The Rise Of The Machines...

User avatar
pErvinalia
On the good stuff
Posts: 60693
Joined: Tue Feb 23, 2010 11:08 pm
About me: Spelling 'were' 'where'
Location: dystopia
Contact:

Re: The Rise Of The Machines...

Post by pErvinalia » Fri Jun 20, 2025 3:08 am

Screenshot_20250620-130947_Facebook.jpg
Sent from my penis using wankertalk.
"The Western world is fucking awesome because of mostly white men" - DaveDodo007.
"Socialized medicine is just exactly as morally defensible as gassing and cooking Jews" - Seth. Yes, he really did say that..
"Seth you are a boon to this community" - Cunt.
"I am seriously thinking of going on a spree killing" - Svartalf.

User avatar
JimC
The sentimental bloke
Posts: 74112
Joined: Thu Feb 26, 2009 7:58 am
About me: To be serious about gin requires years of dedicated research.
Location: Melbourne, Australia
Contact:

Re: The Rise Of The Machines...

Post by JimC » Fri Jun 20, 2025 4:49 am

A perfect example of a closed system trap, where incestuous combining of increasingly degraded information leads to system collapse. An analogy would be excessive interbreeding in a biological population...
Nurse, where the fuck's my cardigan?
And my gin!

User avatar
L'Emmerdeur
Posts: 6206
Joined: Wed Apr 06, 2011 11:04 pm
About me: Yuh wust nightmaya!
Contact:

Re: The Rise Of The Machines...

Post by L'Emmerdeur » Sat Jun 21, 2025 12:14 am

There's a piece in The Register about this. If model collapse is inescapable, then what? GIGO and AI text becomes completely unreliable. The fall of the machines?
"Our concern, and why we're raising this now, is that there's quite a degree of irreversibility. If you've completely contaminated all your datasets, all the data environments, and there'll be several of them, if they're completely contaminated, it's very hard to undo.

"Now, it's not clear to what extent model collapse will be a problem, but if it is a problem, and we've contaminated this data environment, cleaning is going to be prohibitively expensive, probably impossible."

[source]

User avatar
pErvinalia
On the good stuff
Posts: 60693
Joined: Tue Feb 23, 2010 11:08 pm
About me: Spelling 'were' 'where'
Location: dystopia
Contact:

Re: The Rise Of The Machines...

Post by pErvinalia » Sat Jun 21, 2025 4:44 am

They might have to start again. Train them on pre-2022 data, then tag all new AI data based on that model, and only that AI data can continue training AI. Or something like that.
Sent from my penis using wankertalk.
"The Western world is fucking awesome because of mostly white men" - DaveDodo007.
"Socialized medicine is just exactly as morally defensible as gassing and cooking Jews" - Seth. Yes, he really did say that..
"Seth you are a boon to this community" - Cunt.
"I am seriously thinking of going on a spree killing" - Svartalf.

User avatar
L'Emmerdeur
Posts: 6206
Joined: Wed Apr 06, 2011 11:04 pm
About me: Yuh wust nightmaya!
Contact:

Re: The Rise Of The Machines...

Post by L'Emmerdeur » Sat Jun 21, 2025 5:33 am

They'll have to solve the hallucination problem before they try that though or the cycle will just repeat. I'd think at some point you have to allow data-scraping again anyway.

User avatar
JimC
The sentimental bloke
Posts: 74112
Joined: Thu Feb 26, 2009 7:58 am
About me: To be serious about gin requires years of dedicated research.
Location: Melbourne, Australia
Contact:

Re: The Rise Of The Machines...

Post by JimC » Sat Jun 21, 2025 5:58 am

Hallucinating AI! Takes me back to the 70s... :{D
Nurse, where the fuck's my cardigan?
And my gin!

User avatar
pErvinalia
On the good stuff
Posts: 60693
Joined: Tue Feb 23, 2010 11:08 pm
About me: Spelling 'were' 'where'
Location: dystopia
Contact:

Re: The Rise Of The Machines...

Post by pErvinalia » Sat Jun 21, 2025 8:11 am

Next they'll be wearing cardigans!
Sent from my penis using wankertalk.
"The Western world is fucking awesome because of mostly white men" - DaveDodo007.
"Socialized medicine is just exactly as morally defensible as gassing and cooking Jews" - Seth. Yes, he really did say that..
"Seth you are a boon to this community" - Cunt.
"I am seriously thinking of going on a spree killing" - Svartalf.

