Kristian Kersting

Excited to share that our paper "Synthesizing Visual Concepts as Vision-Language Programs" has been accepted to #CVPR2026! 🎉

We propose a novel method that combines VLMs with symbolic program synthesis to learn reliable programs of visual concepts.

🌐 ml-research.github.io/vision-langu...

Happy new year / frohes neues Jahr

Thanks 🙏

The AI and ML Lab wishes you a joyful holiday season and an insightful start to 2026

🚨 Just one week left to apply! W2 Tenure-track or W3 Tenured professor for Ethical and Safe AI. 🤖 🙏 Please share widely!

Check out the work here: ml-research.github.io/vision-langu...

Work together with my great co-authors @wolfstammer.bsky.social, Hikaru Shindo, Lukas Helff, @devendradhami.bsky.social , @kerstingaiml.bsky.social 💫

🚨 New paper alert!
We introduce Vision-Language Programs (VLP), a neuro-symbolic framework that combines the perceptual power of VLMs with program synthesis for robust visual reasoning.

Projekt SOOFI stärkt europäische KI-Souveränität Europa verfügt bislang über kein eigenes offenes Basis-Sprachmodell für anspruchsvolle industrielle KI-Anwendungen. Viele Unternehmen greifen deshalb auf Systeme aus den USA oder Asien zurück. Das neu...

www.tu-darmstadt.de/universitaet...
Im Rahmen von SOOFI („Sovereign Open Source Foundational Models for European Intelligence“) entsteht in Deutschland ein offenes KI-Grundlagenmodell für die industrielle Anwendung. Von der TUDa ist das Fachgebiet von @kerstingaiml.bsky.social beteiligt.

🚀 Join @tuda.bsky.social and @hessianai.bsky.social as Professor for Ethical & Safe AI to advance the algorithmic foundations of ethical and safe AI and to shape “Reasonable AI."

Apply now and make an impact where AI meets society.
👉 buff.ly/bIvq7Mb

We show applications to long-term outcomes and decision making, developing our works on 'causal parrots' openreview.net/pdf?id=tv46t... and meta-causality openreview.net/pdf?id=J9Vog... .

Thanks to my coauthors @philosotim.bsky.social, @devendradhami.bsky.social and @kerstingaiml.bsky.social! 2/4

As part of ELIZA, doctoral researchers join a vibrant AI community across Germany, supervised by leading scientists including @igurevych.bsky.social, @kerstingaiml.bsky.social, Jonas Peters & Mira Mezini.

(3/3)

#PhD #AIResearch #MachineLearning #NLP #ELIZA #UKPLab #TUDarmstadt

AI Generated Image of https://netzwerkrecherche.org/blog/algorithmen-wer-kontrolliert-die-neuen-machthaber/

10 (zehn !!1!) Jahre ist der NR-Talk von @spielkamp.bsky.social , @nettwerkerin.bsky.social, @kerstingaiml.bsky.social und mir her, um über Algorithmen & Kontrolle zu reden.

Es war einer der Impulse für @algorithmwatch.org, bei mir Anfang eines kleinen ;-) Umwegs zur KI. But: Finally there!

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅

EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu

eurips.cc

📢 Present your NeurIPS paper in Europe!

Join EurIPS 2025 + ELLIS UnConference in Copenhagen for in-person talks, posters, workshops and more. Registration opens soon; save the date:

📅 Dec 2–7, 2025
📍 Copenhagen 🇩🇰
🔗eurips.cc

#EurIPS
@euripsconf.bsky.social‬

Can concept-based models handle complex, object-rich images? We think so! Meet Object-Centric Concept Bottlenecks (OCB) — adding object-awareness to interpretable AI. Led by David Steinmann w/ @toniwuest.bsky.social & @kerstingaiml.bsky.social .
📄 arxiv.org/abs/2505.244...
#AI #XAI #NeSy #CBM #ML

@kerstingaiml.bsky.social (TU Darmstadt), Frauke Kreuter (LMU Munich, MCML, @zuseschoolrelai.bsky.social ), Mira Mezini (TU Darmstadt), @igurevych.bsky.social (TU Darmstadt, @athenecenter.bsky.social), [...]
(5/🧵)

@kerstingaiml.bsky.social ist Professor für KI und Maschinelles Lernen am @cs-tudarmstadt.bsky.social der @tuda.bsky.social und leitet dort das Artificial Intelligence and Machine Learning Lab (AIML).

🙏 for the invitation to give a talk at the Causal Neuro-Symbolic AI (Causal-NeSy) Workshop at ESWC. Integrating causality, neuro-symbolic reasoning, and machine learning to build more robust, interpretable, and generalizable AI systems is so important.

