There is one outstandingly wrong, critique of most LLMs :
"they don't have an intuition".
Well, intuition is NEARLY ALL they have - just open it up and look at the data structure. This is because they are trained as probabilistic consolidations of all past uses of each word in known history, vis-a-vis "unsupervised learning". What they do NOT have is rules, because they are not trained with hard counterexamples, vis-a-vis "supervised learning", let alone trained to have strict rules, vis-a-vis "logic programming".
Thanks for the thoughtful comment, haha. Without getting too much into the weeds, I think : Each of our perspectives on what AGI / or "I" in general means is shaped by unique personal experience, a bound set for each of us. These sets may or may not be commensurable - we don't currently know if the categories or your experience, and the categories of mine are even comparable. This has political consequences, mainly related to how seriously we take each other's speech.
Some of the technological steps you mentioned however, I must say are "operational improvements" which improve cache hierarchy and introduce compression in some cases -without really changing the overall "architectural" approach employed by the currently trendy transformer gang.
I think it remains useful to query both, other humans, and the results of what we can make machines do ... in improving not just "what we think intelligence is", but also how we can communicate what we think it is.
A parent asked, if others had asked their children, what their children wanted from education - and how to judge if the child's preferences or the parents' preferences were better.
I elected to share a canonical rant : before I went to Year 1 of elementary school, my parents asked if I wanted to study in a school with two, or three languages. I said two, because I wanted to be lazy. They sent me to the school with three. So for the rest of my life, I have been wary of fools asking meaningless questions.
Aside from that, of course life has provided many disparate experiences. Many have involved speaking with people who tell me, nicely or otherwise, that they believe I am not well-informed or well-opinionated. The problem with me, I find, is that while I am quite willing to accept these views, and submit to compromise, more often than not, I find that I should have paid less attention to them, because the speaker was foolish. I can only speak from my own experience, which is quite limited.
I do not believe this makes me intelligent, but sometimes intelligence is something you have whether you want to have it or not. And others lack it, whether you wish they had it or not.
And sometimes you are just stupid. :)
Day 1 back at elementary number theory ( day 1155 of sabbatical2, day ~15,547 on earth ). I am revising this slide on the notation for exponentiation and logarithms which I made year ago.
TIL :(or am reminded, because I probably forgot) :
At rest :
Firing :
Muscle Diseases, and ordinary states which become diseases if not reversed within ordinary timeframes :
I'm guilty of avoiding AI over the past decades because the techniques I've heard about still seem rather primitive. The last decade in particular has been characterised by VCs throwing obscene amounts of money at a fundamentally inefficient approach.
After noodling around in the current tech for a week or so, just to make sure I understand what's going on, I think it's safe to say that the the titular concern of this post stands firm.
An analogy. At the heart of trendy LLMs is a giant Y times Y list of known words in English ( also other languages, but nevermind those for now ), forming an enormous 2D table. It's Y times Y because both H and V axes have the same list of words - sure, half the table is redundant, so you can think of it as a trangular half-rectangle of unique pairs. If you lookup the junction between any two words, you find a WINDOW to a realm of many, about 10^(3 +/- 1 ), dimensions of information about each pair of words.
Based on the "best" available public information, this is the data structure for storing everything the LLM knows about the world - which it knows ONLY FROM READING TEXT ( except for MLLMs, which we'll also ignore for now ). Even in the case of MLLMs, sensory spatio-temporal data can be understood to be stored within the realm accessed by each WINDOW.
Of course, when you ask an LLM a question like "What is the meaning of life?" it doesn't need to peek through ALL the WINDOWS and all the realms above, rather it only has to peek through a subset of windows, W_n. However it's still fundamentally inefficient, because : it doesn't JUST look at W_n and answer your question.
Oh no.
It looks at your question, goes to a set of windows, W_n1, takes a peek, grimaces in deep thought, spits out ONE word, then pats itself on the head and goes to W_n2, a COMPLETELY DIFFERENT set of windows, spits out a SECOND word, and goes on and on until W_nM to obtain M words. While there is a cache, this is exactly what it sounds like - quite a bit of work.
I'm not a very smart fellow, so I'm probably going to be wrong about this. But, it would seem that the thin red circle that I am drawing around the entire architecture that drives most of our current AI tech is going to have to collapse, and be replaced by more efficient methods soon enough.
When? Who knows. I've been waiting since 2003, and "they" haven't figured it out yet. Some of "them" are pretty close, I think. I like the JEPA and MLLM approaches, and am eager to see where they end up.
One thing's for sure - if you create a model of the world based on this Cartesian product approach to windows upon a realm of thousands of dimensions, and have to crawl through the whole library, to peek through a different SET of windows, once per answer word ... even if it gets you the correct answer to a question, it's bloody tiring.
And to BUILD that library of windows to that realm ... it still takes WAY more money than it takes to build a human brain.
