Show newer

= * * Data Collection and * * Labeling Techniques for Machine Learning

#Fediverse as training data... [ #AI ] 

" as training data... [ ]

"And as far as I can tell, nothing in their supplemental privacy policy for Threads would prevent them from using fediverse posts as training data for LLaMA."

- L. Rhodes /
@lrhodes

Source: Commodifying the fediverse
destructured.net/federated-mon

#Automation #Quote #AI #Industry = more bullshit jobs. 

"Automation was to help us achieve a 20 hour work week and more leisure time but now we have to work overtime to pay the bills.

The "paperless office" was supposed to streamline bureaucracy and cut red tape but now billions of us work bullshit jobs.

The " economy" promised and but it instead resulted in the precarious economy of sharecroppers.

AI, as an extension/retooling of the / hype train is of course no exception, except perhaps that the hype has become so blatant."

- Mark Shane Hayden quote
@msh / coales.co/@msh/112773418864965

Reasoning vs. Expectations... 

Logical reasoning often fails in situations where prior expectations of "goals" come in conflict with "reality" and "how people felt then and now"...

(Basically people didn't care how you felt then or now, and care only in the now about what did or didn't happen because of some expectation or unreasonable / unrealistic goal[s])

TLDR: Shit changes. And so should mentality of bosses!

Logical vs. ...

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

Title inspired by “Language models also perform imperfectly on logical reasoning tasks and more often fail in situations where humans fail—when stimuli become too abstract or conflict with prior expectations.”

"Language models, like humans, show content effects on reasoning tasks, PNAS Nexus, Volume 3, Issue 7, July 2024, pgae233, doi.org/10.1093/pnasnexus/pgae

Language models, like humans, show content effects on reasoning tasks

People digging digital holes.... Lack of practise with humans... 

People digging digital holes....

Holes are dug as people work when they don't do human interaction for anything / everything

If is repulsive to you, you can see it's not far to say the whole computing world from storage and beefier Google servers, more storage space everywhere for everything... and generally "normal" became not so normal (bloated / out of control)

becomes steroided 🍔 Hamburger fast-food consumption like McDonalds while people starve from attention <--- ok that's good

I start to think of these computer people as less human. AND ON TOP BUYING TONS OF American webspace like McDonalds... no no no... even if double ram for your server is $8 - don't click. Have a smaller rig run more audio / voice chats between people or a better ratio if you need steroids for computing.

It's like this equation for a diagram or comic...

the more they use or computers in general
AND... the less I see them responding on their replies
THEN YOU ARE ALMOST 💯 full-on computer person

SO = a bad practice of humanity / missing humanity (even if I imagine you do other things in life)

Your habit needs separation (from your computer to common communication and being part of other people's lives more directly even via voice / video).

This helps practise of human expression and emotions in more things and eventually fixes people (them or you) in everything which gradually the computers STUNT YOUR GROWTH 🌱 and you mind 🧠 unless you think your as good as a human as you are with computers... sorry make the uphill struggle with us! We too need practise! :thumbsup_hmn_h2:

Mechanise life and you take away humanity almost proportionately... 

As you mechanise you take away humanity and their activities which become replaced, taken over and eventually not needed.
Some is natural but we need to balance this back.

Because Machines at some stage render the need for humans obsolete, which is a terrible world for humans. And if there are must pay their way then not having a job is as close to a death (hunger) sentence as you can get.

Machines at some stage render the need for human's obsolete.

And the opportunities and environments are as important as habitat where what we do and touch, what we are around is important.

Mechanise life and you take away humanity almost proportionately or as if the two were ties together...

#FreeSchool is the best answer I have to working on "Real Humans"... 

*****Somehow***** to help improve THEM! ... getting somewhere personally is important as we've lost almost all other space for talking,,,
Like a in real life, serious coffee, theatre or solid protest place...

If you want to enjoy things beyond the "Virtual Computing" and "Artificial Layers" presented to us by Governments - then we all better finding more time to working on "Real Humans" *****somehow***** and to help improve THEM!

vs.

Training is not needed in many things that are actually worth keeping simple / human / even away from over-thinking / being too efficient etc as we live and enjoy things beyond AI and it's owners directives or rational.

"Weapons of Math Destruction" #book #Math #Banks #Cheating ? 

Indeed! A good title for a book...

"Weapons of Math Destruction" ➕ ➖ ✖️ ➗ :Pipe_right: 💲

PDF Download link appreciated...

========
My Note:
========
I've long since said the "science" of selling debt is working again man, but often that's what science is for (whether you like it or not - but can make the change by not using so often shops that sell money for debt! are a root cause of negativity! 🏦 Literally!... )


vs.
va vs.

"ChatGPT and other generative models are not magic - they rely on massive supply chains of human labour" ⛓️ 👥 👫 🚼 👫 👥 ⛓️

invidious.fdn.fr/watch?v=ug_p2

#AI #Colonialism - Same Shit different Tech 

(sorry it's getting to that stage it's RIGHT IN YOUR FUCKING FACE)

Not even angry... The text is :)

Take two, sorry let me try again...