Diagnostic Reasoning in Natural Language: Computational Model and Application Diagnostic reasoning is a key component of expert work in many domains. It is a hard, time-consuming activity that requires expertise, and AI research has investigated the ways automated systems can s...

»Diagnostic Reasoning in Natural Language: Computational Model and Application« by @nilsdy.bsky.social, Matej Zečević, Ilia Kuznetsov, @kerstingaiml.bsky.social, @igurevych.bsky.social

📄 arxiv.org/abs/2409.05367
(8/🧵)

☺️ Glad that you are part of the team 💪🏻

Thanks ☺️🙏

Reasonable and safe: The future of artificial intelligence The researchers in the new Cluster of Excellence “Reasonable Artificial Intelligence” (RAI) , which is headed by TU Darmstadt, are carrying out research into a new generation of artificial…

🚀 TU Darmstadt leads the new Cluster of Excellence "Reasonable AI" – advancing trustworthy, efficient & adaptive AI grounded in common sense. A big thank you to the entire team, the university, and the state of Hesse for their tremendous work & support!

Multilingual Text-to-Image Generation Magnifies Gender Stereotypes and Prompt Engineering May Not Help You Text-to-image generation models have recently achieved astonishing results in image quality, flexibility, and text alignment, and are consequently employed in a fast-growing number of applications. Th...

👉 Multilingual Text-to-Image Generation Magnifies Gender Stereotypes and Prompt Engineering May Not Help You
Joint work with folks from TU Darmstadt, TU Munich, @kathaem.bsky.social, @jlibovicky.bsky.social, @mbrackaiml.bsky.social, @kerstingaiml.bsky.social
arxiv.org/abs/2401.16092

Eine letzte Chance Podcast-Folge · bto – der Ökonomie-Podcast von Dr. Daniel Stelter · 11.05.2025 · 1 Std. 24 Min.

Wer tiefer einsteigen will: Im Podcast „Eine letzte Chance“ spricht @thinkBTO über das Digitalministerium – mit Bezug zu unserem Whitepaper zur digitalen Staatlichkeit.
Hören lohnt sich! 🎧
👉 podcasts.apple.com/de/podcast/b...

Beyond the obvious: Digitalisierung „made in Europe“ ist die letzte Chance Dass es in den 1980er-Jahren gelungen ist, den Mobilfunk-Standard GSM als globalen Standard zu etablieren, lag am gemeinsamen europäischen Vorgehen und einer Entscheidung.

Ich freue mich, in der aktuellen Beyond the Obvious-Kolumne von @thinkBTO im @handelsblatt zum neuen Digitalministerium aufzutauchen. Digitalisierung braucht endlich Rückenwind – jetzt kommt es auf kluge Umsetzung und echten Gestaltungswillen an.
👉

Excited to share that our paper got accepted at #ICML2025!! 🎉

We challenge Vision-Language Models like OpenAI’s o1 with Bongard problems, classic visual reasoning challenges and uncover surprising shortcomings.

Check out the paper: arxiv.org/abs/2410.19546
& read more below 👇

Consider following the authors Indraneil Paul (@ineil77.bsky.social), Haoyi Yang, Goran Glavaš, Kristian Kersting (@kerstingaiml.bsky.social) & Iryna Gurevych (@igurevych.bsky.social) if you are interested in more information or an exchange of ideas. (7/7)

See you this week in Singapore! #ICLR2025

I'm excited to present our spotlight on meta-causal models at #ICLR2025 next week.

We model evolving causal graphs in dynamic systems. Applications to inference and attribution of agent actions. Paper: openreview.net/forum?id=J9V...

Visit our poster #441 during the Sat 3pm session.

ObscuraCoder: Powering Efficient Code LM Pre-Training Via... Language models (LMs) have become a staple of the code-writing toolbox. Their pre-training recipe has, however, remained stagnant over recent years, barring the occasional changes in data sourcing...

»ObscuraCoder: Powering Efficient Code LM Pre-Training Via Obfuscation Grounding« by Indraneil Paul (@ineil77.bsky.social), Haoyi Yang, Goran Glavaš, Kristian Kersting (@kerstingaiml.bsky.social) & Iryna Gurevych (@igurevych.bsky.social)

📄https://openreview.net/forum?id=VYvxrD7aS0 (3/🧵)

Exploring Neural Granger Causality with xLSTMs: Unveiling Temporal Dependencies in Complex Data Causality in time series can be difficult to determine, especially in the presence of non-linear dependencies. The concept of Granger causality helps analyze potential relationships between variables,...

xLSTM for time series with Granger causality: arxiv.org/abs/2502.09981

xLSTM again shows superb performance at time series analysis.

"Our experimental evaluations on three datasets demonstrate the overall efficacy of our proposed GC-xLSTM model."