-------------------------------------------------------------------------------------------------------
"TL/DR: short  30sec recap: @ LLM NeuralNetwork tech's benefits overshadowed by massive hoarding of computation resources and unethical and aggresive data appropriation that spreads disincentives to the creative soul of the internet and kill the pc market. All pathalogical aspects accelerating due to big state support. Although LLM represents a small fraction of AI research it's  ultra-hyped-posing as AI to further accelerate funding and support through FOMO and psyops. The large scale of resources needed to sustain its usefullness highlight how inefficient it is.
-------------------------------------------------------------------------------------------------------

Alt titles

How to get a good education for your Digital LLM driven Neural Network for Free?
Welcome to The magnificent Era of the Ultra-hype engines and the comp-resource hoarders and data stealers  ...

Analogies

In a dead poet society Sam wants to become a member. Sam comes to meeting fully equiped with recording tech to secretly train his LLM-NN to produce the best poems. Do you keep him into your society? What if ALL members had the same equipement and goal ? But wasnt that the goal from the start? To train your brain to the best poems ? But now a member wants to train ALSO his digital NN ? Is that bad ? Even it he doesnt do it secretly ?

You train your cyberpet (with LLMNN module) when you take walk with it at places with crowds . (as you would do with your kid) . Is that bad ? Illegal ? You train a swarm of cyberpets (all owned by you) again in public spaces . Is that bad ? Illegal ?  Again you are in a public sphere ? What significance it has if you train one cyberpet from training a thousand or a million ? Maybe the imbalance of control that YOU can aquire and exert on you enviroment ?

Bad news

I read increasingly reports-complaints of websites creators and maintainters on the negative consequences (mainly less traffic and mass scraping that create also stability issues) of aggresive AI crawlers and search engines moving away from the links list mode to an AI-centric UI. (and of course the most important -a wave of layoffs in various tech related sectors).
What signifies as aggresive behavior on cyberspace ? Can a crawler misbehave ? Is there justice on the cybersphere or a way to fight power imbalances ?

Neo Luddites ?

One natural question : So hasnt that happen before?  New tech creating new jobs , replacing old ones? Opposing AI means to be a new version of Luddites ?
Wont AI offers us a new brighter future so lets not allow petty complaints to hold us back ? Shouldnt we play along and make our sites more AI access friendly ? Or maybe become youtube content creators-workers ?   Wasnt web build from CERN scientist to interlink-interelate all accessible to internet knowledge .

Concentration of comp power ?

I will argue in favor of a position that the biggest change that is happening in our lifes in not a new-tech transformation but a significant era of comp-related resources hoarding of the powerfull amidst what seems to be a prelude to a world war. A society transformation to less liberal forms . A 'transfomation' that kills the soul of the truly biggest tech of  2000s , thaf of the internet. A transfomation that opposes creative spirits and upholds control as the highest value.That transfomation (a hoarding accelaration) is possibly because it stands on the shoulders of already pathological wealth and power concentration that had been created in prior time . And surely the current data stealing is facilitated from the eco-hostile smartphones!!! The smartphone era is the prelude to the current era. Massive personal data is comming from the smarfphones. Without smartphones all data centers would die. LLM is a tool to access more easily all the huge data . BUT you need ram to access it and store it and gpu's to process it. And another creek leading to the 'AI' river is the cloud era. OSes like windows, ads trackers , and many programs collect vast data and 'telemetry' that needs to be stored and processed. So we have an explosion of digital data creation.

Can it match a billion years biological evolution ?

So i dont buy the ultra-hyped LLM tech as fundamentally AI innovation .
It's a complex LLM that for the moment  needs access to vast comp resources as ram and gpus and ofcourse data in order to create a usefull service.
And for me that exposes it's weakness of the digital computing. It's doesnt align well with complexity. It drains resources .
I think a single cell of a bee is computational way more capable  and utilizes resources many orders of mangitute more efficiently than al l the LLM data centers. Is that an hallucination of me? wishfull thinking?  I think AI hype works better as the world becomes more dump and enumerated. But i think it's not a peer to the biological world that has millions of evolution on it's back . Billionaires surf on a media concetraded world that acts as an ego echo-chamber and hides the wrong doing of their actions. See for example climate instability  not to mention playing global war with nuke threats!!

Is it just another episode of lust for power ?

