Mardi 3 juin, découvrez comment l’évaluation des modèles IA évolue et soulève un enjeu politique majeur.

Une présentation d’ @annaschjoett.bsky.social et @dluitse.bsky.social suivie d’un échange sur le lancement de la collection « The Politics of ML Evaluation » dans Digital Society.

Plus d'info 🔽

The introduction to our SI on bridging the research-practice gap in journalism has now been published!

You can read here: journals.sagepub.com/doi/10.1177/...

It was a pleasure to co-edit this special issue with @annaschjoett.bsky.social, @nadjaschaetz.bsky.social, and @claudiamellado.bsky.social

✨ Very happy to share that my new paper on data annotation practices in affective computing is out ✨ link.springer.com/article/10.1...

🙏 thanks to the fantastic editors of this SI on the politics of machine learning evaluation @dluitse.bsky.social @annaschjoett.bsky.social & Tobias Blanke

The second paper, 'An Accidental Benchmark: The History, Contingent Power, and Lasting Traces of the GTZAN Dataset' by Allison Jerzak traces the history of the GTZAN dataset, which remains the most widely used dataset for Music Genre Recognition. link.springer.com/article/10.1...

The first paper is a short reflection by @alexcampolo.bsky.social titled 'State-of-the-Art: The Temporal Order of Benchmarking Culture', which explores how benchmarks produce certain temporal values and expectations link.springer.com/article/10.1...

✨The first two papers in our Topical Collection on the 'Politics of Machine Learning Evaluation' are out in Digital Society!

The collection is edited together with @dluitse.bsky.social and Tobias Blanke, with the great support of Digital Society and Federica Russo.
link.springer.com/collections/...

I have a short reflection on benchmarking in machine learning, titled "State-of-the-Art." Beyond questions of construct validity and rankings, how do benchmarks and model evaluation produce orientations towards progress or the future? Open access: doi.org/10.1007/s442...

This came out of an excellent conference, organized by @dluitse.bsky.social and @annaschjoett.bsky.social of the University of Amsterdam titled - "The Politics of Machine Learning Evaluation." Look out for the other contributions to this forum, some of which are forthcoming. Super important topic.

The seminar series is organised together with @dluitse.bsky.social and Tobias Blanke. More information can be found here: www.create.humanities.uva.nl/seminar-seri...

On May 14, we are delighted to host @whkchun.bsky.social and Matt Canute as the final speakers in our Critical AI Seminar Series.

The seminar is online and open to everyone. To receive the link, sign up via the link below.

🗓️ May 14, 5:30-7 PM (CEST). www.create.humanities.uva.nl/events/5-cri...

✨Interested in questions of synthetic vision and algorithmic violence?

Then join the online #CriticalAI Seminar series next Wednesday (March 12) from 5:30-7PM with @ikkibop.bsky.social and Francesco Ragazzi.

Organised with @dluitse.bsky.social and Tobias Blanke. Open to everyone, sign up below: