Institute for AI

The aim of the Institute for Artificial Intelligence at the University of Stuttgart (@Uni_Stuttgart) is to research fundamental questions about AI, to reflect on the benefits for society, and to promote the transfer of AI applications to business and society.

Institute for AI boosted:
2025-06-02

🤝 In collaboration with the (@UniStuttgartAI) Institute for Artificial Intelligence at the Universität Stuttgart, it is our great pleasure to highlight the following event:
🎤 Engineering Safe Systems with AI
🗓️ June 5, 2025 | 15:45 | Room U32.101, Universitätsstraße 32
We’re pleased to support this talk by Dr. Reinhard Stolle Deputy Director at Fraunhofer IKS, on how to engineer safe AI-enabled systems without compromising innovation.
In his talk, “Engineering Safe Systems with AI”, Dr. Stolle will explore two key perspectives on safety: a safety-centricand an AI-centric view. He will present his team’s approach to combining the strengths of both, introducing a model for continuous safety engineering for high-risk AI systems—explicitly modeling and propagating uncertainties and confidences during both design and operation.
📣 Students, staff, and all interested guests are warmly invited to attend this exciting and insightful session!

👤 About the Speaker
Dr. Reinhard Stolle is Deputy Director of Fraunhofer IKS and Head of the Mobility Business Unit. He studied computer science at FAU Erlangen and the University of Colorado at Boulder, earned his master’s and Ph.D. in AI, and completed postdoctoral research at Stanford. His career spans AI research at Xerox PARC, 14 years in software and autonomous driving at BMW, and leadership roles at AID (VW Group) and Argo AI, focusing on Level 4 autonomous vehicles.
#AI
#SafeAI
#Engineering
#FraunhoferIKS
#AIsafety
#AutonomousSystems
#TechTalk
#Innovation
#ContinuousEngineering
#AIethics
#KIInstitut
#AIresearch

IRIS Board of Directors
Prof. Dr. André Bächtiger
Prof. Dr. Reinhold Bauer
Prof. Dr. Sibylle Baumbach
Dr. Miriam K.
Prof. Dr. Steffen Staab @ai
Jun.-Prof. Dr. Maria Wirzberger

Institute for AI boosted:
Daniel Hernándezdaniel@mstdn.degu.cl
2025-05-13

Cosimo Gregucci (PhD student at @UniStuttgartAI @Uni_Stuttgart) will present our paper "Is Complex Query Answering Really Complex?" at the International Conference on Machine Learning, #ICML2025. We observed that we can answer many of the current benchmarks for complex queries by reducing queries into basic ones. Are they really complex? How could we advance the field by proposing a more complex set of queries?

Old preprint: doi.org/10.48550/arXiv.2410.12

A screenshot of the paper showing the structures of queries, depending on the links that were seen in training and the links that are missing. Queries are depicted with directed graphs, where missing links are represented with dotted lines and seen links are represented with regular lines.
Institute for AI boosted:
2025-05-13

🌍 Brave Conversations is coming to Stuttgart! 🚀
🟢 5/7/2025, 10am to 04pm
🟢 @Aquarium Campus Vaihingen
🟢 Pfaffenwaldring 45

Since 2017, Brave Conversations has created spaces around the world for people to explore how emerging technologies are transforming our lives—and how we are shaping them in return. Rooted in the spirit of the Web Science Conferences, these events bring critical discussions into the public sphere, inviting diverse voices to reflect, question, and co-create our digital future.

🗓️ Join us for Brave Conversations: Stuttgart 2025 – a dynamic, day-long event open to everyone interested in the societal impact of digital innovation.
We at IRIS – the 🔵 Interchange Forum for Reflecting on Intelligent Systems are proud and excited to co-organize this event alongside School for Talents (with Lisa Kohler), Stuttgarter Change Labs (Spela Setzen), and Intersticia, and to host it at the Universität Stuttgart.
Whether you're a student, researcher, creative thinker, or simply curious about how humans and technologies shape one another — your voice matters in these conversations.

💡 Expect thoughtful dialogue, hands-on learning, and a welcoming atmosphere led by an experienced team of facilitators, including Anni Rowland-Campbell, Hannah Stewart, and Ghada Ibrahim, with contributions from our brilliant team at the University of Stuttgart.

