Announcement: Exciting PhD Position in Human-Centered Explainable AI

KU Leuven is pleased to announce an exciting opportunity for a fully-funded PhD position in Human-Centered Explainable AI (XAI), referenced as BAP-2024-533. This position is a key part of a groundbreaking interdisciplinary research project involving the Augment team at the Department of Computer Science, the LIRIS research group at the Faculty of Economics and Business, and the tutorial services of the Faculty of Engineering Science. The supervisors for this project are Dr. Katrien Verbert, Dr. Monique Snoeck, and Dr. Tinne De Laet.

Project Overview

Explainable AI (XAI) has gained significant attention in recent years as researchers and practitioners strive to make complex AI models more transparent and understandable. Historically, the AI and Machine Learning (ML) fields have explored various explanation methods, but with the rise of increasingly complex models, the need for innovative approaches has become more pressing. This PhD project aims to address the limitations of current explanation methods by focusing on enhancing the interpretability of AI models for non-expert users. The research will build on the existing body of work, which categorizes explanation methods into several types. These include:

1. Visualizations: Graphical representations of a model’s inner workings.
2. Knowledge Extraction: Methods that infer rules to approximate the model’s decision-making process.
3. Influence Methods: Techniques that estimate the relevance of features affecting the model’s outcome.
4. Example-Based Explanations: Methods that explain model outcomes by comparing them to similar or different individual cases.

Despite these advancements, current methods often provide limited insight into the models' reasoning processes, particularly for users with little AI expertise. This project aims to develop the next generation of interactive explainability methods that will improve user understanding of model outcomes and enhance their ability to provide feedback for model improvement.

Research Goals

The primary objectives of this research are:

• Development of Interactive Explainability Methods: Creating new approaches that allow users with minimal AI knowledge to understand complex model outcomes effectively.
• Integration and Extension of Explanation Methods: Combining various types of explanation methods to provide a more comprehensive understanding of model behavior and underlying data.
• Empowerment of Non-Expert Users: Enabling users to provide meaningful feedback to improve AI models based on their understanding of the explanations provided.

Candidate Profile

We are looking for highly motivated candidates with:

• An excellent Master’s degree in Computer Science or a related discipline.
• Strong programming skills and a proven ability to conduct independent research.
• Interest in human-computer interaction research and language models.
• Strong commitment and the ability to work effectively in a team.
• High proficiency in English, both spoken and written.

Offer and Application

Funding for this PhD position is available immediately. The research will be conducted at KU Leuven's Department of Computer Science and the Faculty of Economics and Business, located on the Heverlee campus.Interested candidates should apply by August 28, 2024. For more information or to apply, please visit the online application portal. 

For specific inquiries, contact Prof. Dr. Katrien Verbert at +32 16 32 82 86 or via email at katrien.verbert@kuleuven.be, Prof. Dr. Ir. Tinne De Laet at +32 16 32 70 75 or tinne.delaet@kuleuven.be, and Prof. Dr. Monique Snoeck at +32 16 32 68 79 or monique.snoeck@kuleuven.be.

Commitment to Diversity

KU Leuven is committed to fostering an inclusive and respectful environment. We embrace diversity and recognize the value it brings to research and education. We are dedicated to equal opportunity and do not tolerate any form of discrimination. For questions about accessibility or support, please contact us for assistance.

This PhD opportunity represents a significant step forward in the field of Human-Centered Explainable AI, promising to drive innovations that make AI more accessible and understandable for all users. We encourage all qualified individuals to apply and join us in this important research endeavor.