Steffen Kühne

409 Followers
110 Following
12 Posts

Tech lead and interactive news developer at a public service broadcaster in Germany (BR).

Building the infrastructure and tools for the journalism of tomorrow.

#dataviz #maps #datascience #ai #automation #creativecoding #generativeart

Websitehttps://stekhn.de
Twitterhttps://twitter.com/stekhn
LinkedInhttps://www.linkedin.com/in/stekhn/
Githubhttps://github.com/stekhn
@janeggers ❤️

Hi fediverse 👋 ,

we are the AI + Automation Lab and part of Bayerischer Rundfunk (BR) and the ARD. The Lab is an interdisciplinary team that publishes stories and products at the crossroads of journalism, computer science and product development: https://www.br.de/extra/ai-automation-lab-english/ai-journalism-100.html

The idea of a decentralized social network and the German public broadcasting system are a natural fit. So we decided that it’s finally time to be part of the fediverse. We found our home on ard.social.

#introduction #ai #ddj

Interdisciplinary team at Bavarian Broadcasting (BR): AI + Automation Lab | BR.de

Since 2020, the AI + Automation Lab is working as a editorial office within Bavarian Broadcasting. The interdisciplinary team publishes stories and products at the crossroads of journalism, computer science and product development.

Hi liebes Fediverse 👋

wir sind jetzt endlich auch hier! Wir sind BR Data, wir machen Datenjournalismus beim Bayerischen Rundfunk. Unsere Recherchen sind daten- und dokumentengetrieben und meist investigativ.

Wir werden unsere Recherchen ab jetzt auch hier posten.

Wir arbeiten eng mit dem @br_ailab zusammen. Im Sommer haben wir gemeinsam ein Whitepaper dazu veröffentlicht, wie wir über KI und Algorithmen berichten: https://medium.com/br-next/blackbox-reporting-wie-journalist-innen-%C3%BCber-ki-und-algorithmen-berichten-k%C3%B6nnen-3983cd0bcf3c

#neuhier #introduction #ddj #journalism

Wie kann man kritisch über KI und Algorithmen berichten? Also das machen, was man Algorithmic Accountability Reporting nennt? Wir haben beim BR AI und Automation Lab die Ansätze der vergangenen Jahre strukturiert und mit vielen Beispielen zusammengefasst:

PDF: https://interaktiv.br.de/paper/AI-Automation-Lab_Blackbox-Reporting.pdf
Kurzfassung auf Medium: https://medium.com/br-next/blackbox-reporting-wie-journalist-innen-%C3%BCber-ki-und-algorithmen-berichten-k%C3%B6nnen-3983cd0bcf3c

#journalism #ddj

We at BR Data investigated the largest freely available training dataset for generative image models from #laion
The story with the relevant aspects privacy, copyright, consent: https://interaktiv.br.de/ki-trainingsdaten/en/index.html

A more technical thread 🧵 on how to tackle those huge quantities of data:

#ai #stablediffusion #trainingdata

@BR24

We Are All Raw Material for AI

Training data for artificial intelligence include enormous amounts of images and text gathered from millions of websites. An analysis performed of LAION datasets (Stable Diffusion) by public broadcaster BR shows that it frequently contains sensitive and private data – usually without the knowledge of those concerned.

BR

@bleeptrack @thierna Yeah, we are already using content automation, but only in areas where no creativity, contextualisation or critical thinking is required.

AI won't replace creative workers. It is a useful tool that allows us to automate tedious and repetitive tasks.

But I agree that the children AI analogy has to go away. At best, AI can mimic human intelligence and behavior.

I love everything about how the Youtube channel "Answer in Progess" researches and presents topics 🥰

The most recent episode is about why Japanese web design is so different from what we usually see on the web: https://www.youtube.com/watch?v=z6ep308goxQ

why Japan's internet is weirdly designed

YouTube

SUCH a cool experiment out of the Harvard Library Innovation Lab: https://archive.social

Allows you to capture a thread from Twitter and archive it in sealed PDFs to attest to legitimacy.

Could be enormously useful to journalists, archivists, etc, particularly with Twitter facing a perilous future.

Save Your Threads

High-fidelity capture of Twitter threads as sealed PDFs on social.perma.cc. An experiment of the Harvard Library Innovation Lab.

@Sonnenbrand Verstehe! Das erscheint mir sinnvoll und reicht mir vollkommen als Erklärung 🙏
@Sonnenbrand Danke für den Hinweis. Ich packe es auf meine ToDo-Liste. Gibt es irgendwo eine Erklärung, warum das sinnvoll ist?