💬 We believe that change begins with dialogue. Come share your perspective and help shape the digital world we want to live in.
📲 Registration: lnkd.in/e4NpUCKz

#BraveConversations
#Stuttgart2025
#IRIS
#DigitalSociety
#WebScience
#HumanTechnology
#DigitalTransformation
#AIandSociety
#TechForGood
#FutureOfTechnology
#EthicsInTech
#DigitalHumanities
#CriticalThinking
#TechDialogue
#DigitalLiteracy
#STEM
#Innovation
#UniversityOfStuttgart
#Intersticia
#ChangeLabs
#SchoolForTalents
#DigitalCulture
#PublicEngagement
#CoCreation
#ReflectivePractices
#CyberValley

IRIS Board of Directors
Prof. Dr. André Bächtiger
Prof. Dr. Reinhold Bauer
Prof. Dr. Sibylle Baumbach
Dr. Miriam K.
Prof. Dr. Steffen Staab
Jun.-Prof. Dr. Maria Wirzberger

If you're part of the IRIS community – like Cyber Valley, @izkt_stuttgart – Internationales Zentrum für Kultur- und Technikforschung der Universität Stuttgart, ARENA2036 or @alexanderbrem we’d love to see you join the conversation!

2025-05-10

@royaheeee @daniel @yuqichengzhu.bsky.social

Here are some pictures of the conference #WWW2025.

A picture of three persons in a row. From left to right: Steffen Staab, Yunjie He, and Daniel Hernández. The picture is inside the International Conference Center in Sydney. In the background there is a big panel printed with the conference logo, and sponsor names.Yunjie He is explaining the paper poster to one of the conference attendants. She is pointing to one of the plots.
Institute for AI boosted:

Several papers from our institute (AC group) have been accepted at top conferences! 🎉 Our colleagues are attending—feel free to reach out if you’d like to connect! #ICLR2025 #WWW2025 #NAACL2025 #AI #NLP #ML

Institute for AI boosted:
Daniel Hernándezdaniel@mstdn.degu.cl
2025-05-01

Yesterday @royaheeee and I presented our paper, DAGE: DAG Query Answering via Relational Combinator with Logical Constraints, at #WWW2025. Existing query answering methods over knowledge graph embedding were limited to decomposable queries, called tree-form queries. Our method provides embeddings for relations that correspond to conjunctions of relations. This idea increases the expressive power of existing query embedding methods.

#KnowledgeGraphs #MachineLearning

dl.acm.org/doi/abs/10.1145/369

Yunjie he is presenting a poster to a woman.Yunjie He is presented a poster to a man.Yunjie he and I are around our poster.
Institute for AI boosted:
2025-04-11

IRIS Insights I Nico Formanek: Are hyperparameters vibes?
April 24, 2025, 2:00 p.m. (CEST)
Our second IRIS Insights talk will take place with Nico Formanek.
🟦
This talk will discuss the role of hyperparameters in optimization methods for model selection (currently often called ML) from a philosophy of science point of view. Special consideration is given to the question of whether there can be principled ways to fix hyperparameters in a maximally agnostic setting.
🟦
This is a WebEx talk to which everyone who is interested is cordially invited. It will take place in English. Our IRIS speaker, Jun.-Prof. Dr. Maria Wirzberger, will moderate it. Following Nico Formanek's presentation, there will be an opportunity to ask questions. We look forward to active participation.
🟦
Please join this Webex talk using the following link:
lnkd.in/eJNiUQKV
🟦
#Hyperparameters #ModelSelection #Optimization #MLMethods #PhilosophyOfScience #ScientificMethod #AgnosticLearning #MachineLearning #InterdisciplinaryResearch #AIandPhilosophy #EthicsInAI #ResponsibleAI #AITheory #WebTalk #OnlineLecture #ResearchTalk #ScienceEvents #OpenInvitation #AICommunity #LinkedInScience #TechPhilosophy #AIConversations

Institute for AI boosted:
Universität StuttgartUni_Stuttgart@bawü.social
2025-04-08

🧐💪Du löst gerne Probleme? Und arbeitest am liebsten kreativ im Team? Dann melde dich jetzt für die Next Generation Energy Challenge #NGEC an und baue dein eigenes #Smarthome!💡 📱

So funktioniert die Challenge:
👉 Bildet ein Team mit 4-6 Personen.
👉 Nach der Online-Anmeldung erhaltet ihr ein Set mit allen möglichen Bauteilen und Werkzeugen, die ihr für euer Smarthome braucht.
👉 Tüftelt gemeinsam an eurer Idee - zum Beispiel in einer Schul-AG oder nach den Hausaufgaben.
👉 Hilfe erhaltet ihr von unseren Wissenschaftlerinnen und Wissenschaftlern in einer wöchentliche Online-Fragestunde.

