Суперкомпютърът Петак I / Юнак I

 

Свръхсметачът Юнак I / Петак I

Пророците на мислещите машини - основен том

• Новини от края на 2025 г. - както споменавам в други части от „Пророците…“, Вседържец, Емил и другите мислещи машини ще работят на такава техника, с каквато разполагаме. Започнах да изграждам „беоулф“ клъстер: самоделен „суперкомпютър“, засега съставен от множество обикновени ПС-та. 
  (...)

https://github.com/Twenkid/SIGI-2025/blob/main/petaki.md

Read More

Super Smartasses - improved colorized version of the graphic satirical story about the semiconductor and high-tech industry. The main volume of The Prophets ... and SIGI-2026 and Genesis: Creating Thinking Machines are coming

An original, unusual and very funny glimpse into a corner of the semiconductor industry,  presented in the pilot episode of a series conceived in 2009 and first published in 2012. 

Have fun and support The Sacred Computer: Thinking Machines, Creativity and Human Development

The main volume of "The Prophets of the Thinking Machines..." will be published soon and the next virtual conference: SIGI-2026 is starting on 1.1.2026. Be an early-bird there, LOL. :))

The next work is "Genesis: Creating Thinking Machines" and it is not designed to be "just" a set of 20-some books, monographs and papers, more than 4000 pages. (...)

 https://twenkid.com/smartasses/2025/

Todor Arnaudov

2 д ·  https://www.facebook.com/twenkid/posts/pfbid02FsMjrhjxtAjod6G9dvJ5biSLJSordKfFtKTc2Nt7qgGKhaB5o1zhCa6FXjus1JN8l?__cft__[0]=AZbmn5X2Pabyo5Xqx5YH0eGZOfV5ZJRR1WomuTYGA6N11fN3O90f8cJWcaw-TtC_g-Q80J3fm4AGTebehGjra9rwLVKyhuZkWwx4OXbGGLJFM_SOdU689xhSPCgPGfBYnbRP32Dj7zcGFgTltVM7gK1SkmnDdeQhkTo1erqQowTWlQ&__tn__=%2CO%2CP-R

Споделено с: Публично

Премиера на оцветената и подобрена версия на пилотния епизод на сатиричния ми комикс "Суперумници" (2012) - весели приключения и абсурди от полупроводниковата индустрия и живота: https://twenkid.com/smartasses/2025/

Първи епизод: "Спътници" (Companions)

* Идея, текст и рисунки с молив: Тош (Идея и описания на няколко епизода/скици на сторибордове: 2009 г.; първи осъществен голям пилотен епизод: лятото на 2012 г.)

* Оцветяване и подобрения на някои от рисунките (12/2025): Тош с Gemini и единични "ръчни" (наложи се едно завъртане: Horizontal-Flip;*) Някои от рисунките от черновата са запазени както са или само леко са оцветени при точно запазени контури и като цяло се стремях да се следва композицията, формите и стила на оригиналите с изключения, където не беше възможно или пък ми хареса както е. От линковете горе може да се види и оригиналната версия с молив.

* Сюжетът е вдъхновен от легендарна компания, която по-късно беше закупена от "Broadcom", след като много години работеха в тясно сътрудничество. "Броудком" днес се води с капитализация около 1.67 трилиона долара, около 2008-2009 - 3-4 милиарда (по-късно ги купуват за 37 милиарда през 2015 г., и после купуват и въпросната друга компания през 2016 г. и приемат името ѝ) Дали някой колега е бил "играч" и си е купил акции навреме?

...

* С повече дорисуване на скици или скиците можеше да се подобрят още образите, но не ми се занимаваше достатъчно за, само пробвах някои подсказки и не се получаваха. ChatGPT translation:

Premiere of the colorized and enhanced version of the pilot episode of my satirical comic “Superumintsi” (2012) — cheerful adventures and absurdities from the semiconductor industry and from life:
https://twenkid.com/smartasses/2025/

Episode One: “Companions”

• Concept, text, and pencil drawings: Tosh
  (Idea and descriptions for several episodes / storyboard sketches: 2009; first fully realized large pilot episode: summer 2012)

• Coloring and enhancements of some of the drawings (12/2025): Tosh with Gemini, plus a few manual touch-ups (one horizontal flip was required).*
  Some drawings from the draft were kept as they were or only lightly colorized, with the original outlines preserved. Overall, I aimed to follow the composition, shapes, and style of the originals, with exceptions where that wasn’t possible or where I preferred it as is. The original pencil-only version can also be seen via the links above.

• The story is inspired by a legendary company that was later acquired by Broadcom, after many years of close collaboration. Broadcom is now listed with a market capitalization of about $1.67 trillion; around 2008–2009 it was valued at $3–4 billion (they were later acquired for $37 billion in 2015, and then went on to acquire the other company in question in 2016 and adopt its name).
  Did any colleague happen to be a “player” and buy shares in time?

