Tootfinder

Bref, l'intelligence artificielle n'a que celle que lui prête le miroir de notre idiotie. Quelle surprise.
Ben ouais, l'IA est toujours incapable d'entendement et d'interaction. Je dis encore une fois Š quel point il est atterrant qu'on résume l'intelligence au calcul ?
https://

Kashmir Hill (@kashhill@mastodon.social)
Attached: 1 image After writing about people going into delusional spirals with ChatGPT and having what look like mental breakdowns, I wanted to understand exactly how it happens. A corporate recruiter in Toronto who spent 3 weeks convinced by ChatGPT that he was essentially Tony Stark from Iron Man, agreed to share his transcript after breaking free of the delusion. We analyzed the transcript & shared it with experts. Now you can see the interactions & how delusional spirals happen: https…

I was trying to package #FlexiBLAS for #Gentoo, and to be honest, it doesn't look that good.
The first red flag is lack of an open bug tracker. Apparently, there is the tracker on GitLab that's limited to "members of their group and selected external contributors", but it doesn't seem to be used much. So it's "send us an email", and wonder how many people sent us the same bug report before.
The git repository is currently at something tagged 3.4.80 that seems to be prerelease, and its build system is quite broken. Not exactly the best path to verify that the bugs you are hitting are still there.
Now, upstream seems to insist on either using vendored netlib #LAPACK, or statically linking to the system library (we don't install the static libraries). Apparently I can specify the shared libraries instead, but it doesn't work — and it's unclear to me whether it doesn't work because I'm using the shared libraries, or because it doesn't support my LAPACK version. If I build LAPACK without deprecated symbols, it refuses to load it at runtime because of missing symbols. And if I build it with deprecated symbols, it fails to find some symbols at CMake time.
Honestly, I feel like I've spent too much time on this project already, especially given that its future is entirely unclear to me — the current git is quite broken, I have no clue how many issues were reported already and whether my bug reports will receive any reply. It definitely doesn't fare well for a package that we might start to rely heavily on. We don't want a cathedral there.
https://www.mpi-magdeburg.mpg.de/projects/flexiblas
https://gitlab.mpi-magdeburg.mpg.de/software/flexiblas-release

I randomly bought this book in a quirky bookshop in Copenhagen for the sole reason that it said all the wrong things right on the cover.
(Sales: the single most important profession. NLP™: not natural language processing but neuro-linguistic programming. Meta: the Meta Model™ and Meta Publications™.)
I just started reading it and boy oh boy, I was not disappointed. It's outrageously hilarious.
"Persuasion engineering".

Persuasion engineering by Richard Bandler | Open Library
Persuasion engineering by Richard Bandler, unknown edition,

Balancing Privacy and Efficiency: Music Information Retrieval via Additive Homomorphic Encryption
William Zerong Wang, Dongfang Zhao
https://arxiv.org/abs/2508.07044 https://

Balancing Privacy and Efficiency: Music Information Retrieval via Additive Homomorphic Encryption
In the era of generative AI, ensuring the privacy of music data presents unique challenges: unlike static artworks such as images, music data is inherently temporal and multimodal, and it is sampled, transformed, and remixed at an unprecedented scale. These characteristics make its core vector embeddings, i.e, the numerical representations of the music, highly susceptible to being learned, misused, or even stolen by models without accessing the original audio files. Traditional methods like cop…

I'd like to follow some trainings on SAML/OIDC, understand how these things work, how they should be setup. Any recomendations?
#sysadmin #oidc #saml

“But the insight was never lived. It was stored”
“I stopped wondering and started processing”
“Merlin Donald … argues that human intelligence emerged not from static memory storage but from external symbolic representation: tools like language, gesture, and writing that allowed us to rehearse, share, and restructure thought. Culture became a collective memory system - not to archive knowledge, but to keep it alive, replayed, and reworked”

Revisiting Gossip Protocols: A Vision for Emergent Coordination in Agentic Multi-Agent Systems
Mansura Habiba, Nafiul I. Khan
https://arxiv.org/abs/2508.01531 https://

Revisiting Gossip Protocols: A Vision for Emergent Coordination in Agentic Multi-Agent Systems
As agentic platforms scale, agents are evolving beyond static roles and fixed toolchains, creating a growing need for flexible, decentralized coordination. Today's structured communication protocols (e.g., direct agent-to-agent messaging) excel at reliability and task delegation, but they fall short in enabling emergent, swarm-like intelligence, where distributed agents continuously learn, adapt, and communicate to form collective cognition. This paper revisits gossip protocols, long valued in …