AI fails

[swee'pea99]

Doesn't surprise. As to why,...

FADE IN: INT. THE PRESIDENT'S PRIVATE STUDY - DAY
Bartlet is talking with Dr. Stanley Keyworth.

BARTLET
There's a new international math and science study. We rank 19 out of 21
countries.

STANLEY
That's not very good.

BARTLET
This is what I'm talking about.

STANLEY
Who did worse?

BARTLET
South Africa and Cyprus.

STANLEY
See, that's questionable company 'cause you don't think of Cyprus as a leader
in the field.

BARTLET
No, you don't. Nor us anymore, for that matter. One-third of all math teachers
and half of science teachers didn't major or minor in that subject.

 

[swee'pea99]

Consider me a sceptic. (I know that you already do)
Last thing these people want is an intelligent electorate in a democracy.

I've been known to muse along similar lines on the criminalisation of recreational drugs...

 

[briantrumpet]

Just in case you think AI companies haven't completely lost their minds.

 

[briantrumpet]

https://arxiv.org/pdf/2604.04721

https://news.berkeley.edu/2026/10/0...rodes-your-ability-to-persist-at-hard-things/

 

[Psamathe]

Maybe this post is one for the AI Success thread but there are so many such threads and this one seems the main discussion.
At last. Somebody realising that building all these US Datacentres in the US using US money full of US hardware and running US software gives us no resilliance as the "Off switch" will be in the US (and from ICC experiences we now see how vindictive a rogue President can be).

So at least an AI expert calling for UK to develop its own versions of AI systems

UK must not be beholden to foreign AI, says head of Alan Turing Institute​

The UK must not become dependent on foreign AI systems that can be switched off at short notice and must develop its own versions of the technology, according to the head of the Alan Turing Institute.
...
“As AI becomes more embedded in critical services, in domains like defence, national security, critical infrastructure and healthcare we shouldn’t depend solely on systems we can’t inspect or control, or that could be restricted or even switched off,” said Williamson.

Yet UK Government seems determined to hand ever more power to questionable US "powers" eg switching NHS systems to Palantir without wondering what might happen should those in Palantir with more extreme political aims or seeking tax exemptions start to threaten withdrawing it on 3 hrs notice, or rogue President classifies elements critical under the Arms Export Control Act (AECA) and 10 mins later NHS loses access to its Federated Data Platform.

 

[AndyRM]

It still baffles me that anyone thinks something called "Palantir" is a good idea...

 

[briantrumpet]

You couldn't make it up. Well, Gemini does.

 

[briantrumpet]

A really interesting article on sniffing out AI-aided prize-winning literary works. (Dig those correctly used hyphens!)

https://www.theatlantic.com/books/2...t=Afjo8ZWiYsxozi9wkwT7E8a0pPg88r7sKZqWyJbWbZQ

When I read “The Blue Fighter,” I was struck by the proliferation of language that is common in AI-produced writing: the not x, but y sentence constructions; the glut of metaphors; the platitudinous aphorisms. “Anger burns out eventually,” the narrator’s mother says at one point. “Hunger doesn’t.” The earlier story by Kundizeza lacked those stylistic qualities, its prose and dialogue far more plainspoken and natural-sounding: “So how are things with my niece, huh?” the intelligence officer asks at one point.

What I also noticed, more disturbingly, was the presence of stereotypes about African life common among outsiders: a blundering, unemployed father; a resilient matriarch who dispenses clichéd approximations of wisdom; an environment of deprivation. Research consistently shows that AI products reproduce cultural tropes, including those having to do with the African continent.

 

[AndyRM]

A really interesting article on sniffing out AI-aided prize-winning literary works. (Dig those correctly used hyphens!)

https://www.theatlantic.com/books/2...t=Afjo8ZWiYsxozi9wkwT7E8a0pPg88r7sKZqWyJbWbZQ

Why do you keep using Gemini?

 

[briantrumpet]

Why do you keep using Gemini?

Have you read my replies to @PurplePenguin? I'm deeply sceptical on many fronts, and will criticise it for major reasons, but do you think that means I should not use any aspect of it whatsoever, because to do so to any extent would be being a traitor to all artistic effort?

 

[AndyRM]

Have you read my replies to @PurplePenguin? I'm deeply sceptical on many fronts, and will criticise it for major reasons, but do you think that means I should not use any aspect of it whatsoever, because to do so to any extent would be being a traitor to all artistic effort?

Some of them, but I don't hold them all in my head.

I wouldn't consider you an artistic traitor, but I don't think AI aids anyone's endeavours in any way.

 

[briantrumpet]

Some of them, but I don't hold them all in my head.

I wouldn't consider you an artistic traitor, but I don't think AI aids anyone's endeavours in any way.

OK.

My view - maybe deluded - is that using Gemini as a flawed Google-with-knobs on isn't a crime against humanity, rather a buggy way to get to sources I can evaluate for myself. And sometimes to get a summary that suggests I'm mistaken. (Believe it or not, I'd rather find out when I'm wrong.) Old simple Google is on its last legs, more's the pity, as I enjoy finding relevant terminology and using Boolean separators to get stuff I'm looking for.

My go-to search engine is Duck Duck Go, with AI turned off, but annoyingly, Boolean separators don't work on it (it seems to be a policy, which I don't understand at all).

I'm probably semi Luddite - they weren't anti progress per se (though I suspect there was an element of that), but deeply unhappy with how 'progress' was being imposed with no regard to the human cost (IIRC).

 

[Ian H]

I don't think Anne Carson uses AI to write.
If I were an early person
I'd look for the limits of human wisdom
by going to sacred oak trees
or the local blind man with lips on fire.
But this is now. This is NYC.
I go to Clive.
We meet in a diner
and queue for the breakfast special.
Clive's British.
He tries to make the large Hispanic short-order cook appreciate "underdone"
French toast. "My wife told me
not to say soggy," says Clive.
We pay. Currie shows up.
We sit and talk
of Clive's next trip to Guantanamo where,
although he's visited 56 times, they're questioning
(this time) his signature.
He laughs.
His current client, a Moroccan man,
has been cleared for release
and also informed
he will never leave.
[Extract from Clive Song]

 

[Psamathe]

... but do you think that means I should not use any aspect of it whatsoever, because to do so to any extent would be being a traitor to all artistic effort?

Maybe think of the climate damage you are doing. And it's not just impact per "question" but all the carbon created from the training it needed to do to answer your question.

 

[Pross]

Some of them, but I don't hold them all in my head.

I wouldn't consider you an artistic traitor, but I don't think AI aids anyone's endeavours in any way.

I ran my CV through CoPilot recently based on a job specification , asking it not to embellish or add anything and it actually did a really good job of making it more succinct and highlighting the relevant elements of my experience. I'm not a fan of AI and can see how it is a bad thing in many areas but it definitely has uses as a tool. Much like some of the software used in my industry that for decades has supposedly automated certain design processes, you need to have a good grasp of what you are asking it to do and be prepared to review the output thoroughly. It also seems to be that when people talk about AI they are generally referring to generative AI which to me seems to be the area that is the biggest risk to creative processes.