link dump - catching up on my online reading
While I am slowly getting back up on my feet after a tough time, I am catching up on emails (will still take a bit!), my RSS feed reader, and several newsletters that have accumulated in my inbox.
Here's what I picked out to read and share:
This Is Capitalism: Apple's Hidden Data Workers at the Shadows of the AI Boom - 19 page paper detailing what data workers actually do and what problems they deal with, written anonymously by a data worker interviewing their colleagues. Parts of the work descriptions remind me of the work in Severance.
Is this really a good reason to triple datacentre capacity in Europe? - online blog post tracing where the idea to triple the EU's data centers comes from that is mentioned in several EU AI strategies. Turns out it's from be a blog post from Savills, a commercial real estate firm who profit off of data centers being built.
Fake US thinktank set up and funded by Israel sought to game AI for propaganda - AI slop meant to absolutely flood the web to be included in AI training funded by the Israel government to spread disinformation in AI answers.
A new force is increasing inequality in America - WaPo article about how AI is not leveling the playing field or bridging the gap between poor and rich, but instead worsening the gap. People making the most use of AI are concentrated in richer urban areas and are already often rather wealthy, while the AI data centers are in poorer neighborhoods and the data workers are often migrants or in the Global South. Rich people can invest into AI and its stock, therefore profiting off of the hype and concentrating even more wealth.
An operational framework for AI literacy in the workplace - 15 page paper from Interface EU addressing the vagueness of "sufficient AI literacy" that is often mentioned in EU AI legislation. It proposes a cumulative three-tier framework based on the nature and consequences of a worker's interaction with AI which then dictates the level of literacy required and how to attain/ensure it and measure it.
- Great quote from it: "No evidence yet shows that these trainings work, and three gaps might explain the reason. The first is motive. Corporate training aims at productivity and teaches people to use the tools well, whereas the law cares whether operators understand how systems fail and cause harm. A workforce fluent in prompting can still be illiterate in the sense a regulator means: trained to produce good output, but not to recognise when a model misleads or to know its duties under data-protection and risk rules."
The Quiet Erosion of Collective Action Under Digital Surveillance - article on chilling effects of permanent surveillance and the feeling of constant suspicion which continues to erode activism.
Flipping the kill switch: I survived 72 hours without US tech - online article about an experiment to live without reliance on US tech. The US has such a strong monopoly that almost all online services and tech are unusable with this rule.
Gone in one click - assessing the socio-economic impact of browser-level consent in Europe - small informative flyer style PDF by the Implement Group showing figures about cookie consent rates and ad industry revenues depending on the mode of consent.
Inside the growing vigilante movement to knock out Flock surveillance cameras - online article by The Guardian. I admire these people, and we need more civil disobedience now, everywhere. Not just against Flock; against Meta glasses wearers, against Ring camera owners (Yes, you too! None of you are the "good ones" with "valid" reasons!) and more. There's many ways to affect these devices that you can find online or just get creative with it. Keep yourself safe, don't write about it, don't record yourself doing it, don't discuss it via digital means, leave your devices at home, and leave no fingerprints.
Did someone wearing Meta Glasses film you today? Are you sure? - another Guardian online article, this time about the spy glasses and the people who enable hiding the recording light on them. The man behind GhostMeta is actually so vile and disgusting; anything else I could say would violate the Code of Conduct this blog is hosted on.
German links:
HeißeLuft.org - German website with interactive map showing where AI data centers are planned, in development, and paused, together with information on protests. Made me discover that they are planning on building one not too far away from me...
Deutsche Post trainiert ihre KI mit Ausweisfotos - article about how Deutsche Post is training their AI with ID pictures they get via digital identification procedures. It's not voluntary as they claim, as you need to give permission before being allowed to proceed.
Verhaltensscanner in Berlin: Harte Kritik an der KI-Überwachung - online article about the new camera installed in Berlin that will analyze all people in that area via AI surveillance software by Adesso, with more cameras to follow. They want to put them up in high crime rate areas, but keep secret what the standards for this are, which enables a mass roll-out of them if they wanted to without any oversight or control. There has already been one mix-up leading to higher crime reported in an area than actually happened. These cameras already also exist in Hamburg and Mannheim.
Möglicher AfD-Sieg in Sachsen-Anhalt - online article detailing the fear of queer people and people of color of an upcoming potential win of the AfD in their state.
Afd-Gutachten.de - website containing some stats and a PDF report of a legal assessment on the chances of a successful AfD ban.
„Gipfel gegen Linksextremismus“: Mit Trump gegen die Antifa - online article about the cooperation of Germany with the US on its fight against antifascism. Unfortunately it has continued, with the German government realigning to focus more on supposed "leftist extremism" and even re-distributing money away from leftist projects, which mostly hits projects aimed at helping queer people and migrants.
Wie weit ist Deutschland beim digitalen Gewaltschutz? - an online article about the really embarrassingly low standards of protection against digital violence, especially image-based ones like deepfake nudes and revenge porn, in Germany. Lots needs to be done in general, but especially to even meet the new EU standards.
Wer ist für Straftaten der KI verantwortlich? - legal article about the criminal liability of autonomous AI in Germany, and how crimes done by AI agents are pushing the legal system to its limit as we only legislate for humans.
Published