Colloquia and Seminars

While a significant portion of AI scientists and engineers believe we are on the verge of achieving highly general forms of AI, I offer a critical appraisal of this view through a causal lens. In particular, building on foundational developments in the field, I will present my perspective on the relationship between intelligence and causality, and the central role of the latter in building intelligent systems and advancing credible data science.

I frame this discussion in terms of five core capabilities that we should expect from an intelligent AI system: performing causal reasoning and articulating explanations; making precise, surgical, and sample-efficient decisions; generalizing across changing conditions and environments; generating and simulating in a causally consistent manner; and learning causal structures and variables.

In this talk, I will elaborate on this perspective and share current progress toward building causally intelligent AI systems. A more detailed discussion of this thesis is provided in my forthcoming textbook, a draft of which is available here: https://causalai-book.net/.

Bio: Elias Bareinboim is a professor in the Department of Computer Science at Columbia University and the director of the Causal Artificial Intelligence (CausalAI) Laboratory. His research develops causality as a foundation for artificial intelligence, with contributions to causal and counterfactual reasoning, data fusion, learning, generalizability, and decision-making, along with applications in biomedical and social domains. His honors include AAAI Fellow recognition, IEEE "AI's 10 to Watch," and young-investigator awards from NSF, DARPA, and ONR. Bareinboim serves as editor-in-chief of the Journal of Causal Inference and as an action editor of the Journal of Machine Learning Research.

Technion Host: Sarah Keren

Selected for an oral presentation at CVPR 2026; Image classification is a well-studied task in computer vision, and yet it remains challenging under high-uncertainty conditions, such as when input images are corrupted or training data are limited. Conventional classification approaches typically train models to directly predict class labels from input images, but this might lead to suboptimal performance in such scenarios. To address this issue, we propose Discrete Diffusion Classification Modeling (DiDiCM), a novel framework that leverages a diffusion-based procedure to model the posterior distribution of class labels conditioned on the input image. DiDiCM supports diffusion-based predictions either on class probabilities or on discrete class labels, providing flexibility in computation and memory trade-offs. We conduct a comprehensive empirical study demonstrating the superior performance of DiDiCM over standard classifiers, showing that a few diffusion iterations achieve higher classification accuracy on the ImageNet dataset compared to baselines, with accuracy gains increasing as the task becomes more challenging.

In the Degree Realization problem with respect to a family P of graphs the input is a non-increasing sequence d = (d1, . . . , dn) of positive integers, and the goal is to decide whether there exists a simple undirected graph G ∈ P, whose degrees correspond to d, i.e., such that deg(G) = d. In this paper we consider the version of Degree Realization in which the realization is required to be a forest (i.e., P is the family for forests). We consider optimized Degree Realization in which the goal is to obtain a realization that minimizes an objective function f. That is, the goal is to find a realization G that minimizes f(G) among the realizations of the given input sequence. More specifically, we focus on the following functions: the size of an optimal vertex cover and the size of an optimal dominating set. We also consider the total and paired versions of both Min Vertex Cover and Min Dominating Set. We provide characterizations and linear time realization algorithms for all the above-mentioned problems.