Das ist der Zeitplan:
🧑‍💻 Einführungstermin am 05.05. um 18 Uhr
🔧 Start der Challenge am 12.05.
📱 Baut euer Smarthome bis zum 30.07.
🏆 Präsentiert euer Smarthome an #EneryDay am 02.10. Es gibt tolle Preise zu gewinnen!

Neugierig? Dann sichere dir und deinem Team jetzt einen Platz! Mehr Infos findest du auf unserer Webseite 👉 sohub.io/lgjb

Institute for AI boosted:
2025-03-14

We cordially invite all interested parties at the Universität Stuttgart to our IRIS Colloquium on March 26, at 2 p.m., in Room 101 at Universitätsstr. 32.
🟦
Mara Seyfert will give her lecture, „Uncertainty and robustness against persuasion in large language models.“
🟦
Today's large language models (LLMs) excel at providing convincing answers across a broad spectrum of inquiries, with their conversational capabilities enabling them to closely align with users' needs. However, this adaptability is beneficial only to the extent as models remain robust to adopting wrong statements from user inputs.
🟦
Recent research demonstrates that even advanced models like GPT-4 can shift from initially correct answers to incorrect ones during multi-turn conversations solely due to user input. In my talk, I will present my research exploring how uncertainty in LLMs can provide insights into their robustness against persuasion while highlighting the specific challenges of quantifying uncertainty in these models.
🟦
The lecture is held in English. Registration is not necessary.
🟦
#LargeLanguageModels #AI #GPT #GPT4 #robustnessagainstpersuasion #AIResearch #MachineLearning #LLMs #ArtificialIntelligence #TechTalk #RobustAI #DataScience #AIethics #Innovation

2025-03-14

How easy is to persuade a large language model?

The Interchange Forum for Reflecting on
Intelligent Systems (IRIS @Stuttgart_IRIS @Uni_Stuttgart) invites all interested students and university staff to their next IRIS colloquium on March 26 at 2 PM in the room UN 302.101, where Mara Seyfert will talk about uncertainty and robustness against persuasion in large language models.

#LargeLanguageModels #AI #GPT #GPT4 #RobustnessAgainstPersuasion #AIResearch #MachineLearning #LLMs #ArtificialIntelligence #TechTalk #RobustAI #DataScience #AIethics #Innovation

Poster for the IRIS colloquium. It mentions the title, the speaker, and the place. It has a drawing of a circle semi full of water. The water has waves and the word colloquium is written over the water and reflected in the water.
2025-02-22

Controlling a computer with your head or foot: Semanux, a spin-off from the Analytic Computing department of the Artificial Intelligence Institute of the University of Stuttgart (@Uni_Stuttgart), has taken a major step forward by becoming part of the Alfa Group of companies from Karlsruhe. This move strengthens Semanux’s ability to develop groundbreaking solutions for hands-free computer control, making digital accessibility more intuitive and inclusive.

ki.uni-stuttgart.de/institute/

beschaeftigte.uni-stuttgart.de

semanux.com/en/blog/2024-11-28

Five persons in an office are looking at the camera. Two persons shake hands in agreement, and showing a folder including the agreement document.
Institute for AI boosted:
2025-02-21

Why is the integration of robots into the construction industry considered crucial? How can human-robot collaboration (HRC) be enhanced in the construction industry? What new perspectives does feminist technoscience offer on human-robot collaboration in construction and prefabrication?

Gili Ron, Thomas Wortmann, Cordula Kropp, and Achim Menges (IRIS members) published an article about "Human-Robot Reconfigurations: Advancing Feminist Technoscience Perspectives for Human-Robot-Collaboration in Architecture and Construction" in M. Kanaani (Ed.), The Routledge Companion to Smart Design Thinking in Architecture & Urbanism Fora Sustainable, Living Planet (1st edition, pp. 669–679). Routledge, Taylor & Francis Group.