Misanthropy ( maybe a good startup name ) in cognitive sciences.
1. Many humans celebrate the act of living. This coalesces into a treatment of the brain as a joyful machine - lovingly and wonderfully operational. So, there is this current of trying to make machines more like humans, because there is a prejudice that humans are the best template.
2. On the other hand, I think life is boring. And I'll be damned if my brain hasn't been the dumbest piece of shit machinery I've ever had the chance to work with. So my prejudices in cognitive sciences (and politics in general) tends to be about fixing all the things that human brains are bad at.
I think each camp finds the other superficially appalling :P
Walaupun tak sedap dengar, dan disebut kasar, rasanya :
TIL current frontier "transformer" LLMs ...
1.
The terms "encoder, decoder" apparently, don't apply to the encoding of end-user queries. "Encoder-only" and "encoder-decoder" LLMs thus refer to entirely other things.
Frontier LLMs are "decoder-only" text generators, but they DO ENCODE the user's query into "token" space.
Simplifying the dumb parts : the following model process applies to any data, sensory or verbal - all those inputs would be first translated into "tokens", and input the same way. After processing, all results are spat out into the target (sensory or verbal) human language.
2.
Frontier "decoder-only" models function "autoregressively", they "produce results based on historical data". The important part of the model's thought process is broadly - grab a PENCIL and SKETCH it out - 2A looks something like this :
Input -->[ 2A1 --> 2A2 ] times, ( 80 to 120 minus skipped ) layers --> Output
2A.
Input data activates a process of 80 to 120 sequential "layers". ( Fortunately at this point in history, BRANCHING LOGIC allows some to be skipped. ) Each layer is internally composed of,
2A1.
... an initial "attention" network, of PARALLELISED weight-holding memory cells ( stacked to varying depth ), terminating in 64 to 128 "attention heads" ( architectural variations : MHA, GQA, MQA, MLA, etc. ), whose purpose is to CLUSTER the input data, followed by
2A2.
... a subsequent "feed forward network" block, of BRANCHING weight-holding memory cells ( stacked to varying depth ), whose purpose is to LOGICALLY JUDGE the input data.
3.
During the model's "training" period, each loop through all the layers of 2A "feeds forward" data through all ( 80 to 120 ) layers in sequence, and from 2A1 to 2A2 within every layer, then checks the correctness of the result, then "backpropagates" corrections to the weights of 2A1 and 2A2 in all ( 80 to 120 minus ) layers - which make future "feed forwards" more correct. Many cycles happen, to maximise correctness.
4.
During the model's "inference" period, a "feed forward" loop through all ( 80 to 120 minus skipped ) layers of 2A happens ONCE FOR EVERY SINGLE WORD GENERATED. Like this ...
[ Query ] -> Loop1 -> adds word1
[ Query + word1 ] -> Loop2 -> adds word2
[ Query + word1 + word2 ] -> LoopN -> adds wordN
Can you see how incredibly stupid this is?
It would (should?) be much simpler to have a model that actually understands the query in terms of a sensory-spatiotemporal model, generates an answer in the same space ( which it then reads just once), and then outputs just once into the target language. Maybe this is something JEPA, and future models will fix.
5.
Because 2A1 is just about clustering the data and not judging it, all 64 to 128 attention heads and their network cells are hit, during any "feed forward" loop, unless a layer is skipped.
6.
But because 2A2 allows branching, over 90% of cells can be skipped during each "inference" loop, though none can be skipped during "training" loops.
👷🏼戴上阴谋论帽子：当然，标准的阴谋论是这样的：
1. 马来西亚所有种族极化的政党实际上都与对立阵营的政党勾结。
2. 这些种族极化的政党并非天真地发表容易被攻击的言论，而是相互协调，轮流扮演受害者。
3. 这使得媒体中种族极化的刻板印象得以延续，支撑了种族极化政治纲领的短期支持，却牺牲了对经济增长和种族间文化融合的长期关注。
🙋🏼摘下阴谋论帽子：当这种情况在没有明确协调的情况下发生时，正如马祖基博士在最近的播客中所说，在学术界，我们称之为结构性问题。
👷🏼Dài shàng yīnmóu lùn màozǐ: Dāngrán, biāozhǔn dì yīnmóu lùn shì zhèyàng de: 1. Mǎláixīyà suǒyǒu zhǒngzú jí huà de zhèngdǎng shíjì shang dōu yǔ duìlì zhènyíng de zhèngdǎng gōujié. 2. Zhèxiē zhǒngzú jí huà de zhèngdǎng bìngfēi tiānzhēn dì fābiǎo róngyì bèi gōngjí de yánlùn, ér shì xiānghù xiétiáo, lúnliú bànyǎn shòuhài zhě. 3. Zhè shǐdé méitǐ zhōng zhǒngzú jí huà de kèbǎn yìnxiàng déyǐ yánxù, zhīchēngle zhǒngzú jí huà zhèngzhì gānglǐng de duǎnqí zhīchí, què xīshēngle duì jīngjì zēngzhǎng hé zhǒngzú jiān wénhuà rónghé de cháng qī guānzhù. 🙋🏼Zhāi xià yīnmóu lùn màozǐ: Dāng zhè zhǒng qíngkuàng zài méiyǒu míngquè xiétiáo de qíngkuàng xià fāshēng shí, zhèngrú mǎzǔ jī bóshì zài zuìjìn de bòkè zhōng suǒ shuō, zài xuéshù jiè, wǒmen chēng zhī wèi jiégòu xìng wèntí.
