iiinova — where your digital self keeps evolving

Stefan Szeider

@szeider.bsky.social

Professor & Chair at TU Wien, leading the Algorithms and Complexity Group. Specializing in symbolic and neurosymbolic methods to address complex computational problems in AI, automated reasoning, and combinatorial optimization. (ac.tuwien.ac.at)

Posts Replies Media

Stefan Szeider @szeider.bsky.social · 13/09/2026

It was cool to be invited to present my AI installation, Spot On, at the Ars Electronica Festival 2026 and to see the reactions of over a thousand visitors who had the courage to try it. Many thanks to the Ars Electronica team for all their support. ars.electronica.art

camera, screen, and speakers of the Spot On installation

Someone sitting on the stool and let the AI analyze

from the festival cataloge: Spot On poses the unsettling question: What does it feel like when a machine claims to know you? The work transforms AI’s promise of understanding us into an intimate, disquieting encounter. A monitor displays visitors’ own reflections and invites them to pose. As soon as someone takes a seat, a calm voice from the speaker beside the camera delivers a psychological interpretation. A multimodal AI model generates these assessments live from the camera image and speaks them aloud with unwavering confidence. It performs the very authority we often attribute to AI systems that claim to know the truth. The machine is authoritative, intimate—and often wrong.
Spot On confronts visitors with their desire to be seen and truly understood, while exposing the problem of confabulation in AI systems. The installation draws attention to the power of technological systems not only to describe reality but to actively produce it. By interpreting identity, emotions and intentions with apparent certainty, the work questions who has the authority to define people—and what relations of power are inscribed within these technological interpretations.

Stefan Szeider in front of the Ars Electronica banner, wearing an "Artist" festival pass

030

Stefan Szeider @szeider.bsky.social · 18/07/2026

I‘m on my way to FLoC in Lisbon and look forward to giving quite a few talks… #FLoC2026

Invited talk at WHOOPS: From Verdicts to Theorems: Importing LRAT and VeriPB Certificates into Lean 4
Invited talk at CompBench: Benchmarking for and with Language Models
Paper presentations at
PoS: PBLean: Pseudo-Boolean Proof Certificates for Lean 4 and Reverse-Engineering Constraint Models from CNFs: An Agentic Approach
ML4SP: Algorithm Selection with Zero Domain Knowledge via Text Embeddings
CP: VIPR Certificate Construction from Black-Box ILP Solvers
ICLP: ASP-Bench: From Natural Language to Logic Programs
My students will present
CP Smart Cubing for Graph Search: A Comparative Study
ICLP Streamliners for Answer Set Programming
I will be a panelist at CP 2026 Panel on Thirty Years of Progress Towards the Holy Grail.

040

Stefan Szeider @szeider.bsky.social · 13/03/2026

Many thanks to the Digital Humanism Initiative for inviting me to give a public talk on "LLMs When Left Alone." Among other things, I present a Placebo experiment I conducted on AI agents. See the recording on YouTube youtu.be/qY1aaX-I1LU #AI #AGI #AIethics #AIsafety #AIconsciousness

youtu.be

Stefan Szeider: “LLMs When Left Alone”

YouTube video by Digital Humanism

010

Stefan Szeider @szeider.bsky.social · 26/02/2025

I'm looking forward to todays' BRIDGE program @ #AAAI25, in particular to the CP+ML program (ucc.ie/CPML2025/schedule.html) and the ML for solvers program (ml-for-solvers.github.io) I will be a panelist in two panels: CP+ML panel (4pm) LLMs for logical reasoning panel (5pm) will be fun!

020

Stefan Szeider @szeider.bsky.social · 07/01/2025

Join us on January 14 for the 2025 Vienna World Logic Day Lecture with Joao Marques-Silva, who speaks about Trustable Explainable AI - A Redemption by Logic Register: logicday.vcla.at/vienna-logic... #XAI #LogicInAI #WorldLogicDay2025

131

Stefan Szeider @szeider.bsky.social · 18/12/2024

Computer Science as the continuation of Logic with other means. Georg Gottlob gives a talk at TU Wien on the occasion of the Helmut Veith Stipend Award.

020

Stefan Szeider @szeider.bsky.social · 05/12/2024

BILAI Cluster of Excellence retreat at Otto Wagner's Postsparkasse. Art Nouveau meets AI research. @BilateralAI

010

Stefan Szeider @szeider.bsky.social · 28/11/2024

Sepp Hochreiter gives an overview of what Subymbolic AI Can and Cannot Do (Paradigm Shift in Computer Science workshop at TU Wien informatics.tuwien.ac.at/stories/2745/)

Sepp Hochreiter at the Paradigm Shift in Computer Science workshop @ TU Wien

020

Stefan Szeider @szeider.bsky.social · 28/11/2024

Thomas Eiter gives an overview of Symbolic AI – What It Can and Cannot Do, at the workshop A Paradigm Shift in Computer Science? at TU Wien. There is a livestream. informatics.tuwien.ac.at/stories/2745/

Thomas Eiter at workshop on a Paradigm Shift in Computer Science.

031

Stefan Szeider @szeider.bsky.social · 20/11/2024

Somewhat related to this topic are Knuth's essays Computer Science and its Relation to Mathematics (1973), and Algorithmic Thinking and Mathematical Thinking (1985) - they are pretty old but still an interesting read doi.org/10.2307/2318... doi.org/10.2307/2322...

doi.org

Algorithmic Thinking and Mathematical Thinking

Published in The American Mathematical Monthly (Vol. 92, No. 3, 1985)

043