Künstliche Intelligenz & Desinformation im #Superwahljahr ? Eine reale Bedrohung aber (noch) kein Grund zur Panik.
Ein ausgezeichneter #Doublecheck von Nadja Hahn mit Wortspenden von Ingrid Brodnig, Felix M. Simon, Flo 🅿. Schmidt & meiner Wenigkeit.

oe1.orf.at/artikel/710219…

I don’t have the chops to assess this pre-print but if their findings prove to be correct and are borne out by future work it seems like we need to correct current beliefs that foundation models trained on data increasingly laced with their own output are inherently doomed

Today’s the day to vote!

Pouring some cold water on the 'AI-generated search answers will be too expensive and therefore not happen' argument

📢PUBLICATION ALERT🧵
Our new research in Journal of Quantitative Description: Digital Media maps the global landscape of political microtargeting across 95 countries during 113 elections. Simon Kruschinski▲@meinungsfuehrer.bsky.social, Mads Hove, Tom Dobber, NHelberger is @[email protected], Claes de Vreese, and I analyze 2.5M FB & IG ads that ran from 2020 to 2022.

Online political targeting is widespread, but contra recurring fear-mongering of 'sophisticated' algorithmic-based micro targeting - remember Cambridge Analytica? - , is usually based on a single (and simple) targeting criterion, such as age or sex or location

Who needs deepfakes when you have animation, narratives developed+reinforced for years in multiple formats by prominent public figures facing often pliant news media, and a highly motivated partisan community happy to push them on every platform available? thewire.in/communalism/co…