Abstract
Integrating robots in the construction industry is an essential requirement for urgently needed improvements in productivity and sustainability. This integration requires new perspectives on human-robot collaboration (HRC) to retain skilled staff and support skills and decision-making while considering conditions on-site. This chapter reflects on HRC for prefabrication and construction from the perspective of feminist technoscience, a transdisciplinary field offering distinct ways of thinking about society, technologies, bodies, power, and the environment. Specifically, the chapter proposes new perspectives on agents, agency, and collaboration. Applied to HRC, these perspectives can for example imply (1) situated interfaces supporting multi-agent collaborations of skilled workers and robots, (2) human training of robots on how to adapt to changing conditions and uncertainties on-site enabled by machine learning, and (3) collaborative prefabrication and construction processes supporting on-the-fly design decisions. Such approaches reconsider “agency” as a collaborative unit of human and non-human agents and foster trust and sensible task distribution for improved efficiency in fabrication.

Read more here: lnkd.in/eda6Qxw7

Institute for AI boosted:
Finnish Center for AIFCAI@mastodon.online
2025-02-15

In the latest FCAI newsletter:

- ELLIS Institute Finland PI recruitment is ongoing! Apply by March 9 ( @finland )
- ELLIS member Vikas Garg shares his perspective on doing research in Finland
- Events: #LLM seminar, #GenerativeAI and copyright, Inverse Problems SIG, and more

Read more here: mailchi.mp/22e737034e61/fcai-n

#newsletter #artificialintelligence #machinelearning #hiring #getfedihired #jobs #academicjobs #research
@ELLISforEurope

Text: ELLIS Institute Finland; photo of people with laptops sitting around a table; illustration of two human figures holding puzzle pieces; logo of Finnish Center for Artificial Intelligence FCAI
Institute for AI boosted:

Der Bericht zum Programm #OpenAccess Publikationskosten für 2024 ist veröffentlicht: Zum Stichtag 4. Mai sind 55 Anträge auf Förderung bei der DFG eingegangen, Antragsvolumen: 59,1 Mio €. Antragsberechtigt (auch 2025) sind Hochschulen und außeruniversitäre Einrichtungen.
zenodo.org/records/14808601

#LoveDataWeek

2025-02-10

Unsupervised Model Monitoring through Explanation Shift - Transactions on Machine Learning Research (TMLR) paper introduces a novel instrument of explanation distributions.

The benefits of machine-learned models only apply if the training and the test data come from the same distributions. In many application scenarios, distribution shifts outdate the learned models—often surprisingly quickly. Model monitoring is required not to build your application on wrong assumptions. In the best world, up-to-date validation data would allow for fully informed model monitoring. In realistic scenarios, such up-to-date validation data is frequently missing, requiring methods for unsupervised model monitoring.

Established unsupervised model monitoring approaches check for shifts of input data or predicted outcomes, but both lead to too many false positives and negatives.

Attribution approaches (e.g., Shapley Values) evaluate how a learned model picks up or ignores features of an input data record. An explanation distribution, thus, represents how a learned model handles data. The core novelty of our approach is to introduce this notion of explanation distribution and check for shifts of explanation distributions from training to test data, i.e., monitor a potential explanation shift.

Check out the paper: Carlos Mougan, Klaus Broelemann, Gjergji Kasneci, Thanassis Tiropanis, Steffen Staab. Explanation Shift: How Did the Distribution Shift Impact the Model? In Transactions on Machine Learning Research, 2025. openreview.net/forum?id=MO1slf

Also on ki.uni-stuttgart.de/institute/

This figure shows the matrix with the strength analysis of features driving the change in the model, on the y-axis are the features, and
on the x-axis are the different states. The matrix cells with higher values are filled with lighter colors.
2025-02-06

I know what you did on my user interface! No, we have not invented a new method to spy on you. Rather, we suggest a novel paradigm that helps usability experts understand in which situations users interacting with a rich Website progress smoothly and in which they don't. The key to understanding users' problems is summarizing comparable situations that various users encounter at different points of their website journey.