User avatar
rainbow
Posts: 13752
Joined: Fri Jun 08, 2012 8:10 am
About me: Egal wie dicht du bist, Goethe war Dichter
Where ever you are, Goethe was a Poet.
Location: Africa
Contact:

Re: The Rise Of The Machines...

Post by rainbow » Sat Jun 21, 2025 1:14 pm

Nobody can tell me the point of AI. What is it?
I call bullshit - Alfred E Einstein
BArF−4

User avatar
macdoc
Twitcher
Posts: 8942
Joined: Tue Feb 23, 2010 3:20 pm
Location: BirdWing Home FNQ
Contact:

Re: The Rise Of The Machines...

Post by macdoc » Sat Jun 21, 2025 1:51 pm

AI Overview
The primary point of AI is to simulate human intelligence in machines, enabling them to perform tasks that typically require human-like cognitive abilities. This includes reasoning, learning, problem-solving, and decision-making. AI aims to enhance efficiency, automate processes, and provide insights across various fields, ultimately improving how we live and work.
and in certain things it far exceeds individual human capability ...seeing things, deriving patterns a human cannot.
Yeah its got problems but....no putting this genie back in the bottle. :pop:
Resident in Cairns Australia • Current ride> 2014 Honda CB500F • Travel photos https://500px.com/p/macdoc?view=galleries

User avatar
JimC
The sentimental bloke
Posts: 74112
Joined: Thu Feb 26, 2009 7:58 am
About me: To be serious about gin requires years of dedicated research.
Location: Melbourne, Australia
Contact:

Re: The Rise Of The Machines...

Post by JimC » Sat Jun 21, 2025 9:08 pm

https://www.abc.net.au/news/2025-06-21/ ... /105440772
Australian workers are facing a major upheaval as artificial intelligence (AI) becomes a cheaper alternative to employing humans.

While the full impact of AI is yet to be reflected in job ads or official employment statistics, both employers and employees warn the technology is already reshaping the nation's labour market.

It took recent data science graduate Tien Hung Nguyen 30 applications and an internship to land his first full-time job.

"I feel privileged to have secured this position. I'm going to give everything I've got," he says.

Most of his friends are still looking for work — and he says artificial intelligence is a big reason why.

"Since AI appeared, for example, a team might have needed three or four juniors and a senior. Now, it's one junior and AI," Mr Nguyen explains.

"There are fewer opportunities for young graduates."
Nurse, where the fuck's my cardigan?
And my gin!

User avatar
Tero
Just saying
Posts: 51149
Joined: Sun Jul 04, 2010 9:50 pm
About me: 15-32-25
Location: USA
Contact:

Re: The Rise Of The Machines...

Post by Tero » Sat Jun 21, 2025 11:39 pm

rainbow wrote:
Sat Jun 21, 2025 1:14 pm
Nobody can tell me the point of AI. What is it?
It will help you find the needle in a haystack.
International disaster, gonna be a blaster
Gonna rearrange our lives
International disaster, send for the master
Don't wait to see the white of his eyes
International disaster, international disaster
Price of silver droppin' so do yer Christmas shopping
Before you lose the chance to score (Pembroke)

User avatar
Brian Peacock
Tipping cows since 1946
Posts: 39875
Joined: Thu Mar 05, 2009 11:44 am
About me: Ablate me:
Location: Location: Location:
Contact:

Re: The Rise Of The Machines...

Post by Brian Peacock » Mon Jun 23, 2025 11:38 am

JimC wrote:
Fri Jun 20, 2025 4:49 am
A perfect example of a closed system trap, where incestuous combining of increasingly degraded information leads to system collapse. An analogy would be excessive interbreeding in a biological population...
...more like relying on your own excrement as a primary food source!
Rationalia relies on voluntary donations. There is no obligation of course, but if you value this place and want to see it continue please consider making a small donation towards the forum's running costs.
Details on how to do that can be found here.

.

"It isn't necessary to imagine the world ending in fire or ice.
There are two other possibilities: one is paperwork, and the other is nostalgia."

Frank Zappa

"This is how humanity ends; bickering over the irrelevant."
Clinton Huxley » 21 Jun 2012 » 14:10:36 GMT
.

User avatar
Brian Peacock
Tipping cows since 1946
Posts: 39875
Joined: Thu Mar 05, 2009 11:44 am
About me: Ablate me:
Location: Location: Location:
Contact:

Re: The Rise Of The Machines...

Post by Brian Peacock » Mon Jun 23, 2025 11:44 am

JimC wrote:
Sat Jun 21, 2025 9:08 pm
https://www.abc.net.au/news/2025-06-21/ ... /105440772
Australian workers are facing a major upheaval as artificial intelligence (AI) becomes a cheaper alternative to employing humans.