-
👷🏼pakailah topi komplot jap : teori konspirasi rujukan tentulah
1. Setiap parti politik Malaysia yang menunggang kaum, sebenarnya bersekongkol dengan parti-parti yang tunggangannya bertentangan.
2. Parti-parti yang menunggang kaum tidak secara naif membuat kenyataan yang mudah diserang, sebaliknya mereka sepakat dan bergilir-gilir menjadi mangsa.
3. Ini mengekalkan persekitaran media kiasan terpolarisasi kaum : maka, menegakkan sokongan jangka pendek bagi perjuangan politik berpolarisasi kaum, dengan mengorbankan tumpuan bersepadu jangka panjang terhadap pembangunan ekonomi dan pergabungan budaya antara kaum.
🙋🏼Tanggal jap topi komplot: apabila ini berlaku TANPA PENYELARASAN TERSURAT, seperti yang dikatakan oleh Dr. Marzuki dalam podcast baru-baru ini: dalam akademik kita menggelarnya sebagai keadaan berkerangka.
-
👷🏼Tinfoil hats ON : the standard conspiracy theory of course, is
1. Each Malaysian political party that is racially polarised, is actually in cahoots with parties of opposite polarity.
2. The racially polarised parties are not naively making statements which are easy to attack, rather they coordinate and take turns to play the victim.
3. This perpetuates the media landscape of racially polarised tropes, propping up the short-term support for racially polarised political platforms, at the expense of long-term integrated focus on economic growth and cultural integration between races.
🙋🏼 Tinfoil hats OFF : when this happens WITHOUT EXPLICIT COORDINATION, as Dr. Marzuki said in a recent podcast : in academia we call it a structural situation.
From chat with an influence:
3. About the world's PAST AND PRESENT interaction with Islam, I think we're generally stuck at the nopope problem. Which is, ( not sure if you or someone else reiterated it lately ) that there is no central governance for Islam, so any Tom, Dick, and Chin Lee can wake up tomorrow and start doing things in the name of Islam, without any formal body for everyone to complain to or measure them against. This is pretty much the bottleneck for progress on global attitudes towards Islam, both good and bad, Islamophilic-phobic whatnot.
... or "how any brains, human or artificial, can do maths" : including the creation of novel concepts for nouns, verbs, and adjectives in maths.
I started writing this as a comment on Valerio's post, but it overflowed so I made it a repost, then it overflowed, so I had to articleise it.
I think categorically, "mathematics" isn't "a thing that can be solved" as a whole. Mathematical QUESTIONS yes, but there is no end to the questions.
Mathematics ITSELF is a [ software program ], for [ representing ways to reason about phenomenology in general ] - in other words, maths is a programming language, or compiler toolchain - with specific focus on [ phenomenology that is objective ].
( This applies to mathematics as a social activity - in private, any mind can perform subjective quantification upon phenomena it perceives, even though that activity is not directly observable by other minds. Because at this point in history, we still do not have [ social / common access ] to the [ subjective phenomenological experience of humans ], since that memory space is embedded within an ill-understood meat net. )
I think the matter of coming up with new [ mathematics, and science in general ] is really just a generative process of creating ontologies of language where existing ontologies are insufficient. It's just mutate, and cull, loop. What gives rise to the DEMAND for new ontology / concepts, depends a bit more on the field.
In physics, ontological DEMAND is a matter of "we have some observational SENSORY data that does a quantifiable pattern, but we don't have a name for the pattern, though we can point at the pattern. Also we don't know how this new pattern relates to old patterns, and we need to do more tests to find out."
In mathematics, ontological DEMAND is a matter of "basically everything that physicists do, but swap the object of research from SENSORY phenomena in general to [ formal INSENSIBLE verbal concepts, attached to some SENSIBLE notation that we use to represent it ]".
While physicists have to tolerate the limit of "hypotheses are falsifiable, and resistant to falsification under duress," mathematicians tend to need to show that "theses are necessarily true/false".
In the realm of testing the combinatorial play-niceness of known fundamental relations and entities, that's fairly straightforward.
The discovery of new fundamental relations and entities ... is also driven, from the same sensory source that drives such discovery in physics.
