Martin Ruskov @mapto@qoto.org

If you're a journalist writing about this Azure outage, don't forget to ask Microsoft how much of their code is "AI-generated".

Satya Nadella claims 30%:
https://www.cnbc.com/2025/04/29/satya-nadella-says-as-much-as-30percent-of-microsoft-code-is-written-by-ai.html

#Outage #Azure #Microsoft

But an analysis of social media suggests something else. Many people and groups on the radical and far-right are harnessing a process known as audience capture in order to influence political policy.

A group of anonymous X accounts is said to follow a “posting-to-policy” strategy. These accounts – some of which are run by disaffected Westminster professionals – post to inject their grievances into online discourse.

To explore this dynamic, and how Reform’s recent u-turn has been shaped by it, we analysed the online networks that drove conversation about “mass deportations” on X over the past year. Using computational methods, we identified four distinct sub-communities defined by their retweet relationships. These sub-communities were formed around far-right influencers, radical right influencers, Advance UK/free-marketeer influencers – and around the Reform party.

Discussion of mass deportations in 2024 was almost exclusively dominated by the far-right and the anonymous accounts of the radical right. Fast forward to April 2025 and we find Lowe, Habib and a wider range of rightwing influencers have entered the conversation in support of the policy.

Finally, in September, following Reform’s August announcement, you can see Farage and key Reform personnel supplant the influencers as players in a movement they had little role in creating. In doing so, the party has aligned itself with a policy that less than a year ago it vehemently rejected.

A group of scientists set out to study quick learners. Then they discovered they don't exist | KQED https://www.kqed.org/mindshift/62750/a-group-of-scientists-set-out-to-study-quick-learners-then-they-discovered-they-dont-exist

As I'm screwing with a device trying to get wifi working, it occurs to me: Why are we still dutifully setting MAC addresses from flash rather than letting the ethernet card take a random one?

MAC addresses are the fucking Digital ID that everyone keeps railing against, the thing is written in the factory, it's logged in the DB, so it's connected to your identity when you buy the device, then it's being emitted out on any LAN you connect to.

Successful nonviolent movements for change require more than street protests. A human rights expert outlines 10 effective actions citizens can take: https://buff.ly/kpfD4ke

Total cost of today’s AWS outage will be in the billions of dollars. https://www.cnn.com/business/live-news/amazon-tech-outage-10-20-25-intl?post-id=cmgzdye9q00003b6ngdv69nyh

With the AWS outage, now‘s as good a time as any to repost this old strip

It got bright. #aurora #auroraborealis #northernlights #norrsken

Milano, la città dove la sosta selvaggia ha un suo spazio nei disegni tecnici

MM sta lavorando per completare la corsia preferenziale per la linea filoviaria circolare 90/91. Si tratta di una importantissima infrastruttura di collegamento urbano, che è
penalizzata da alcuni tratti in cui è ostacolata dal #traffico di auto.

Ebbene, nei disegni tecnici sullo stato attuale (da questo post), prima dei lavori, sono riportati gli spazi usati per la sosta irregolare. La #sostaSelvaggia è ufficiale, protocollata.

È un po' come se in un disegno tecnico dei giardini pubblici fosse riportato punto in cui si spaccia.

@milano

@heidilifeldman Canceled #Spotify a few weeks ago. Would cancel again if I could!

Ian Bogost asked Final Round AI to supply him with a final round job interview question

"It returned a lengthy, milquetoast answer... The entire thing was plausible in the way LLM responses often are; if an aspiring writer provided this response during a genuine interview, it wouldn’t be wrong so much as uninspired. It is the sound of a person performing the role of a job candidate, rather than one actually pursuing a job."
https://www.theatlantic.com/technology/2025/10/ai-cheating-job-interviews-fraud/684568/
@genai@lemmy.graphics

Quanti muoiono in scontri stradali che avvengono da soli, ovvero scontri autonomi senza coinvolgimento di un altro veicolo?

https://benzinazero.wordpress.com/2025/10/18/quanti-muoiono-in-scontri-stradali-che-avvengono-da-soli-ovvero-senza-coinvolgimento-di-un-altro-veicolo/ #incidente #incidentestradale #incidenti #sicurezza #mobilità #strade

@davidgerard , not sure if you saw this great post about ML models in radiology?

The boosters love to say ‘look how good AI is in medical diagnosis, therefore LLMs are good’. Only, it turns out (from the article):

while models beat humans on benchmarks, the standardized tests designed to measure AI performance, they struggle to replicate this performance in hospital conditions. Most tools can only diagnose abnormalities that are common in training data, and models often don’t work as well outside of their test conditions

It also highlights a problem that’s actually quite general in medicine: we have far more data about unhealthy people than healthy ones. I was talking to a cardiologist almost ten years ago who was very excited about the data things like the Apple Watch could collect. Apparently they know that a lot of people who have heart attack have arrhythmia, but they have no idea if this is a meaningful correlation. Healthy people tend to have their heart monitored for a minute or less on a visit to a doctor every few years. People with known heart problems wear heart monitors that can record a load of things, so you have very good data on their heart rhythms but no baseline to compare it against.

This is also true for radiology. You really want to do anomaly detection: take a few million scans of healthy people, wait a few years to see if any of them have undiagnosed conditions, and then use that dataset to train a model of what a healthy lung (or whatever) looks like. Then feed new scans to the model, have it flag anomalies, and loop in an expert to figure out what kind of anomaly it is and whether it’s important.

But what you have is a load of very examples of things that are wrong, in very specific ways. And these also have artefacts that are specific to individual devices, so it’s easy for a model to learn that people who are scanned with this class of machine have this condition.

And that’s just the start of the issues they discuss.

HOP Art

Looking for some good No Kings protest frog 'HOP' poster art, but concerned that most of the stuff floating around is AI slop?

Artist Edgar Wathert @eWalthert has produced this certified AI free beauty.

https://typo.social/@eWalthert/115373600558696862

@renewedresistance

#NoKings #OperationInflation #FrogArt #Frog #MastoArt

Greta Thunberg: “They kicked me every time the flag touched my face” https://www.aftonbladet.se/nyheter/a/25LgKq/greta-thunberg-they-kicked-me-every-time-the-flag-touched-my-face