While the full impact of AI is yet to be reflected in job ads or official employment statistics, both employers and employees warn the technology is already reshaping the nation's labour market.

It took recent data science graduate Tien Hung Nguyen 30 applications and an internship to land his first full-time job.

"I feel privileged to have secured this position. I'm going to give everything I've got," he says.

Most of his friends are still looking for work — and he says artificial intelligence is a big reason why.

"Since AI appeared, for example, a team might have needed three or four juniors and a senior. Now, it's one junior and AI," Mr Nguyen explains.

"There are fewer opportunities for young graduates."
Seems a lot of the jobs were just nodes in an algorithm anyway. But AI isn't going to lift you in and out of the bath or shimmy your front door at 2am when you've locked the keys inside.
Rationalia relies on voluntary donations. There is no obligation of course, but if you value this place and want to see it continue please consider making a small donation towards the forum's running costs.
Details on how to do that can be found here.

.

"It isn't necessary to imagine the world ending in fire or ice.
There are two other possibilities: one is paperwork, and the other is nostalgia."

Frank Zappa

"This is how humanity ends; bickering over the irrelevant."
Clinton Huxley » 21 Jun 2012 » 14:10:36 GMT
.

User avatar
rainbow
Posts: 13752
Joined: Fri Jun 08, 2012 8:10 am
About me: Egal wie dicht du bist, Goethe war Dichter
Where ever you are, Goethe was a Poet.
Location: Africa
Contact:

Re: The Rise Of The Machines...

Post by rainbow » Tue Jun 24, 2025 2:50 am

Tero wrote:
Sat Jun 21, 2025 11:39 pm
rainbow wrote:
Sat Jun 21, 2025 1:14 pm
Nobody can tell me the point of AI. What is it?
It will help you find the needle in a haystack.
...or a noodle in a Hubei snack.

Image
I call bullshit - Alfred E Einstein
BArF−4

User avatar
L'Emmerdeur
Posts: 6206
Joined: Wed Apr 06, 2011 11:04 pm
About me: Yuh wust nightmaya!
Contact:

Re: The Rise Of The Machines...

Post by L'Emmerdeur » Fri Jul 04, 2025 5:12 pm

A description and name for a particular version of AI (LLM) failure-- 'potemkin understanding.'
Researchers from MIT, Harvard, and the University of Chicago have proposed the term "potemkin understanding" to describe a newly identified failure mode in large language models that ace conceptual benchmarks but lack the true grasp needed to apply those concepts in practice.

It comes from accounts of fake villages – Potemkin villages – constructed at the behest of Russian military leader Grigory Potemkin to impress Empress Catherine II.

The academics are differentiating "potemkins" from "hallucination," which is used to describe AI model errors or mispredictions. In fact, there's more to AI incompetence than factual mistakes; AI models lack the ability to understand concepts the way people do, a tendency suggested by the widely used disparaging epithet for large language models, "stochastic parrots."

...

Here's one example of "potemkin understanding" cited in the paper. Asked to explain the ABAB rhyming scheme, OpenAI's GPT-4o did so accurately, responding, "An ABAB scheme alternates rhymes: first and third lines rhyme, second and fourth rhyme."

Yet when asked to provide a blank word in a four-line poem using the ABAB rhyming scheme, the model responded with a word that didn't rhyme appropriately. In other words, the model correctly predicted the tokens to explain the ABAB rhyme scheme without the understanding it would have needed to reproduce it.

[source]
A preprint version of the paper is available.
Abstract:

Large language models (LLMs) are regularly evaluated using benchmark datasets. But what justifies making inferences about an LLM's capabilities based on its answers to a curated set of questions? This paper first introduces a formal framework to address this question. The key is to note that the benchmarks used to test LLMs -- such as AP exams -- are also those used to test people. However, this raises an implication: these benchmarks are only valid tests if LLMs misunderstand concepts in ways that mirror human misunderstandings. Otherwise, success on benchmarks only demonstrates potemkin understanding: the illusion of understanding driven by answers irreconcilable with how any human would interpret a concept. We present two procedures for quantifying the existence of potemkins: one using a specially designed benchmark in three domains, the other using a general procedure that provides a lower-bound on their prevalence. We find that potemkins are ubiquitous across models, tasks, and domains. We also find that these failures reflect not just incorrect understanding, but deeper internal incoherence in concept representations.

Post Reply

Who is online

Users browsing this forum: No registered users and 13 guests