Physics and mathematics are both "sensory", it's just that in physics the sensory data is of [ intrinsic ] interest, whereas in maths the sensory data is [ just symbolic of insensible interests ].
So, in the same way machines can be programmed to be creative in hypotheses about [ intrinsically interesting sensory data ], they can do the same for [ sensory data, i.e. language morphemes and syntax, which are extrinsically interesting ].
After all, in and of itself, [ the latent intuition about how logic, space, and time, work ] is essentially the brain's physical activity, as a [ subconscious memory space ]. Data from there then intrudes, from time to time, upon our also physical [ imaginations, as conscious memory spaces ].
Both memory spaces simply process [ sensory data structures, which we may call qualia when they move into the conscious memory space ].
A practiced thinker knows how to completely map, FROM qualia, TO analogous signals or other data structures which are known to be objectively quantifiable, through the social practice of science.
Of course, I might be venturing a bit too far here, in terms of asserting that meat brains are just sub/conscious physical information processes.
Haha
- - -
Discussion
2026-05-26 :
You can deterministically block things from happening. But in order to do that, your underlying architecture must have deterministic outcomes for all corner cases.
In the event when you, by nature of not having enough computational resources to check all corner cases of a deterministic system, must treat a deterministic system as non-deterministic ... it's just a black box! Like a person!
Similar approaches to governance then apply.
https://www.linkedin.com/posts/chuck-h-securityexecutive_cracksinyourfoundationmodels-aisecurity-math-share-7463025773600333824-VgO0?
for If you live in Malaysia, you must know this! The law of Malaysia EXPLICITLY points at adherence to EXTRA-JUDICIAL NORMS.
In anthropology, this admission of informality creates what is called "a high-context culture".
Some people feel that this is a rich part of our culture, and prefer to have things perpetually maintained in a limbo between formality and informality.
Others feel that this limbo is counterproductive.
In any event, it is important to understand that for the time being, we live in this situation. The consideration of how valuable this is, the public discourse about it, and the perpetual progress of laws through time, is inevitably part of what makes Malaysia, Malaysia.
---
Google for translation, because I'm lazy today, and edits marked *as such* :
Jika anda tinggal di Malaysia, anda mesti tahu perkara ini! Undang-undang Malaysia SECARA JELAS menunjukkan pematuhan kepada NORMA *LUAR-KEHAKIMAN*.
Dalam antropologi, pengakuan tentang *ketidakrasmian* ini mewujudkan apa yang dipanggil "budaya konteks tinggi".
Sesetengah orang merasakan bahawa ini adalah bahagian yang kaya dalam budaya kita, dan lebih suka perkara-perkara sentiasa dikekalkan dalam keadaan tersimpang antara *rasmi* dan tidak *rasmi*.
Sesetengah orang pula merasakan bahawa ketidakformalan ini adalah tidak produktif.
Walau apa pun, adalah penting untuk memahami bahawa buat masa ini, kita hidup dalam situasi ini. Pertimbangan tentang betapa berharganya ini, wacana awam mengenainya, dan kemajuan undang-undang yang berterusan dari semasa ke semasa, pastinya merupakan sebahagian daripada apa yang menjadikan Malaysia, Malaysia.
---
NotebookLM for laziness again :
Here is the reformatted translation of the source material. Each row contains a maximum of twelve ideograms, with alternating rows of Pinyin:
**如果你住在马来西亚，**
(Rúguǒ nǐ zhù zài Mǎláixīyà,)
**你必须知道这一点！**
(nǐ bìxū zhīdào zhè yīdiǎn!)
**马来西亚法律明确指出**
(Mǎláixīyà fǎlǜ míngquè zhǐchū)
**要遵守法外规范。**
(yào zūnshǒu fǎwài guīfàn.)
**1. （任何人）的感受**
(1. (Rènhé rén) de gǎnshòu)
**在法律中没有定义。**
(zài fǎlǜ zhōng méiyǒu dìngyì.)
**但已经通过了法律**
(Dàn yǐjīng tōngguòle fǎlǜ)
**来保护（任何人）**
(lái bǎohù (rènhé rén))
**免受情感伤害，**
(miǎn shòu qínggǎn shānghài,)
**通过惩罚（伤害**
(tōngguò chéngfá (shānghài)
**任何人情感的人）。**
(rènhé rén qínggǎn de rén).)
**2. “马来人”的习俗**
(2. “Mǎláirén” de xísú)
**在法律中没有定义。**
(zài fǎlǜ zhōng méiyǒu dìngyì.)
**但联邦宪法任命**
(Dàn liánbāng xiànfǎ rènmìng)
**马来统治者为**
(Mǎlái tǒngzhìzhě wéi)
**这些习俗的保护者。**
(zhèxiē xísú de bǎohùzhě.)