…

• With more redraws or further refinement of the sketches, the characters could have been improved even more, but I didn’t feel like investing the extra effort— I only tried a few prompts, and they didn’t turn out well.

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

The Sacred Computer's Lab In the meantime, now that I’ve finished the major “time waste” of the Biblical book surveys and defending the priority of my contributions, it’s time for real work. I want the handicap I gave to the AI opportunists back! :))

You can help me if you wish: donate hardware or services, support me, connect and help in the development and experiments.

According to early 2010s "wishful" ideas and designs, while working on the Infrastructure, and to what was counted as an AGI, universal artificial intelligence, a thinking machine, capable to think and create at a single human/super-human level/speed/scale/performance/quality/complexity level*, were "dreamed" to work on a quad-core PC with 32 GB RAM and a desktop GPU from that era - equipped with sensors: one or several cameras, microphones, Internet and other sensors, used for bootstrapping - without pre-trained and existing modules which are now available, we may not count them in the requirements. :) 

Whether that would be enough - we will see.

Ask if you are interested in these articles or search the blog in 2013-2014, Issues on AGIRI list, "What's going on - Everything", "The infrastructure" (now called Vsy, Jack of All Trades, Вседържец etc.)
* I.e such a machine is not required to be capable of generating photorealistic 4K videos in real time - this is vastly superhuman. It should be at least as fast as me or reasonably slower. 

Read More

The excessive spending for "The AI Bubble" is not *a mistake* and it is not about AI - that is how the system works

A recent: post from Linkedin: https://www.linkedin.com/feed/update/urn:li:activity:7405447973154844672/ 

Todor ArnaudovTodor Arnaudov • YouYouThe Sacred Computer: Pioneer in Artificial General Intelligence: Thinking Machines, Creativity and Human Development [Looking for Partners, Collaborators etc.]The Sacred Computer: Pioneer in Artificial General Intelligence: Thinking Machines, Creativity and Human Development [Looking for Partners, Collaborators etc.]1w • 1 week ago • Visible to anyone on or off LinkedIn

The excessive spending* for "The AI Bubble" is not *a mistake*. It is not because "they don't understand the laws of scaling" etc. (see the comments to the Gary Marcus' post). This is [just] capitalism. (...) The goal of economy & the "leaders" of the "AI bubble" and other bubbles or apparently excessive overspendings and mindless expansions is not for being economical, to save resources etc. overall. The same is with spending for the Military ("Defense"), e.g. the EU program etc. - build weapons, apply them "creatively" and then rebuild = "efficiency", LOL - it generates "economical growth". Apparently the goal is to MAXIMIZE THE EXPENSES (which come from another source) and to cheat and deceive your "host" source that you optimize and that you spend the least possible (not enough), in order to take even more resources and to spend even more: this is [also a pattern of the] "consumer culture" and capitalism, not [of] "AI", LOL. 

In the case of the military spending: [the goal is] to cause maximum losses for the enemy,

[But on the way, the weapon producers cause "economical losses" for the suppliers of the resources for the production as well, because they extract raw resources, "workforce", energy, money, material from them and send the produced "goods" to the war theatre in order the production to be "utilized" and cleared, which helps to  create room for more new production and to justify the need for more resources.]

E.g. ML globally is paradoxically similar - its goal is presented as "optimizations", but how many times the same datasets are "relearned" from scratch on how many places and how many GPUs, how many useless benchmarks are testing things which are already known, improving the "SOTA" with 0.4 points, 4.2 points on MMLU, XYZA*. Thousands of papers [are] repeating almost the same, all countries [are] having "National AI strategies" and institutes (clones of the 2003 original with exponential financial requirements), yet the most of what's done everywhere is replication & multiplication with minimal variation and maximum spending - just like with the living organisms and their goals.

A lot of the investments, [especially some huge ones], are possibly money laundry or a cash-"circulation" mechanism, as with the famous cases [with the biggest players in this game]. The investors may be aware that they will not get return of their investment from "sales", however they might not care either. :)))

In addition, a lot of the recent money are probably printed from nothing - thus spending fake money by making real workers produce money from other branches of the economy and [then] redistributing the wealth. [For the ones who "produce the money" the nominal value don't matter - the smaller ones lose buying power.]

If many "small players" enter the "stock game" and invest "real money", earned by hard work, salary etc. & the bubble bursts, the loss will be mostly theirs. The big sharks won't sink, just the money of some people will be "reset" in order to eliminate them "from the game" [ - they will have to work hard in order to earn them again - if "there are any jobs already".]