Because of the complexity of modern Web GUIs, this requires the analyses of videos recording user interactions on websites. Our novel approach to discovering visual stimuli employs machine learning to judge the similarity between recorded situations. Our paper, accepted on ACM Transaction on the Web accepted our paper about a novel paradigm for studying Website users, evaluates several case studies to demonstrate the effectiveness of our proposed method. It also surveys usability experts who confirm the usefulness of our suggested approach.

The paper results from a long-running collaboration between (i) Steffen Staab from Analytic Computing, (ii) Raphael Menges from Semanux GmbH, a spin-off from Analytic Computing, (iii) Christoph Schäfer and Tina Walber from EyeVido GmbH, a spin-off from Staab’s former research group at University of Koblenz and now a part of Tobii AB, the leading manufacturer of remote eye trackers, and (iv) Chandan Kumar, a team lead at Fraunhofer IAO, and previously a postdoc at Analytic Computing.

Read more: ki.uni-stuttgart.de/institute/

The paper: doi.org/10.1145/3715881

Graph that shows the relations between the phases in usability analysis and the sections of this article.
Institute for AI boosted:
Universität StuttgartUni_Stuttgart@bawü.social
2025-02-05

Wie wehrhaft ist unser #Grundgesetz? Darüber diskutieren Expert*innen auf der #Tagung mit #Podiumsgespräch „Verfassungsresilienz - Das Grundgesetz vor seinen Feinden schützen“ 🏛️🔍

📅 10. Februar 2025
⏰ 14:00 - 20:30 Uhr
📍 Uni Stuttgart, Campus Stadtmitte

Die Veranstaltung findet in Kooperation mit der Landeszentrale für politische Bildung Baden-Württemberg statt.

Jetzt anmelden. Alle Infos 👉 sohub.io/do73

#Demokratie #UniStuttgart @lpb

Institute for AI boosted:
ELLIS Institute Finlandfinland@ellis.social
2025-02-05

ELLIS Institute Finland is hiring Principal Investigators in AI + machine learning. World-class resources for research incl. LUMI supercomputer, generous starting package & professorship affiliation with a university in the world’s happiest country!

Apply by March 9: ellisinstitute.fi/PI-recruit

@samikaski
@ELLISforEurope

Logo of ELLIS Institute Finland (line drawn map of Europe)
2025-02-04

Machine learning turns volunteers’ lower-quality #OpenStreetMap data into quality indicators for official, authoritative data. Novel method and case studies accepted for publication in ACM Transactions on Spatial Algorithms and Systems (#TSAS).

Fighting fires, containing floods, or police missions, these and other situations require the availability of high-quality map data. Authoritative data exhibits such high quality at entry time, but expert geographers update it only every few years by scanning the whole environment, leading to many stale and misrepresented geographical regions. Volunteered map data, such as open street maps, contains many mistakes, as it is not captured by experts. However, volunteers are often very motivated to refresh map data quickly. We suggest to combine the best of both worlds and exploit machine learning to point experts to regions that need attention of data curation.

Insitute website: ki.uni-stuttgart.de/institute/

Paper: dl.acm.org/doi/10.1145/3715910

A low quality OpenStreetMap detail of a mapa. The image includes streets and buildings.
2025-01-29

A new NAACL paper proposes a principled way to estimate the uncertainty of knowledge graph link predictions.

A knowledge graph is a set of relationships between entities. For example, the relationship "Alice is diagnosed with common cold" relates the entities "Alice" and "common cold" stating a diagnosis. The link prediction task consists of predicting a missing entity in a relationship. For example, the x in "Alice is diagnosed with x" can be filled with "common cold" or "lung cancer."

Existing methods provide a score to every entity so that one can select the first k entities with a higher score. However, they do not provide uncertainty information, are not calibrated, and lack probabilistic information.

In a paper accepted at the 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025), "Conformalized Answer Set Prediction for Knowledge Graph Embedding," Yuquicheng Zhu (@yuqichengzhu.bsky.social), Nico Potyka, Jiarong Pan, Bo Xiong, Yunjie He (@royaheeee), Evgeny Kharlamov, and Steffen Staab, study how many entities we need to guarantee coverage of the actual answer at a pre-defined confidence level (e.g., 90%). The more entities we need, the more uncertain the model is about its predictions.

Read more on ki.uni-stuttgart.de/institute/

Figure showing the entities and their scores, and the coverage set for the actual answer with a giving certainty.

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst