derrassenc
@derrassenc.bsky.social
Ponent, Terraça, Camp, al-Maghrib, la perfide Albion. Flaques comptades. No necessàriament d'acord amb favs i RTs, però noblesa obliga 🐺 Tinc compte B.
Reposted by derrassenc
Bueno, amijos, como lo prometido es deuda, pues hoy os voy a contar la historia de la señora que empotró a tanta gente y se batió tantas veces en duelo que el rey tuvo que perdonarle la vida… dos veces. Hilo:
December 12, 2025 at 5:58 PM
Bueno, amijos, como lo prometido es deuda, pues hoy os voy a contar la historia de la señora que empotró a tanta gente y se batió tantas veces en duelo que el rey tuvo que perdonarle la vida… dos veces. Hilo:
Que la intel·ligència artificial és algo ideal per acompanyar la burrície natural dels que hi confien, doncs ja es repetia als anys 80 a les facultats de Matemàtiques.
Josep Maria, hi ha milions de llibres de cuina al món que valen la pena, i a la internet no feta malbé també un bon grapat.
Josep Maria, hi ha milions de llibres de cuina al món que valen la pena, i a la internet no feta malbé també un bon grapat.
December 14, 2025 at 8:51 AM
Que la intel·ligència artificial és algo ideal per acompanyar la burrície natural dels que hi confien, doncs ja es repetia als anys 80 a les facultats de Matemàtiques.
Josep Maria, hi ha milions de llibres de cuina al món que valen la pena, i a la internet no feta malbé també un bon grapat.
Josep Maria, hi ha milions de llibres de cuina al món que valen la pena, i a la internet no feta malbé també un bon grapat.
Reposted by derrassenc
Lmfao every AI defense piece is like "AYY I FUCKIN SUCK BRO HOPEFULLY THIS FUCKIN STEAL YOUR SHIT MACHINE WILL MAKE ME SUCK SLIGHTLY LESS OK"
December 14, 2025 at 2:41 AM
Lmfao every AI defense piece is like "AYY I FUCKIN SUCK BRO HOPEFULLY THIS FUCKIN STEAL YOUR SHIT MACHINE WILL MAKE ME SUCK SLIGHTLY LESS OK"
Reposted by derrassenc
When the AI boom began, copywriters were singled out as one of the jobs most vulnerable to AI. Now, three years later, I wanted to hear from workers on the frontlines of the industry, to hear what had actually taken place on the ground.
For many, it was even worse than they'd feared.
For many, it was even worse than they'd feared.
"I was forced to use AI until the day I was laid off." Copywriters reveal how AI has decimated their industry
Copywriters were one of the first to have their jobs targeted by AI firms. These are their stories, three years into the AI era.
www.bloodinthemachine.com
December 12, 2025 at 6:47 PM
When the AI boom began, copywriters were singled out as one of the jobs most vulnerable to AI. Now, three years later, I wanted to hear from workers on the frontlines of the industry, to hear what had actually taken place on the ground.
For many, it was even worse than they'd feared.
For many, it was even worse than they'd feared.
Reposted by derrassenc
The people swayed by nonsense tend to be lacking in humility and very sure they personally don't believe any nonsense. Knowing that we can also be idiots and make mistakes because we're human and all is intellectually protective. Rationality is itself an illusion on a neurobiological level.
December 13, 2025 at 5:16 AM
The people swayed by nonsense tend to be lacking in humility and very sure they personally don't believe any nonsense. Knowing that we can also be idiots and make mistakes because we're human and all is intellectually protective. Rationality is itself an illusion on a neurobiological level.
Reposted by derrassenc
For those who are interested, @prisonculture.bsky.social and I talk about political psychology and its role in organizing a bit in our book Let This Radicalize You.
Let This Radicalize You
Doing Justice: The Revolution of Rescue and Reciprocal Care is a practical and imaginative resource for activists and organizers building power in an era of destabilization and catastrophe.
www.haymarketbooks.org
December 13, 2025 at 3:49 AM
For those who are interested, @prisonculture.bsky.social and I talk about political psychology and its role in organizing a bit in our book Let This Radicalize You.
Avui el tren ha atropellat el senglar a la zona de Les Fonts, així que els fgc arriben puntuals a Rubí però ens ajuntem dos trens allí per arribar a Terrassa.
Arribo deu minuts més tard, veges. No m'imagino un incident així entre Reus i Tarragona, o Vilanova i Sitges.
Arribo deu minuts més tard, veges. No m'imagino un incident així entre Reus i Tarragona, o Vilanova i Sitges.
Que al ramal de Sabadell fa estona un senglar ha topat amb un tren dels ferrocarrils i els horaris es van ajustant, però servidora va a Terrassa i ja hem deixat Rubí enrere a plena puntualitat, i només queda el tram de Les Fonts per tenir la mala sort dels cocapitalins.