(it's getting to that stage, right so it's...

RIGHT IN YOUR FUCKING FACEBOOK FACE !!!! :facebook: :meta: :data :petabytes ) invidious.fdn.fr/watch?v=ug_p2

#AI disposable workers: Remote human data labour like sweatshops underpaid = problems VIDEO 

The Invisible Workers Behind AI:
Exposing Underpaid Tech Labor

invidious.fdn.fr/watch?v=SWU7I

disposable work AND underpaid people VIDEO

“Meet the invisible workforce behind tech giants like Google, Facebook, Amazon, and Uber.

These underpaid and disposable workers label images, moderate content, and train the AI systems, only to be dumped later - often earning less than minimum wage (of course similar to 'sweatshops).

Their work is essential to be in the shadows (we don't see them but call it AI). Unacknowledged by companies they also have less rights abroad. Disposable at any time. not exactly reassuring


using to ABUSE labour or human

#AI #Notes "How Tech Companies Are Fooling Us..." 

TLDR; AI uses humans more than ever in the background to "show" / pretend about the tech in front and wow people into investing...

AI is the same - devaluing people even more by using remote people abroad in the background like you would find is (exactly India and similar countries)...

====================
Here's How Tech Companies Are Fooling Us... notes from video:

invidious.fdn.fr/watch?v=rX8MP

====================
and
====================

█ Humans are in the background checking things (tech is shown as the wow in front bells and whistles by Indians abroad ( ) and non-western country standards abroad doing laborious and questionable stuff for the "magic" to work.

█ Misleading claims or advert, like self-driving cars when it's someone somewhere else driving it for you.

█ "AI Washing" when companies exaggerate or even falsify claims of their AI technologies to attract investment or attract market advantage (people stupid enough to invest)

█ Hype trickles down to smaller (and more vulnerable) companies who are eager to rife the AI wave.

█ AI in CPU processors ?

█ Gartner Hype Cycle
en.wikipedia.org/wiki/Gartner_

█ More Hype than substance

█ More crashed cars involving AI

█ The notion of creating more jobs than it displaces might be overwhelming false

█ companies leveraging and restructuring whole companies in favour of AI

█ Those who use AI will replace those who don't - favouring AI trained people

█ Big estimates made for millions to change jobs by 2030

█ 12.07 Optimistic projection = hype of AI's "potential" and is almost purely to *pretend* to "create millions of jobs and boost GDP", when the reality is more complex and actually opposite

(since when does any company want to create more jobs more than a hot minute and then cut back?)

█⭐ Raising questions about who benefits and who is left behind. Emptying work itself using AI.

Comments from page:

█ "Humans are emotionally driven."

█ Its very easy when the prototype is wheeled out to get excited, the real time to start cheering is when its being manufactured and deployed

█ We shouldn’t even be excited about A.I.

The is "How to avoid AI being integrated when it's from the tops of food chains (employment, industry etc) ? ❓

Choice is also yours, not completely the industries...

AVOID AI / don't buy AI products.

██████████████
KEYWORDS LEARNED
██████████████

AI Washing
Remote Work
Contracted "Quality Control" (not on same soil)
Displaced Responsibility
Hype / Misleading
Human intensive / slave labour

AI Colonialism with ChatGPT - Generative models are not magic, they rely heavily on raw human underpaid labour...! 

Same-but-newer model.... AI Outsourced to places like

invidious.fdn.fr/watch?v=ug_p2

"problems"
- for example vacuum with cameras are checked by humans that are abroad in less controlled conditions. They "data label" objects or where to hoover, map things, toys and people in the way etc).

Cameras and AI see private and personal pictures of you and your home and are access by employees abroad who are NOT subject to the same rights and quality of privacy (also bad for them checking crude data) and doesn't ensure Western employment standards which is why they outsource in first place and can get away skipping privacy purposes and leveraging another culture to do it for them.

Company's outsource so they can leave it to contractors to cover this responsibility.

Employees have publicly shared vacuum pictures of people the toilet on Facebook :facebook: / Meta. It's so simple- the camera, the pictures/video and the connected internet+AI - all together as a privacy spaghetti 🍝

OTHER NOTES FROM VIDEO:

2.20 Some are calling this "AI Colonialism of global south"

Timnit Gebru @timnitGebru also was mentioned at this time of the video and is on Fediverse.

"LLMs, by contrast, work with probabilities, not binary correct/incorrect judgments. Based on a huge map of words created by processing huge amounts of text, LLMs decide which words would most likely follow from the words used in a prompt."

- Gerry McGovern @gerrymcgovern
mastodon.green/@gerrymcgovern/

"Pursuit of endless growth is pursuit of a cancer.

Moore's Law is dead and we have reached a happy plateau. We dont need more computing power. "

- Tinker ☀️ @tinker :thumbsup_hmn_h2:
infosec.exchange/@tinker/11265

Show older

FreeSchool=Co-Op+Friendship🔁's choices:

Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.