Tootfinder

«KI-Modelle sind anfällig für wiederholte Angriffe:
Laut Forschern von Cisco versagen KI-Modelle bei realistischen Multi-Turn-Angriffen und lassen an Sicherheits-Benchmarks auf Basis weniger Prompts zweifeln.»
Der moderne Widerspruch ist die KI oder was ist es sonnst? So klug wie KI angeboten wird ist es einfach nicht.
🤖

KI-Modelle sind anfällig für wiederholte Angriffe
Laut Forschern von Cisco versagen KI-Modelle bei realistischen Multi-Turn-Angriffen und lassen an Sicherheits-Benchmarks auf Basis weniger Prompts zweifeln.

from my link log —
42 is an answer to the question, what is the sum of three cubes?
https://aperiodical.com/2019/09/42-is-the-answer-to-the-question-what-is-80538738812075974³-8043575…

42 is the answer to the question “what is (-80538738812075974)³ + 80435758145817515³ + 12602123297335631³?”
We now know that the number 42 can be written as the sum of three cubes: \[ 42 = (-80538738812075974)^3 + 80435758145817515^3 + 12602123297335631^3 \] This computational breakthrough was achieved i…

The whole #LLM ROI thing reveals something interesting. It's basically impossible to figure out the ROI of an LLM. That makes it impossible for bean counters to make a comparison between human work and LLM work, or human work without an LLM and LLM-assisted work, to determine if the incredibly high price is worth it. But it's also impossible because you can't measure the ROI of a human, especially for skilled labor.
You can't measure the ROI of a human, because managers have no idea what people do. There's an eternally expanding amount of work designed to address this problem. But no matter how closely people are surveilled, interrogated, analyzed, there's never any real answer.
I've talked in the past about in relation to medical care. One of the dirty secrets of hospitals is that they have no way to figure out how much individual treatment costs. It's easy to understand at scale. You can know exactly how much something costs society. You can even identify patterns, using public health models, and decrease costs for society by trying to get people to avoid risky behavior (stop smoking, use protection during sex, etc). But it is absolutely not possible, at all, in any way, to figure out how much a single visit costs. This is similar to the problem of predicting climate change vs predicting the weather tomorrow in Amsterdam at 15:00. One is possible, the other is simply not.
But what is becoming painfully clear now is that this is true *everywhere*. It's trivial to know how much an industry costs. It's possible to figure out it's ROI for society. It is not possible to figure out how much value any individual worker provides. LLM ROI and cost comparison is an instance of this larger problem.
This is a problem for capitalism because it shows that the fundamental assumptions behind capitalism, that product value and labor value are quantifiable, that people can actually make comparisons between competing products, etc, are completely bullshit. The capitalist apologetics that makes up so much of economics, the lies that are told that hold this system together, are crumbling before our eyes.
If you make a lot of money, it's because you've been lucky. You have the right social networks, you have become good at convincing people to give you money. There is absolutely no way to connect that to actually providing value to society. If you make a lot of money, internalize that. Understand that you are not special, and things can change. If you don't make a lot of money, it's not because you don't provide value. Don't forget that. The system is a lie built to destroy you. Don't let it.
The ideology is sick, something something time of monsters and all that, we are together in this dying machine. We need to understand the lies. Your value can never be quantified. The way we have always figured out how to do the right thing for each other is through each other. Social connection has always guided us. But now the most socially disconnected people on the planet have hijacked the system. They direct the resources of the world, and game the system to avoid personal responsibility.
We have to build a system where everyone is accountable. We can't use abstract numbers and lies to figure things out for us. We have to build systems around people and accountability. There is no other solution.

The Computer Science Fetish https://mail.cyberneticforests.com/the-computer-science-fetish/

The Computer Science Fetish
On the Valorization of Technical Authority 💡Nerd Alert: 1/5. Readable, but also mostly venting about weird encounters at a conference in San Francisco. The academic AI critic is often either bored or angry. Bored because everyone keeps saying the same things, or angry because all the same things need

Hop on interdit l'argent dans le sport ...
Règlements de comptes entre ultras du PSG : de Munich Š l’A11, l’histoire secrète d’une bagarre en trois rounds - Le Parisien
https://www.

Règlements de comptes entre ultras du PSG : de Munich à l’A11, l’histoire secrète d’une bagarre en trois rounds
Samedi, un groupe de supporters ultras du PSG en a attaqué un autre lors d’un déplacement vers Angers. Un règlement de comptes qui trouve son origine à la dernière finale de Ligue des champions en Allemagne.

Anker announces Thus, a compute-in-memory chip it says will bring on-device AI to its products and accessories, starting with its upcoming Soundcore earbuds (John Higgins/The Verge)
https://www.theverge.com/tech/916463/anker-thus-chip-announcement

Anker made its own chip to bring AI to all its products
The Thus chip will come to earbuds first before being rolled out to other Anker products.

Training data generation for context-dependent rubric-based short answer grading
Pavel \v{S}indel\'a\v{r}, D\'avid Slivka, Christopher Bouma, Filip Pr\'a\v{s}il, Ond\v{r}ej Bojar
https://arxiv.org/abs/2603.28537 https://arxiv.org/pdf/2603.28537 https://arxiv.org/html/2603.28537
arXiv:2603.28537v1 Announce Type: new
Abstract: Every 4 years, the PISA test is administered by the OECD to test the knowledge of teenage students worldwide and allow for comparisons of educational systems. However, having to avoid language differences and annotator bias makes the grading of student answers challenging. For these reasons, it would be interesting to compare methods of automatic student answer grading. To train some of these methods, which require machine learning, or to compute parameters or select hyperparameters for those that do not, a large amount of domain-specific data is needed. In this work, we explore a small number of methods for creating a large-scale training dataset using only a relatively small confidential dataset as a reference, leveraging a set of very simple derived text formats to preserve confidentiality. Using these methods, we successfully created three surrogate datasets that are, at the very least, superficially more similar to the reference dataset than purely the result of prompt-based generation. Early experiments suggest one of these approaches might also lead to improved model training.
toXiv_bot_toot

2026's Costco-value late-round NFL draft picks, plus examining the 49ers' confidence https://www.nytimes.com/athletic/7240037/2026/04/29/2026-nfl-draft-late-round-values-49ers-reach-consensus/

Has Technology Ever Reduced Labor?
Has technology ever reduced labor? The question sounds rhetorical. We carry small computers that answer any factual query in seconds, our laundry tumbles itself clean while we sleep, our cars drive themselves on highways our great-grandparents traveled by mule. Of course technology has reduced labor. The question barely needs asking.

Has Technology Ever Reduced Labor? | David Boles, Blogs
Has technology ever reduced labor? The question sounds rhetorical. We carry small computers that answer any factual query in seconds, our laundry tumbles itself clean while we sleep, our cars drive…

A look at Politico's Dasha Burns, soon to be its Global Anchor in addition to other journalistic roles, as the outlet molds her into a creator-like figure (Corbin Bolies/The Wrap)
https://www.thewrap.com/media-platforms/journalism/dasha-burns-politico-i…

Dasha Burns Brings a Creator Mindset to Politico: ‘On Your TV, In Your Ear, In Your Inbox’
The Playbook co-author on becoming Politico’s first global anchor, covering Trump’s combative second term — and why she welcomes his appearance at Saturday’s White House Correspondents’ Dinner.