Really excited to present our #Neurips2023 paper “Do Not Marginalize Mechanisms, Rather Consolidate!” (arxiv.org/abs/2310.08377), tomorrow (Thu) during poster session 6, poster #1012. Feel free to stop by!

The unreliable "causal mind" of LLMs. We show that causal judgments of LLMs are biased in a very human-like fashion by examining two structures: Canonical Chain (A→B→C) & Common Cause (A←B→C). openreview.net/forum?id=544P6… Re-Align@ICLR2024 representational-alignment.github.io

Heading to #ICLR2024? Interested in causal reasoning? Check out our latest findings on a new causal bias (chains vs. common causes) in humans and LLMs. We show how model temperature affects these biases using earth mover's distance. Join the discussion at the Re-Align Workshop!

Great talk Antonia Wüst . If your interested in learning more about visual concepts via program synthesis check out our poster UAI 2025 tomorrow at 2 PM. paper: arxiv.org/pdf/2402.08280 #UAI2024

In prior work, we argued that LLMs recite causal knowledge embedded in the data. At #AIME2024, we now show how to combine this with structure learning and dedicated new data to induce medical models! Causal Parrots 👉 openreview.net/forum?id=tv46t… AIME 👉 link.springer.com/chapter/10.100…

Late announcement due to JMLR reviews and AC duties for NeurIPS. Florian Peter Busch will present our work on scaling causal inference to large scale real world data sets using probabilistic circuits at PGM 2024. Very excited for this line of research. Paper: raw.githubusercontent.com/mlresearch/v24…

Our Continual Causality Bridge is going into its 3rd edition at AAAI-25 AAAI ! 🎉 Join us for a spectacular 2 days of continual learning & causality by submitting your original 4 page work (for proceedings or non-archival tracks) until Nov 25 AOE! continualcausality.org/cfp/

The first “ConCon” challenge, tackling continual confounding, has just been launched!💡 Submit 3-page reports (non-archival) as part of the third Cont. Causal. Bridge at AAAI@2025. (Early deadline is Dec 6, 2024 AoE) Check out the full announcement here: continualcausality.org/challenge/

🔥More Continual Learning meets Causality 🔥 In addition to our call for papers (by Nov 25), our Continual Causality Bridge at AAAI-25 has now opened the first continual confounding challenge! Submit challenge solutions by Jan 7 on Kaggle! More info: continualcausality.org/challenge/

Don't miss this! Also remember, that we have an associated Continual Confounding Challenge running: kaggle.com/competitions/c… . Early challenge deadline is today AoE, but submissions will be open until 7. January. Selected papers will be given the chance to present at the bridge.

Three days left to apply for a PhD in my group! ⬇️⬇️ Come join us to work on the forefront of adaptive AI & lifelong machine learning. We’ve got lots of exciting ideas in store 🤗

Thrilled to share our #ICLR2025 work on Meta-Causal States! 🌟 Causal graphs evolve with dynamic systems & agent actions. We show how to cluster causal models by qualitative behavior, revealing hidden dynamics & emergent relationships 🚀 #Causality #ML arxiv.org/abs/2410.13054

🧠Foundation models are powerful—but what happens when they remember too much? Join us at #ICML2025 for our workshop on “The Impact of Memorization on Trustworthy Foundation Models” 👉icml2025memfm.github.io Let’s talk about memorization & what it takes to build trustworthy AI!

Hesse plans major cuts to university funding. This article, featuring my colleague Stefan Roth (TU Darmstadt), highlights the serious consequences. It risks widening the gap with states like Bavaria & Baden-Württemberg—and with leading high-tech nations: bit.ly/44prCWO

Not at #ICML2025 myself, but Florian & Roshni are presenting our spotlight poster on continual confounding + our respective ConCon dataset on Thursday 17 Jul 4:30 pm - 7 pm PDT in East Exhibition Hall A-B: E-3309 icml.cc/virtual/2025/p… Please talk to them! :) Details below ⬇️

There is some confusion among readers of #Bookofwhy regarding the impressive "causal understanding" LLM's, which seems to defy the theoretical prediction of the Ladder of Causation. The Ladder predicts that, regardless of data size, no learning machine could correctly answer

Judea Pearl Andrew Lampinen The meta concerns factors that lead to the establishment of causal edges. Think of the necessary condition of two physical objects being close to interact. Here, proximity is a meta-causal factor. In our robot example, robot A's decision to follow B establishes a causal link.

Can the new GPT-5 model finally solve Bongard Problems? 👉Not quite yet! Using our ICML Bongard in Wonderland setup, it solved 64/100 problems - the best score so far! 📈 However, some issues still persist ⬇️