**2a. 作为推论，**
(2a. Zuòwéi tuīlùn,)
**马来统治者的定义**
(Mǎlái tǒngzhìzhě de dìngyì)
**被下放到九个州的法律中，**
(bèi xiàfàng dào jiǔ gè zhōu de fǎlǜ zhōng,)
**而这些州中的每一个**
(ér zhèxiē zhōu zhōng de měi yīgè)
**本身都有一部法律宪法，**
(běnshēn dōu yǒuyī bù fǎlǜ xiànfǎ,)
**指向（在某些情况下**
(zhǐxiàng (zài mǒu xiē qíngkuàng xià)
**是法外的）习俗，**
(shì fǎwài de) xísú,)
**这导致了谁是和**
(zhè dǎozhìle shéi shì hé)
**谁不是马来统治者的执行。**
(shéi bùshì Mǎlái tǒngzhìzhě de zhíxíng.)
**3. “土著”的概念**
(3. “Tǔzhù” de gàiniàn)
**(bumiputera)**
((bumiputera))
**在联邦宪法中**
(zài liánbāng xiànfǎ zhōng)
**没有详细说明，**
(méiyǒu xiángxì shuōmíng,)
**因此我们有附属立法**
(yīncǐ wǒmen yǒu fùshǔ lìfǎ)
**和行政机构与**
(hé xíngzhèng jīgòu yǔ)
**非政府组织合作**
(fēi zhèngfǔ zǔzhī hézuò)
**来解决这个问题。**
(lái jiějué zhè gè wèntí.)
**宪法对谁拥有特权**
(Xiànfǎ duì shéi yǒngyǒu tèquán)
**有一个极其简短的定义。**
(yǒu yīgè jíqí jiǎnduǎn de dìngyì.)
**在人类学中，**
(Zài rénlèixué zhōng,)
**这种对非正式性的承认**
(zhè zhǒng duì fēi zhèngshì xìng de chéngrèn)
**创造了所谓的**
(chuàngzàole suǒwèide)
**“高语境文化”。**
(“gāo yǔjìng wénhuà”.)
**有些人认为这是**
(Yǒuxiē rén rènwéi zhè shì)
**我们文化的丰富组成部分，**
(wǒmen wénhuà de fēngfù zǔchéng bùfèn,)
**并希望事情永远保持**
(bìng xīwàng shìqíng yǒngyuǎn bǎochí)
**在正式与非正式**
(zài zhèngshì yǔ fēi zhèngshì)
**之间的模糊地带。**
(zhī jiān de móhú dìdài.)
**另一些人则认为**
(Lìng yīxiē rén zé rènwéi)
**这种模糊地带会适得其反。**
(zhè zhǒng móhú dìdài huì shìdéqífǎn.)
**无论如何，重要的是**
(Wúlùn rúhé, zhòngyào de shì)
**要明白，**
(yào míngbái,)
**目前我们生活在**
(mùqián wǒmen shēnghuó zài)
**这种情况下。**
(zhè zhǒng qíngkuàng xià.)
**对这一价值的考量、**
(Duì zhè yī jiàzhí de kǎoliàng,)
**相关的公共讨论**
(xiāngguān de gōnggòng tǎolùn)
**以及法律随时间的**
(yǐjí fǎlǜ suí shíjiān de)
**不断进步，**
(bùduàn jìnbù,)
**不可避免地构成了**
(bùkě bìmiǎn de gòuchéngle)
**马来西亚之所以成为**
(Mǎláixīyà zhī suǒyǐ chéngwéi)
**马来西亚的一部分。**
(Mǎláixīyà de yībùfèn.)
---
Additionally you can :
"tak faham kenapa ada orang marah - bende itu sudah termaktub"
ekshellii ... saya ada pandangan lain sikit.
1. Perundangan apa-apa pun, boleh ditukar, SEKIRANYA, adanya sokongan majoriti. Kita abaikan topik ini dulu.
2. Bahasa tidak diwajibkan bagi penggunaan rakyat dalam perundangan. Ia mampu diundangkan begitu. Saya galakkan orang ramai cuba mengundangkannya - sebab saya pun dah penat dengan ...
3. ... isu orang A tak selesa dengan pendirian orang B, tapi tak nak mengundangkan isu itu. Isu hati kurang diwajibkan dalam perundangan. Jadi kita sepatutnya abaikan isu hati orang lain. KALAU satu isu penting, ia patut DIUNDANGKAN bukan biar sebagai isu hati saja. Contohnya : "sayang", "hormat", "setia" - perkataan ini semua tiada maksud, kecuali dalam otak setiap pemikir berasingan. Kalau nak membawa maksud serentak bagi rakyat, kena pakai undang, "jangan pijak bendera - kalau pijak masuk penjara" dan selainnya.