December 12, 2025 at 8:48 PM
Avui el tren ha atropellat el senglar a la zona de Les Fonts, així que els fgc arriben puntuals a Rubí però ens ajuntem dos trens allí per arribar a Terrassa.
Arribo deu minuts més tard, veges. No m'imagino un incident així entre Reus i Tarragona, o Vilanova i Sitges.
Arribo deu minuts més tard, veges. No m'imagino un incident així entre Reus i Tarragona, o Vilanova i Sitges.
Reposted by derrassenc
Nice one, Spidey.
December 12, 2025 at 11:36 AM
Nice one, Spidey.
Reposted by derrassenc
Interesting article on some of the recent excitement about LLMs for physics: superposer.substack.com/p/we-are-in-...
We are in the era of Science Slop
(and it's exciting)
superposer.substack.com
December 11, 2025 at 1:10 PM
Interesting article on some of the recent excitement about LLMs for physics: superposer.substack.com/p/we-are-in-...
Donant pel sac cada dia, oferint sense que li hagi demanat re briefings i key points de m que ni necessito ni obro, i quan demanes algo perquè no saps on para la icona del Teams al mòbil i dius, gastem una mica d'aigua, va i no ho pot fer.
Vaia cagada. Com a mínim el clip de Microsoft era graciós.
Vaia cagada. Com a mínim el clip de Microsoft era graciós.
December 12, 2025 at 5:00 AM
Donant pel sac cada dia, oferint sense que li hagi demanat re briefings i key points de m que ni necessito ni obro, i quan demanes algo perquè no saps on para la icona del Teams al mòbil i dius, gastem una mica d'aigua, va i no ho pot fer.
Vaia cagada. Com a mínim el clip de Microsoft era graciós.
Vaia cagada. Com a mínim el clip de Microsoft era graciós.
Santa Llúcia no és fins dissabte, però la logística va fer que ahir ja plantéssim el tió, abrigat amb la samarra d'emergència que fa hivern a la biblio.
Avui tindrà el platet de mandarines i fruits secs per passar el cap de setmana.
Avui tindrà el platet de mandarines i fruits secs per passar el cap de setmana.
December 12, 2025 at 4:48 AM
Santa Llúcia no és fins dissabte, però la logística va fer que ahir ja plantéssim el tió, abrigat amb la samarra d'emergència que fa hivern a la biblio.
Avui tindrà el platet de mandarines i fruits secs per passar el cap de setmana.
Avui tindrà el platet de mandarines i fruits secs per passar el cap de setmana.
Reposted by derrassenc
Zinaida Serebriakova (Ukrainian/Russian, 1884-1967)
Self-Portrait With a Brush, 1924
Oil on canvas
57 x 80 cm
Kyiv National Museum of Russian Art
#art #painting #artist #BlueSkyArt
Self-Portrait With a Brush, 1924
Oil on canvas
57 x 80 cm
Kyiv National Museum of Russian Art
#art #painting #artist #BlueSkyArt
December 11, 2025 at 8:19 AM
Zinaida Serebriakova (Ukrainian/Russian, 1884-1967)
Self-Portrait With a Brush, 1924
Oil on canvas
57 x 80 cm
Kyiv National Museum of Russian Art
#art #painting #artist #BlueSkyArt
Self-Portrait With a Brush, 1924
Oil on canvas
57 x 80 cm
Kyiv National Museum of Russian Art
#art #painting #artist #BlueSkyArt
Reposted by derrassenc
Demà, per cert. No sé si donaran rebujitos.
December 10, 2025 at 9:14 PM
Demà, per cert. No sé si donaran rebujitos.
Que al ramal de Sabadell fa estona un senglar ha topat amb un tren dels ferrocarrils i els horaris es van ajustant, però servidora va a Terrassa i ja hem deixat Rubí enrere a plena puntualitat, i només queda el tram de Les Fonts per tenir la mala sort dels cocapitalins.
December 10, 2025 at 8:41 PM
Que al ramal de Sabadell fa estona un senglar ha topat amb un tren dels ferrocarrils i els horaris es van ajustant, però servidora va a Terrassa i ja hem deixat Rubí enrere a plena puntualitat, i només queda el tram de Les Fonts per tenir la mala sort dels cocapitalins.
Em va agradar, no puc tornar-hi.
December 10, 2025 at 8:35 PM
Em va agradar, no puc tornar-hi.
No només la comercial. Empire of AI va a la pila de Reixos, també.
karendhao.com
karendhao.com
December 10, 2025 at 8:22 PM
No només la comercial. Empire of AI va a la pila de Reixos, també.
karendhao.com
karendhao.com
Reposted by derrassenc
In all of history, the book banners were never the good guys.