* The broader and apparently paradoxical maximization of the inefficiency is explained in a more universal sense in the small monograph "Is Mortal Computation Required for the Creation of Thinking Machines?" at SIGI-2025* 

Globally nothing is "optimized". All is a trade-off - converting one currency to another one and cheating by accounting only a cherry-picked tiny fraction. (... https://twenkid.com/agi/download.php?file=Arnaudov-Is-Mortal-Computation-Required-For-Thinking-Machines-17-4-2025.pdf

hashtag#ai hashtag#aibubble hashtag#agi hashtag#thinkingmachines hashtag#sacredcomputer hashtag#sigi2025
Self-Improving General Intelligence 2025: https://twenkid.com/agi/

* Comment: "Purnima Gauthron • 2nd Engineering | Finance I “Artificial Intelligence” Technology Fact-checking (edited) 4d

Gary Marcus This is how Scale is All you Need looks: More than ~1500 AI startups with valuations of over $100 million, including ~500 AI “unicorns” are merely scaling their proprietary large-scale foundation AI models — the "operating systems of the future" so to speak — consuming enormous amounts of resources and compute on an infrastructure that will cost $1 trillion a year.
"
 ....

 https://www.linkedin.com/feed/update/urn:li:activity:7403859708693913600?commentUrn=urn%3Ali%3Acomment%3A%28activity%3A7403859708693913600%2C7403981868578369537%29&dashCommentUrn=urn%3Ali%3Afsd_comment%3A%287403981868578369537%2Curn%3Ali%3Aactivity%3A7403859708693913600%29


https://www.linkedin.com/posts/gary-marcus-b6384b4_scale-is-all-you-need-is-dead-activity-7403859708693913600-0Hcj?utm_source=share&utm_medium=member_desktop&rcm=ACoAAADvqZ8BN8iTwroeQOUlHeE-

Read More

Power Overrides Intelligence - AGI List and SIGI-2025

Power Overrides Intelligence:

Answers to Matt Mahoney’s summary of LessWrong’s

“The Problem” regarding the so called existential risks of

Artificial General Intelligence and Superintelligence

in August 2025

Todor Arnaudov, James Bowery1  vs Matt Mahoney1

https://twenkid.com/agi/download.php?file=Power_Overrides_Intelligence_Todor_on_AI_Aligners_SIGI-2025.pdf 

See also my long letter to Oxford University institute from 2.2012, where I discussed ethical issues and topics, which later became current: https://artificial-mind.blogspot.com/2012/02/philosophical-and-interdisciplinary.html  

In one of the threads I concluded that: Wisdom is powerless against the physical force. 

Read More

Intel Will Lay Off 15000 Real Engineers: The Bulgarian Predictions about the Semiconductor industry

 

A funny part from the "Bulgarian Predictions" (or Bulgarian Prophecies) which were mostly about AGI, AI, Theory of Universe and Mind,  generative AI, the phenomenon that "Creative intelligence will be first surpassed and blown-away by the thinking machines ... and intellectual work will be done in 1 ms for free - the smart "asses" will be useless - not that we WERE useful before that" (2013)

Yesterday I realized another prophecy of mine, which guessed one exact number* in the pilot episode of  my satirical comics series "Super Smartasses", that presented the world of the Semiconductor industry  in late 2000s.


First drawn and published in the summer of 2012 (colorized recently with Gemini)

MENTEL and AND lay off 15K Real Engineers!

On 1.7.2008 the newspaper reports (with filled words for more complete syntax):

" ... The VP of Mentel ... Sorry, I need to save ... I wish them all the best - they are real professioals" - the VP said ... The financial crisys ... World economy ... New Cure 2 CPUs ... But Banks ...  Low Sales ... Revenue
...

Next Issue will be with a reduced print of 15K - pre-order NOW!

Vice President's Private Island 
 (...)"


The draft version of the comics: https://twenkid.com/smartasses/1/en.html


https://www.google.com/search?q=Intel+layoffs+15000&oq=Intel+layoffs+15000&gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIICAEQABgWGB4yCggCEAAYChgWGB4yCAgDEAAYFhgeMggIBBAAGBYYHjINCAUQABiGAxiABBiKBTINCAYQABiGAxiABBiKBTIKCAcQABiABBiiBDIKCAgQABiABBiiBDIKCAkQABiiBBiJBdIBCDYzODFqMGo3qAIAsAIA&sourceid=chrome&ie=UTF-8

12 years later, summer 2024

Intel Will Lay Off 15,000

August 7, 2024 Written by Oregon Business

The semiconductor manufacturer will cut 3,000 employees in Oregon as it looks to trim $10B by the end of next year.

Intel axes 15,000 jobs after sales tumble

2 August 2024

Nick Edser & Natalie ShermanBusiness reporters, BBC News

https://www.bbc.com/news/articles/c16j3318n08o 

I will release the enhanced colorized version of the draft after wrapping it up - I'm still refining some of the images. Nano Banana doesn't always "listen" to all of the instructions and if you get one thing correctly, another onject may disappear etc. 

Read More