4. Ok. Sekarang kita sampai pada isu "mengapa kita takut, marah, duka, amok, dll dengan keselamatan institusi, walaupun sudah ada dalam perundangan". Pendek kata, ini bukan kecuaian. Semua gerakan perkataan di tempat awam, di media massa, surat khabar, TV, radio, medsos, sekolah awam, tempat awam ... BAKAL DIKAWAL SESIAPAPUN, jadi kenapa SESIAPAPUN, tidak nak mempergunakannya bagi mengukuhkan kuasa PARTI SENDIRI ? Kesemua isu ini adalah operasi perhubungan strategik dalam jentera politik sejak zaman dahulu, sejak TunM and mungkin boleh dikatakan sebelumnya. Bende ini memang KHAZANAH ZAMAN PERANG DINGIN.
5. Jadi siapa nak gerak untuk membatalkan politk berasaskan dengki dan pencacian antara kaum? Siapa lagi? Kalau bukan rakyat, yang SEDIA, dan MAMPU, dan DIKURNIAKN DENGAN MANDAT, siapa lagi? Itu sebabnya kita mesti galakkan anak muda untuk FASIH KENEGARAAN : perundangan, jentera politik, ekonomi, tek, dan kesenian perhubungan. Kalau tidak, BIARLAH ORANG LAIN TADBIR.
6. Nampak tak saya tidak memihak kepada ketuanan Melayu atau ketuanan pukul rata ... yang saya ingin sampaikan : dunia bukan tempat yang selamat. Keselamatan hanya dimenangi dengan kerja yang rumit.
Seriously like - fuck this kindergarten country - I would rather be spending my time studying ML :P
But you look at the numero uno country on earth, and it is no different.
So you have no choice : political participation is social work.
Okok, I keep forgetting to jot down these two notes from running ... so before I go on about computer vision, I should do this. These are just operational recoveries of forms lost over the years, so it's a bit of a log of shame, albeit a somewhat technical one.
2026-05-23
Ok - back to what I was INITIALLY thinking about this morning : neuroendocrinology of transitions between sleep and waking.
PRE-READING
Observed :
Inferred :
READING yields ...
TANGENT : ... Mechanotransduction ...
90a. Endurance training -> thyroid etc. hormones -> AMPK-SIRT1-PGC-1a (protein) axis -> mitochondrial biogenesis -> adaptation towards slow, type-1 fibres
90b. Power training -> androgens ( testosterone etc. ) -> hypertrophy of existing fast, type-2a/2b fibres
90c. localised myostatin suppression -> hypertrophy in all of type-1/2a/2b fibres
91a. AMPK-SIRT1-PGC-1a Triggers :
91b. Activated PGC-1a moves to the nucleus, where it coordinates BOTH nuclear and mitochondrial genomes!
92a. Hypertrophy Triggers :
92b. satellite cell fusion
ASMR* : in terms of brain-networks, neurotransmitters, and endocrine response.
( I was reading up on motor neural voltages, between sleep and non-sleep states, and ended up at ASMR. )
Perhaps this is relevant to my preferred state of mind, as a generalist, whereby I allow myself all-dimensional reactions to stimuli.
ASMR apparently reflects a "higher than usual" synchrony between the CEN* and DMN*. The SN* is characteristically less-engaged, whereas it is usually responsible for flagging issues, and raising them into priorities, which then interruptively switches between CEN-dominance and DMN-dominance.
Generally : this appears to be a risk-off state ( see : dACC* ), where sensory inputs of ambiguous risk are not interpreted immediately as threats, but are instead allowed to be contemplated as amusements ( see : NAc, VTA* ). The states of threatened and amused, each involve separate neuroendocrinal cascades, which are available for further reading.
I reflect also on how interoception is often referred to as "subjective feelings" in the literature - whereas I find that subjective feelings may all be quantified in terms of concrete sensory data types.
*
Yeah, one of the things putting me off neuro studies for so long was the word soup.
Pasal "less of everything tu" nampaknya contoh anda lebih pada natrium.
Dapur saya, saya asingkan garam2 ini :
Biasanya : makanan komersil/terproses mengandungi terlalu banyak Na, dan kurang K, Mg, Ca
Jadi unsur2 K, Mg, Ca ... you boleh cuba mainkan sikit lah, sebab ia memang perlu tambah, dan juga rasanya lain.
"Sparse autoencoders / SAEs" - if you run into this jargon, it just means "capability to analyse a complicated issue into multiple simpler components". Basically middle-school reasoning / critical thinking / analytical philosophy.
However, with regards to the implementation of SAEs in "transformer" architectures, it just gives you a better stochastic parrot, as it doesn't introduce things like rules or logic programming at all. It's just an increase in the resolution of a probabilistic reasoning algorithm.
Konsep mukallaf mungkin menggagaskan suatu lokus bimbingan bagi rakan-rakan saya yang berkepercayaan bahawa tuhan menginsankan manusia.
