AI, Trust & Design - Meetup #1 in Leuven
AI is everywhere in the products we build. But do users actually trust it?
Four speakers from research and industry share their inspiring approach to building AI products that are transparent, reliable, and user-centered. Expect concrete patterns, working demonstrators, and insights you can apply in your own projects.
Attendance is free and open to everyone, with plenty of time to network over drinks and bites afterwards.
Program
18:30 - Doors open, drinks & fingerfood 19:00 - Four talks (all in English) 20:30 - Networking & drinks
Talks
Jad Najjar Co-Founder and CEO, Eummena
AI Empowering Learning Experiences
Jad opens the evening with a talk on the role of AI in learning platform design and implementation, focusing on the why rather than the how. Eummena is an EdTech company that grew out of the HCI research group at KU Leuven and helps organizations deploy AI in education and digital transformation.
Rutger De Wilde User Centered Experiences, Odisee
Your LLM Needs a Straitjacket: A practical pattern for building AI conversations that stay on track
Free-form LLM chatbots are easy to build and hard to trust. They hallucinate policies, skip steps, and give advice they shouldn't. The problem isn't the model, it's that we hand it the wheel and hope for the best.
This talk explores what it looks like to wrap an LLM in a deterministic shell without killing what makes it powerful. Built around a real-world demonstrator, and exploring how the same thinking applies to helpdesk bots, webshop assistants, onboarding flows, and more. Relevant for anyone who writes the code, designs the flow, or decides what gets built.
Katrien Verbert Augment, KU Leuven
From transparency to agency: interactive explanations for user empowerment
Despite a long history of research on explanations in Machine Learning, AI, and Recommender Systems, current efforts face unprecedented challenges: contemporary models are more complex and less interpretable than ever. As such models are increasingly embedded in day-to-day applications, justifying their decisions to end users has become increasingly crucial. In addition, several researchers have voiced the need for conversational and interactive explanations as a core requirement to support user empowerment. Such interaction methods enable users to steer models through input and feedback.
In this talk, Katrien will present work on interactive explanation methods tailored to the needs of end users, such as healthcare professionals and job seekers. She will also discuss work on combining data-centric and model-centric explanations to support user feedback, enhance model understanding, and address issues such as bias and trust. The results highlight how interactive and conversational approaches can transform explainability from a passive presentation of information into an active, collaborative process in which users and AI systems learn and adapt together.
Henri Jacobs FORCIT
Navigating the UX hurdles of trustworthy expert management technology
AI can recommend, rank, and automate faster than ever, but in expert-driven environments, speed alone is not enough. If users do not understand why a system suggests a certain expert, challenge framing, or next action, they hesitate to rely on it. And if the experience is not intuitive enough, they bypass it.
In this talk, Henri will share how FORCIT is building expert management technology that helps organizations match and predict their challenges to the right human and/or synthetic experts, structure collaboration, and embed these flows into the daily tools of users such as Slack, ChatGPT, and WhatsApp. He will discuss the UX challenges at the intersection of AI, expert judgment, and workflow design: making recommendations transparent without overloading the interface, balancing automation with human control, and creating experiences that feel fast and seamless while still remaining fair, explainable, and credible.
Who is this for?
Designers, developers, product managers, CTOs and anyone building, evaluating or adopting AI-powered products. You'll meet people from research and industry working on the same questions. A good opportunity to network, learn, and discover how others are tackling the challenges around trustworthy AI.
Getting there
Address: De Hoorn, Sluisstraat 79, 3000 Leuven
Parking: Parking Vaartkom (Engels Plein 32), a 2-minute walk. First hour free, evening rate €3.
Public transport: 15-minute walk from Leuven station. Bus stop Engels Plein at 5 minutes.
Bike: Bike parking available at the venue.
About the organizers
This meetup is organized by ULTRA, a collaboration between Odisee, KU Leuven, and Arteveldehogeschool, funded by VLAIO. We bring together researchers and practitioners around the question of how to design AI interfaces that are user-friendly, transparent and trustworthy.
This is the first edition of a meetup series. Free admission, open to everyone.