Disappointed to discover that there is no “telephone box” emoji. A blue telephone box emoji would be even better, but any port in a storm. I’d recruit @Edent as consultant on Convincing The Unicode Consortium, but the power symbol is rather more justified…
@Edent ah, I wanted it as a Doctor Who reference but figured that a normal red phone box would be sufficient, but there isn’t one. Mostly because the Venn diagram of emoji plus people who remember red phone boxes is just about O O :-)
Fully switched to my Fairphone after @phpday 2024, and it feels like a good phone without Android bloatware (looking at you, Samsung, Google, etc.).
A bit big, but it runs as it should, and I feel better knowing that I'll be able to replace any part of it in very few minutes, if it breaks.
The fairbuds headphones feel amazing too: better than my old Sony WF-1000XM3, for which I had even custom silicone plugs made. Also there, the Sony ones failed due to batteries dying, while here I change 'em
I am really troubled by the proposal from the Tony Blair Institute for a "National Data Trust" (NDT) for health data, particularly by the idea that the next government might actually go for it.
What if, and hear me out here, the public simply don't agree that commercialising access to sensitive health data is a reasonable thing to do? What happens then?
Public engagement without power to make meaningful change over what matters is just participation washing.
➡️ Die Speicherpreise🔋 haben sich in den letzten 2 Jahren halbiert. Jetzt sind sie ohne Förderung wirtschaftlich.
➡️ Es wird Zeit, dass sie netzdienlicher werden und z.B. mit Tibber geladen werden und somit die Netze entlasten.
➡️ Im Sommer ist der Speicher immer voll, im Winter immer leer. Diese ungenutzte Kapazität muss für die Bereitstellung von Regelleistung genutzt werden.
Zwar habe ich eine anderes Modell (2Türme20kW), aber die Dinger haben LAN und ich kann nur über das PV-Tool sehen was die Batterien machen.
Demzufolge wäre ein Nichtgebrauch dieser Möglichkeit schon fast fahrlässig zu nennen.
Weil die Türme nach dem Winter asynchron liefen, musste der Handwerker meiner Installationsfirma die Hotline von BYD anrufen um Einblick zu erhalten...
BLOCAGE D'INTERNET EN CAS DE RÉVOLTE : L’IDÉE FAIT SON CHEMIN
La Kanaky et les banlieues comme laboratoires : chronique du fascisme qui vient -
Couper totalement internet pour priver la population d'un moyen de communication ? Une méthode réservée aux régimes dictatoriaux ou aux périodes de guerre ? En tout cas, c'est ce dont rêve le gouvernement français. Et ce qu'il est en train de se concrétiser.
Florida allowing clinics to perform C-sections outside of the hospital is a horrendous handout to private equity on the backs of women and babies.
For perspective, driving DOWN the C-section rate is literally a quality metric for hospitals. Widespread availability elective C-sections doesn’t improve outcomes, can cause future complications, and of course C-sections are high-risk, often frenetic surgeries. This should not be performed somewhere that doesn’t have NICU and ICU services, period.
Dear people of mastodon,
Does anyone know how I can protect my digital art from being stolen by ai? I've heard about things called Glaze or Nightshade but I have no idea how to use them or what they actually do. Any advice?
@BlackGlitterXx I haven't used them myself, but from domain knowledge:
How you use them: You use them like a regular filter or image processor. Most of the time they take an unprocessed image in and you get an obfuscated image out.
What they do: They subtly alter the pixel data of the image to break the patterns that AIs look for while training. In most cases, you can't visually tell the difference because the alteration is so subtle, but it causes current-gen AIs to absolutely choke when they try to generate an image mirroring the style that was obfuscated.
Note: while these tools do show significant results at this time, future AIs probably won't fall for their tricks and there are already ways to de-obfuscate them (but they're usually pretty lossy, so they're not super viable yet).
Note 2: Only images that are run through the processor are protected. If you post one obfuscated one to one site and one non-obfuscated one to another site, the AI will still likely be able to train off the latter.
@BlackGlitterXx Both tools can be found over here https://glaze.cs.uchicago.edu/ . Last time I checked they required a window computer and a Nvidia graphics card to speed the process but you can run them without it.
They have a little window with all the options so no command line skill required. They will produce a poisoned image based on your input that will prevent ai to extract the model from your work and dumb the algorithm down. Keep in mind that IT doesn't prevent ai copy, just make it very dif
I'm in Oxford! Sandstone buildings, bars, and bookshops!
I did this "self-guided tour for new and prospective students", and enjoyed it a lot! It's very well-produced, and playing audio based on my location worked quite well. https://www.ox.ac.uk/students/new/knowyouroxford