December 9, 2025 at 10:03 PM
In all of history, the book banners were never the good guys.
Reposted by derrassenc
Reposted by derrassenc
It’s fine if people find the term “hallucination” useful as an entry point to explain LLMs, I just think you really have to make the following clear: if anything an LLM produces is a hallucination, everything is.
Otherwise people will take away deep misconceptions about how LLMs work.
Otherwise people will take away deep misconceptions about how LLMs work.
December 9, 2025 at 9:32 AM
It’s fine if people find the term “hallucination” useful as an entry point to explain LLMs, I just think you really have to make the following clear: if anything an LLM produces is a hallucination, everything is.
Otherwise people will take away deep misconceptions about how LLMs work.
Otherwise people will take away deep misconceptions about how LLMs work.
Reposted by derrassenc
needing “human-in-the-loop processes” means needing actual thinking to improve the quality of the simulated thinking that we are told is on the brink of replacing actual thinking 🙃
December 9, 2025 at 1:33 PM
needing “human-in-the-loop processes” means needing actual thinking to improve the quality of the simulated thinking that we are told is on the brink of replacing actual thinking 🙃
Reposted by derrassenc
Excellent explainer thread. I was also reminded of this article: www.computerworld.com/article/4059...
OpenAI admits AI hallucinations are mathematically inevitable, not just engineering flaws
In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limi...
www.computerworld.com
December 9, 2025 at 6:55 AM
Excellent explainer thread. I was also reminded of this article: www.computerworld.com/article/4059...
Reposted by derrassenc
But they are inherently prone to hallucination. It's not a programming error.
You're right that they are trained to please, to give an answer instead of frustrating the user with "I don't know". They are likewise trained to be sycophantic, which means they flatter the user's own ideas.
6/end
You're right that they are trained to please, to give an answer instead of frustrating the user with "I don't know". They are likewise trained to be sycophantic, which means they flatter the user's own ideas.
6/end
December 9, 2025 at 6:45 AM
But they are inherently prone to hallucination. It's not a programming error.
You're right that they are trained to please, to give an answer instead of frustrating the user with "I don't know". They are likewise trained to be sycophantic, which means they flatter the user's own ideas.
6/end
You're right that they are trained to please, to give an answer instead of frustrating the user with "I don't know". They are likewise trained to be sycophantic, which means they flatter the user's own ideas.
6/end
Reposted by derrassenc
My colleague points out here that possibly one in seven requests to our library now may be assisted by LLMs.
When those point to fictional documents, that's a huge staff time drain. There's often no quick way to check if an undigitized primary source doc exists beyond physically pulling some boxes.
When those point to fictional documents, that's a huge staff time drain. There's often no quick way to check if an undigitized primary source doc exists beyond physically pulling some boxes.
December 9, 2025 at 2:06 AM
My colleague points out here that possibly one in seven requests to our library now may be assisted by LLMs.
When those point to fictional documents, that's a huge staff time drain. There's often no quick way to check if an undigitized primary source doc exists beyond physically pulling some boxes.
When those point to fictional documents, that's a huge staff time drain. There's often no quick way to check if an undigitized primary source doc exists beyond physically pulling some boxes.
Reposted by derrassenc
Folks might not know a lot of people are getting *very* upset that the materials they request are not available in the libraries/archives they're requesting from.
We're getting lots here at the Library of Virginia: requests of documents likely surfaced by large language models that *do not exist*.
We're getting lots here at the Library of Virginia: requests of documents likely surfaced by large language models that *do not exist*.
I wish a notice like this wasn't necessary.
www.icrc.org/en/article/i...
www.icrc.org/en/article/i...
Important notice: AI-generated archival references
www.icrc.org
December 8, 2025 at 2:07 PM
Folks might not know a lot of people are getting *very* upset that the materials they request are not available in the libraries/archives they're requesting from.
We're getting lots here at the Library of Virginia: requests of documents likely surfaced by large language models that *do not exist*.
We're getting lots here at the Library of Virginia: requests of documents likely surfaced by large language models that *do not exist*.
Reposted by derrassenc
Voleu llegir el què va publicar la Mariana Enriquez sobre Martin Parr. És del 2006.
www.pagina12.com.ar/diario/suple...
www.pagina12.com.ar/diario/suple...
Página/12 :: radar
Con más de treinta años de trayectoria y un número similar de libros publicados, el británico Martin Parr es uno de los más célebres fotógrafos documentales del mundo, miembro de la mítica agencia Mag...
www.pagina12.com.ar
December 8, 2025 at 9:13 AM
Voleu llegir el què va publicar la Mariana Enriquez sobre Martin Parr. És del 2006.
www.pagina12.com.ar/diario/suple...
www.pagina12.com.ar/diario/suple...