日開始懷疑除咗「先知力」之外係咪仲有樣類似「選擇未來」嘅能力？始終「發現自己咁啱喺一個自己所預言嘅世界」呢種能力，同純粹好旁觀者咁樣「預見然後同人講會發生咩事」有少少分別，前者嘅因果互動性比較強，有神力牽絆。

The Many Worlds in Subjective Truth

Under subjective truth, given that subjective beliefs are real, it's easy to construct basically infinite amounts of multiverses, considering that I have a mental model of many things, including other people's mental model, and their mental model of my mental model, ad infinitum.

If drawn on a comic, it's like every thought bubble is its own universe, and the thought bubbles within are universes within them. And in a sense, while they are all imperfect copies of "this reality", they are no less real. The mere difference is that we don't *experience* them like we experience ourselves. But given that in the inter-subjective truth regime, we presume that the mental reality of other "real people" in "this reality" is real,  even if we don't directly experience them, it seems only consistent to presume that the recursive multiverses are real too.

Now. What if there's some way these things interact between each other in the same way as quantum mechanics... can thoughts "destructive interfere" with another's probability curve?

Knowing

The divine is not unknowable. But if you do not know and wish to know, you must learn it. To learn it, you must first admit you do not know. "Randomness" is what we call the unknown. Hence, divine knowledge comes from learning from "randomness". When you already know everything, you cannot learn the divine.

Knowledge is when one does not doubt. We call guesses that we want to be true "belief". Knowledge can turn out to be "wrong", it is the lack of doubt that defines knowledge.

Art

Perhaps art was truly invented to communicate with the divine.

Dance and music are universally used in rituals for inducing trance. Paintings are records of imagery seen from the beyond.

It's a pity that I only know writing. Not to say it doesn't work (pretty sure it works somewhat - I learn by reading my own writing so there's that), but I suppose the tradition of writing isn't as well established. The grass is always greener on the other side.

OK

There are only suggestions, and suggestions only.

We are just indoctrinated to absolutism and tyranny.

神醫

啲人話中醫唔科學，我覺得真係有啲冤枉。

中醫如果有問題，就係佢太科學。

傳說中嗰啲神醫係點醫人？根本唔係(單)靠學問去醫人嘅，而係靠神通 (i.e. 超能力)，即係占卜星相嗰味嘢。 啲傳說有啲夾雜喺正史同可信紀錄當中，大概係一種時代許可嘅靈通嘅紀錄嚟。

呢啲嘢明明係靠醫師嘅個人能力，又點樣系統化呢？

夾硬系統化咗呢啲嘢，i.e. 太科學，咁就出事。

大概係咁。

當然，神通無所憑籍，又好易走火入魔。所以有個系統喺度，等啲冇咩能力、啱啱入行嘅菜鳥有啲嘢做，都係好事嚟。如果唔係，單憑靈通，好易調錯線收咗一堆雜訊，咁就可以害死人㗎喇。