Saya, yang berkepercayaan bahawa manusia telahpun mencipta gagasan tuhan, meminati isu ini dari segi antropologi kebudayaan.
Kini ahli falsafah, iaitu pengurus risiko, amat menumpukan perhatian terhadap keselamatan pengunaan AI, dan tahap pengiktifrafan personaliti undang-undang AI.
I don't agree with corporal punishment because it's simply not effective. It may stop certain behaviour, but it doesn't modify the cognitive apparatus of the actor, coherently 🙂
Likewise with the death penalty : if we approve of the death penalty, we should kill more weak people in society preemptively also lol.
Just caught up to this 😅 : so OpenAI's Triton is a pythonesque DSL for exposing hardware primitives using a "tile" paradigm.
Handles hardware for you
You handle hardware :
Triton is a dialect of MLIR@LLVM ...
Brief dip into computer vision jargon. Anthropomorphism. Humans learn to recognise gross gestalten, before understanding mereology ... and humans learn to recognise 2D persistence before 3D persistence. But as conscious processes develop, they discover more efficient storage formats for data, by relying on 3D persistence graphs.
So while we could train a vision system to be more human like by beginning with gross gestalten, we may want to skip forward a little bit to endow the system with the capability to immediately begin forming a database of 3D graphs, representing object topologies in the wild.
LFG :
0. Statistical globbing via neural networks should be used as a final optimisation technique, only after all known structural optimisations ("physics") have been ("logically") programmed into the system - not as a lazy shortcut from zero.
1.
1.1. Let us throw away the concept of colour, and begin only with one dimension of magnitude per unit of space - luma / luminance / brightness.
1.2. Let is ignore 3D structure and begin only with the gross phenomenology of the field of vision.
2.
2.1. Now, mereology : we must have a way to describe subsets of any field which are interesting. The smallest item is a point of brightness.
2.2. SCALE invariance : the concept of an item growing to take up a larger fraction of the field, or shrinking to take up a smaller fraction, must exist. See SIFT, SURF, ORB, et al.
2.3. Location : the field of vision must be referenced by an address space.
2.4. Multiplicity : there may be many items of interest.
2.5. Extension in 2D : items may be linked, giving them 2D structure in the field. See "graph convolution networkS / GCNs"
2.6. ROTATIONAL invariance : the 2D structure is different if you spin it around with respect to the address space. Where you apply this to spinning everything around the system, see "sphere node graphs", though in 2D that's probably a "cylinder node graph".
... at this point we have five to six concepts that must be uniquely addressed in the cognitive space of the system. A sort of introductory transcendental aesthetic.
3.
3.1. Extension in 3D : Now we can revisit the notion of 3D space.
3.2. Location ("distance") in 3D is associated with scale in 2D (2.2.).
3.3. Occlusion in 2D is associated with rotation in 3D. See "aspect graphs".
... and, I think that's all we need to get starting building a computer vision system, to undergird the visual ontological comprehension of any anthropomorphic computer.
4. Oh yes, we can then revisit colour and motion in the end.
OK - I've figured out the theoretical solution to Malaysia's public education historical sources controversies.
kita kena ajar pembaca buat lit review la ... sebab ramai orang tak tahu sistem petikan / citation yang berdasarkan peer review tu
... come to think of it, zaman kini : besnya kalau kita start setiap buku teks KPN itu ada citation betul-betul pada akhiran buku
Finally got around to reading a definition of JEPA. Based on this limited skim, my first impression is that it's still operating on data structures which are too abstract, from a top-down approach, instead of just building a world-view from the bottom, upwards.
I'll keep tabs on it, but of course I don't have time to dive in right now. I have already discussed what I consider to the the basics : anthropomorphic AI must have (1) sensory data structures (2) spatio-temporal modelling (3) hypothesis formation and testing.
Meanwhile the probabilistic relationship between high-level abstract ideas like "dogs in walkers" is simply "less reified" or "more abstract", and basically will display the behavioural inefficiencies associated with reasoning about platonic forms in general, without instantiation.
https://youtu.be/oM4neOyZOi0?si=YFY5qnse8vb3MyWh
Update : oh, hypothesis testing is in the 2022 paper, we're good
( a reverse list of what I would prefer to forget first ) :
Phainomechanics. I may start using this word to talk about physical mechanics in phenomenology.
Synesthesia. Mysterious, perhaps. Inexplicable? Absolutely not - it may seem odd that a 1D signal like sound translates into a 2D signal like vision. But hold on ... vision is a 2D phainomechanical process on an intrinsically 1D signal. Essentially we animals hallucinate colour as a matter of convention : the three (or four colour) receptors we have, are discretising samplers of a continuous EMF spectrum.
Imagine if sound was discretised in such a way. Not hard at all, with a bit of practical thought. Imagine if smells, pressure, vibration, heat, and pain were too. Actually we do not really need to imagine ... we simply have many of these discretisations, already.
