Volume 67, Issue 1 - Special Issue: Junior Fellows and Distinguished Dissertation of the GI & AI for Crisis; Guest editors: Ricardo Usbeck, Angelie Kraft, Patrick Westphal

With great power comes great responsibility. As Artificial Intelligence in general and Natural Language Processing, in particular, continue to shape the world we live in, there is an increased need for Computer Scientists in academia and industry to acknowledge the pivotal role our discipline plays in shaping the future of our society. Accepting this role should come with the responsibility to make positive contributions, whether through research, teaching, application, or by contributing to the public discourse surrounding AI technology and its regulation. This text outlines how Natural Language Processing can contribute to shaping a better future in which society at large can benefit from technological advances.

Machine learning (ML) is increasingly deployed in critical domains such as healthcare, finance, and autonomous driving, where the use of sensitive data raises significant privacy challenges. My research places individuals and their data at the center of ML privacy, building systems that protect individuals’ privacy without sacrificing performance. I focus on (1) exploring the threat space in ML privacy to inspire targeted protection, (2) analyzing the root cause of privacy leakage from ML models, and (3) developing individualized privacy guarantees that protect data according to individuals’ unique needs while improving privacy-utility trade-offs. My vision is to advance privacy-preserving ML to address the evolving challenges of increasingly complex ML models and systems. As models grow in scale, integrate diverse data modalities, and become embedded in critical societal applications, protecting individual privacy becomes both more urgent but also more challenging. My goal is to create methods that ensure privacy across a broad spectrum of ML applications, while also addressing the interplay between privacy and other trustworthy ML aspects, and aligning technical privacy measures with legal and societal expectations to meet individual rights.

It is the article’s overall aim to elucidate the contingency and volatility inherent to academic biographies. In order to derive these, but also underpin them, the article begins by outlining the argument that Mareike Lisker makes in her Master’s thesis, which explores the nexus between the disciplines of computer science and philosophy. There, she posits that the demand for more digital literacy places an onerous responsibility on individual users when it comes to the control of their own data. She argues that the individual users are structurally ill-equipped to meet that responsibility in face of all-encompassing tracking infrastructures. In accordance with the aim of the article, the thesis’ topic will be situated within Lisker’s scientific career path. This path will then be traced up until her current PhD project, which focusses on content moderation on decentralized platforms.

As artificial intelligence (AI) increasingly permeates high-stakes domains such as healthcare, transportation, and law enforcement, ensuring its trustworthiness has become a critical challenge. This article proposes an integrative Explainable AI (XAI) framework to address the challenges of interpretability, explainability, interactivity, and robustness. By combining XAI methods, incorporating human-AI interaction and using suitable evaluation techniques, the implementation of this framework serves as a holistic XAI approach. The article discusses the framework’s contribution to trustworthy AI and gives an outlook on open challenges related to interdisciplinary collaboration, AI generalization and AI evaluation.

The author’s dissertation, entitled “Scalable SAT Solving and its Application”, advances the efficient resolution of instances of the propositional satisfiability (SAT) problem, one of the prototypical “hard problems” of computer science with many scientific and industrial real-world applications. A particular focus is put on exploiting massively parallel computational environments, such as high-performance computing (HPC) systems or cloud computing. The dissertation has resulted in world-leading solutions for scalable automated reasoning and in a number of awards from the SAT community, and has most recently been acknowledged with a GI Dissertation Award. The article at hand summarizes the topic, approaches, and central results of the dissertation, estimates the work’s long-term impact and its role for future research, and closes with some personal notes.

When we look at the global knowledge graph landscape, we quickly find that there are billions of interconnected facts that have the potential to answer all kinds of questions. However, a persistent challenge lies in finding corresponding questions that align with these facts. The availability of these questions along with matching SPARQL queries is an important prerequisite for fine-tuning Large Language Models for domain-specific query generation, which is why we propose Queryfy , a novel framework that leverages Large Language Models to automate the task of deriving questions and queries from knowledge graphs, empowering users to harness their full potential.