I see pure peaked greed and immoral content appropriation-stealing . Look the perpetrators. Multi billionaires with the ethical flexibility of saluting like nazis to attract media attentions, fantasizing replacing the workforce of their companies with indians, posing happily near trump and near dictators constantly promoting their ideas thought controlled appearances where they parrot in front of friends or employees how great they are...  (put them in a room for a real - uncontrolled discussion with a philosopher like Slavoj Žižek and see the tech feudokings's ideological hollowness..). So are these people the AI innovators? The nvidia gpu gaming guy who ripped off for decades family budgets and now poses as a tech lord ?  really ?

The hyped AI bandwagon ...

That is no news ofcourse. Remember github and copilot getting in bed with microsoft? They big corps always are after our data. I think the difference now it that big govs are fully onboard the AI train  because i think they believe that it's a matter of state survival and a way to hide many of their ineffeciencies under the new hype tech word..  A gov with the bigger AI survives and wins. Without government money i dont think a company could buy mem - gpu production for the next years or build easily data genters.  So there is a AI ultra hype train that is moving fast and attracts on board  stateactors , aggresive companies and people who see a window of opportunity to tap on water and energy with more ease. That train hoard comp resources and public data very fast - cancer like fast- and nothing stops it. I believe that a crucial fuel of that train is the big hype engiges (that ofcourse belong to the same persons that drive the AI train.. how convenient and illegal and unethical...) . Those engines with constant hype-flows keep the train going full speed attracting resource hoarders . AI or die!! (i hope you see that most propably those on the AI-hyped train wont die... not because they are more AI-smart but because they will have stole massive water , energy , hoard all memory and gpu under the sun.., and ofcourse have access to gov public money and support.)  (any free market funs left ???) (do the free world turns into north korea? or we are in the middle of WW3?)

The state-sponsored hoarding and datacenters expansion that destroyed the pc consumer marked highlights that argument.
Without vast memory and gpu , and energy and water that powers the datacenters   LLM is would be a search engine language driven interface.
(speaking of AI_LLM capabilities.. have they learned to count in images yet?). 

It's the data that train an LLM and makes a 'AI' agent usefull. If an 'AI' is asked tips about living in Berlig it will ultimately have to had be trained on a good reliable blog. Otherwise it'll recite in perfect english just crap.  An LLM agent can write perfect prose but say crap. So the ethical would be ? To buy the data to train your model  . Or to allow third party to use LLM to offer an language driven UI on their work.  Now we see fast movers tapping unethically to data and sell it AS if their agent is smart !! Isnt this an AI parody salespitch of conman sales(wo)man ?

Breaking the AI hype cycle

So for me the solution it to break the AI hype cycle. Expose the crucial element of greed,and data stealing and resource tapping. Expose the superficial hype-building claims of AI (when for example the hard problem of consciousness is out of reach). But that hype is supported by the great hype engines of our time . (facebook,twitter, etc). Breaking the hype ofcourse means engaging in a civilised discourse in public spheres not controlled by the AI-train hoarders. So in the end it boils down for me in a matter of how societies and political systems work. I fear that in our days states tend to align more with a china-concentrated model keeping elections as a thin vale constantly vanishing.

In eras of power concentration in public spheres what should be our response ?

• is encryption combatible with public-open spaces?

• Should we silently keep offering our source code to the cyberpets of Sam and Elon or China's superstate ?

• Should we practice  @ abstinence from the digital public sphere while is infensted ?

• Should we practice partial abstinence from the digital public sphere while is infensted and increase presence in  physical public spheres that build stronger semantic networks that are resistant to automated crawerls ? (physical public spheres can add flavors to languages unknown to crawlers)

• Should we try to avoid feeding some cyberpet swarms ?

• Are there digital content not digestable from automated crawlers?

• A data type-format being less machine-searchable is a non-digestable-crawler data trait.Podcast vs text vs image vs video

• Should we protect content behind authentication ?

• Can we devise processes as @ sentinel species (like the dogs in the Terminator movies that detected infitrator units?)

• Should we inject older tech like phone conversation (like BladeRunner's Voight-Kampff) tests during authentication  to our digital realms ?

• Should we rethink the internet's architecture in order to allow public sub-spheres-spaces to have a way to protect themselfes from centers of accelerating (out of normal pace - cyberflows homeostatis braken) data appropriation ? Or the internet's architecture is fine and we can find solutions inside it?

• Can you create honey pots of garbage and lure a crawler to get that and bypass what is truly significant to a website owners

• Combine non-automated - undigestable to ai-bots authentication , data flow rate , data types ? How can you make authentication really bitchy to automated cyber bots .What would piss of a ultra-greedy webcrawler? How would you treat such a kid in your famliy ? Keep less food(data) in the fridge ? Selling more content in physical format ? Or again .. through telephone conversation ? 

• Should communities engage more in live voice-teleconference conversation using encrypted channels?

• Should we communicate with audio files messages ?

• Elevate the prestige and access rights associated with a physical presence in a community gathering (layered open community)

• Attach valuable data on matter? Reducing its mobility but increasing its control ?

• Should we ask @ William Gibson ? Either for advice or apologies for making us believe that cyberspace was our world when in fact we are the aliens in it!!

• Embrace community's idiosyncrasies @

• Enhance non automated workflows (for example filter account creation through human conversation channels

• Enhance communities help availability

Please dont phone me to tell you my prefered solution.

There was an era when Sierra made money from hot-phone lines selling tips how to help LarryLaffer unzip his pants and selling physical copies of hintbooks .

Refs

@The New Era of Sensory Augmentation
“We have only the most shallow understanding of basic questions like what physical properties give rise to conscious experience, what makes an experience perceptual, and how we come to feel ownership over a sensory stream.” <--- interesting.. how do you make an information flow that is foreign to automated crawlers ? (or is it US that we stepped into the ai-llm's crawlers digital realm and IT can detect US as aliens? )Can we make info tele-flow more organic ? To feel that we own it ? Make data that is not @ fungible ? How would that help discern foreign-unwanted tapping onto a data stream from a friendly data tap ?
@Regain control of AI crawlers

pc: devuan 6
wm : xfce
network manager: (to be filled)

Product name : @Tehnoetic TET-N150 . Which seems basically a simply rebranding of 2010 Taiwan or Chinese  @ Sharp VR-WL2 or @ Wistron NeWeb/ DNUA-93F . (The DNUA-93F is on the sticker but the FCC-ID corresponds to Sharp VR-WL25) .. But it seem from thils list @ that various     AR9271 based wifidongles report to linux the same Ven:Dev ID. So i wonder if all that name jungling has any meaning...

Related product : @ Penguin Wireless N USB Adapter for GNU / Linux (TPE-N150USB), with a nice manual: @

@  (2014) FSF-certification announcement

I tested : Technoethical's N150 mini wifi@
For many years i have in my home turned towards using wired networking . The reasons  being :
1) having in many years tried various adaptors that have problems making the work reliably (even if most of them claimed also linux support).
2) being afraid of the emf pollution .
Anyway i had in my radar for many years technoethical and so i decided to buy and try Techno's N150 that claims a

As for the open firmware i found:

• @ open-ath9k-htc-firmware.

• @ Announcement of the release in the @ ath9k_htc_fw emailing list.

• @ A nov/2011 datasheet of AR9271.

From little i've read it seems credit goes mainly to : Adrian Chadd and Luis R. Rodriguez (@ (2008)Atheros hires ath5k developer)
In the Announcement email they thank management of QCA and even Tensilica .
Here @ (a page orf AR9271 specs) it says:
SoC Architecture: RISC - Tensilica core (Xtensa LX2.1.0). So i guess Tensilica is the microprocessor that executes the firmware.

The T'sN150 page claims:

Well that doesnt hold in Devuan 6 's 6.12.74 kernel.

Initially plugging in various usb3 ports of my Devuan 6 PC  i didnt see any syslog messages (nothing!).
Plugging it in a usb2 port i show a $ lsusb entry :
Bus 003 Device 006: ID 0cf3:9271 Qualcomm Atheros Communications AR9271 802.11n

But after:
# apt-get install firmware-ath9k-htc firmware-atheros 
it works ok in either usb3 (blue) and usb2 (black) ports.

Also i dont know why but my adaptor came not with the fancy logo filled small sticker but with a simpler one with made in china and some small stuff that i  dont remember.

Also according to @ the product name of my dongle is  Wistron NeWeb DNUA-93F and it contains also a usb chipset: Atheros AR9002U . ( The FCC ID on the dongle matches @  Sharp VR-WL25.  .

ps:
I am very curious how that happened. I mean what conditions and efforts allowed a company to accept to license openly a product's firmware.
Also i wonder it the firmware contain blobs.

I want to be able to write other languages (in addition to english).
I tried with no help ;
$ sudo dpkg-reconfigure locales
$ sudo dpkg-reconfigure console-setup
$ sudo dpkg-reconfigure keyboard-configuration

what should i do ?