All human qualia are hallucinations from a physical point of view. The intrinsic nature of qualia are therefore most practically characterised as data types / data structures, in an information system.
Hari2 medsos mengalir kepada kePERLEMBAGAAN berita. Kisah seterusnya tak mungkin- buat bayangan saja. Klo pada 1957, ini permudahannya:
Saya rasa, tumpuan ekonomi 1957-2026 jauh lebih tinggi, kalau bukanlah kenyataan kita, yang saya cirikan, Kedukaan Sebangsa M'sia
Bab perisian sumber terbuka kerajaan ni : masalah teras ialah Malaysia tidak mencita2kan ketamadunan STEM yang berkepimpinan di dunia. Dengan itu, halaju orbit tak tercapai. Unsur2 masalah dasar :
🤣🚧💥
-- I'm not the sort of person to be passionate or sentimental about my country. Commentary like this is, to me, the equivalent of picking up rubbish where I happen to live.
I live in Malaysia ( and related socmed ) so I spend more time discussing with Muslims how Islamophobia works, than I do in European circles.
Subsequently the shortest path to peaceful co-existence, is to make these premises EXPLICIT, in legislation. Western states should recruit citizens with *explicit* frameworks for how [ individuals coming from Premise 2, "Muslims" ] will be treated.
Correspondingly, GLOBAL-scale Islamic authorities should provide explicit frameworks for how [ Muslims ] should engage with [ states asserting Premise 1 ].
All of this sets up PUBLIC protocols for engagement, shifting from high-context ambiguity, fraught with fear, to low-context confidence, and hopes for civilisational progress on a heterogeneous Earth.
This is turning out to be an interesting group of molecules.
Dorang tanya, STEM atau KESENIAN, lebih penting?
Dengan terperinci.
0. Seni
Secara am, semua kegiatan itu ada perseniannya.
KEMAHIRAN = KESENIAN LAH XD
1. Sains (Stem)
Ilm mengutamakan pemantauan zahir / empiris. Jadi kesenian STEM senang mula dikaji dari sini.
2. Matematik + Logik (steM)
Kesenian hisab + mantik adalah kerangka batin, yang menyekat Sains.
3. Kejenteraan (stEm)
Inilah seni kerja amal.
4. Teknologi (sTem)
Sina'ah hanya istilah am bagi kesemuanya.
Yunani, "tekne" => technology;
Latin, "ars" => art.
__TRAVELLER__
"How do I find something firm to latch my feelings and understanding to, on the [ war in Iran ]* ?"
( Where * can be replaced with any complex geosocial phenomena. )
__GUIDE__
There are some separate "planes". You can cluster your study of each plane, then combine them.
Plane 1.
Abstract concepts in : sociology / psychology / anthropology : such as "The Clash of Civilisations", someone mentioned ( in chat ).
Plane 2.
Linear narrative of time's arrow of geopolitics : today, Google gives great bullet point summaries, and links to wikis or academic literature. ( Evaluating academic sources is a separate question, which can spin off to a new chat. )
Plane 3.
Lens of each tribe ( non-linear time, the subjective experience ). Or the effect of 2 upon the fabric of 1. Reading method can be similar to 2.
__PROMPT EXAMPLES__
Plane 2 :
"Who are the main groups of stakeholders, and actors, in [ the current war in Iran ]*"
ANSWER : ABC; now you have to keep your own notes! So ...
Plane 3 :
"What is the history of A's view on topic-X. List major events and shifts in perspective over the last 2000 and 200 years."
ANSWER : DEF
"Tell me more about the details of D."
ANSWER : HGI
__MAP-REDUCE__
So ... your reading will be this huge cube of :
( parties : A+B+C )
* ( Abstract Issues : X+Y+Z )
* ( ( Concrete events : D + E + F ) + ( Details : H+G+I ) )
__APPROACH__
No need to rush. Understanding people takes time.
Today's fun was going through TIOBE and drilling into highly-ranked but otherwise obscure contestants :
* (a) => "apparently"
Perhaps I have too much trauma from past learnings. Like, childhood learnings, before I started self-guided studies at 18. Often enough I run into a list of things I don't know, and I become afraid. Then I recollect myself and procrastinate before finally getting to the study. The study is usually easier than anticipated.
But this is a bit of an oversimplification. Based on my biased view of my own track record, my normal acceptance for stress is something like "75% of the time unsuccessful, with accompanying hormonal and neural responses", and "25% successful, ditto". Which is to say, I spend most of my life depressed, and make up for it more than adequately otherwise.
Then I reflect on the sort of people I think are foolish, who seem to spend the majority of their time talking about success, but then have these minority periods of time when they are depressed. But after all, the averages are equivalent, I suppose. So it is just a preference about how one wishes to spend time.
