Workshops​

Adversarial AI - The Security of AI

Over the past few years, attacks targeting AI systems have become increasingly prevalent. To highlight the significance of this issue, we will examine a specific attack on an AI application, provide a brief overview of various Adversarial AI attack methods, and introduce TAISHA, the TNO AI Security Hazard Advisor. TAISHA is designed to assist experts in identifying potential AI-specific vulnerabilities within their applications.

Public Prosecution Service - AI-driven decision support for handling Mulder objections (Voertaal Nederlands)

The Public Persecution Office processes hundreds of thousands of objection letters annually. In this workshop, we will demonstrate the Mulder Miner, an AI-driven, human-in-the-loop pipeline designed to support decision-making in handling objections to traffic fines. The Mulder Miner aids judicial professionals through argument mining, argument classification, and response letter generation. Additionally, we will explore how this technology can potentially be generalized to other use-cases.

This workshop will be given in Dutch.

Health & Work - Generative AI applications in Health & Work

Abstract: to be announced

Oversight Lab - Joining AI Algorithms and Governance for Responsible AI Adoption

Responsible AI adoption asks for many considerations and decisions at different levels, from technical perspective but also from organizational and ethical viewpoint. These decisions are often intertwined and require an interdisciplinary perspective. In this session we dive into interdisciplinary decision-making about AI, the balance between risks, benefits, resources and values, and applying frameworks in practice. We share lessons learned from public sector usecases and we interactively discover these topics together.

ObjectivEye - AI-driven intelligence for faster, fairer, and skills-based hiring in an evolving job market

This session showcases the Appl.AI use-case of ObjectivEye, a platform that translates TNO’s cutting-edge research into a practical tool for unbiased, skills-based recruitment. As we approach a major milestone—the first real-world pilot with live applicants and hires—we invite you to explore how foundational research in knowledge-driven and responsible AI has been brought to life.

We’ll guide you through the journey from a visionary concept to a working platform, demonstrating how ObjectivEye applies TNO’s expertise in skills-matching and ethical AI design. In the second part, we’ll share a preview of upcoming innovations, including customized language models and responsible AI practices, and explore how we stay connected to research through collaborations like the TNO AI Oversight Lab.

The session will conclude with an open discussion on the role of AI in recruitment, co-creation opportunities, and how government, industry, and research partners can contribute to building fair and future-proof hiring practices.

BioAnalyst - A Foundation Model for Biodiversity Conservation

In this talk, we will introduce BioAnalyst, a large-scale, multimodal AI model, pre-trained on diverse ecological data modalities. BioAnalyst aims to enhance biodiversity monitoring, prediction, and conservation efforts, while being flexible and robust to any kind of downstream task, from classification to prediction. Still, like any other advance AI models, BioAnalyst comes with a series of challenges from data to scaling that requires careful engineering. This talk will highlight the progress made so far, while feature an interactive segment about the opportunities and impact of this modelling paradigm.

AutoAdapt - Towards self-adaptive systems - integrating fleet-based learning and AI-based decision making

Today’s control development is at a turning point for complex, connected systems. Traditional methods lead to high costs and long development times, compromising performance. Future systems must be self-adaptive, accelerating innovation and optimizing performance in real-time for sustainability and competitiveness.

This workshop explores real-time learning from data sources to enhance decision-making. Our innovative approach is illustrated for the use case of an electric distribution trucks, aiming to minimize energy consumption, maximize battery life, and minimize operational costs

Highlights:

• Discover how to manage complexity and real-life variations with self-adaptive concepts;

• Learn how to leverage physics-informed Bayesian Networks to monitor and adapt to changing system capabilities;

• Gain insight into risk-averse decision support to handle varying operational conditions;

• Experience a demo showcasing the challenges in optimal use of battery electric vehicles

FATE - Trustworthy AI decision support systems for healthcare, with Diabetes Type 2 as use case

‘How can AI truly support high-quality healthcare decisions that suit the patient’s needs in a fair and responsible way compliant to relevant regulations? In this interactive workshop, we take you through a scrollytelling story of a patient at risk of diabetes—where AI-driven decision support doesn’t always get it right. Through guided discussion and reflection, we’ll explore critical issues like privacy, fairness, and adaptability in AI for healthcare. This is your chance to reflect, share insights, and help shape AI solutions that truly meet the needs of the healthcare sector. Join the conversation and make an impact!’

GRAIL - Generative responsible AI

In this workshop we will guide you through the developments within the SNOW project. The project aims to close the gap between research and operational application. We will share our vision for the future of the effective and safe deployment of human-machine teams in highly complex environments and use-cases. Starting from the early development of the core-components towards flexible delegation of tasks between human and robot. To top it off we will give you a sneak peek into the mind of our robots to examine the knowledge representations that are used to capture and understand the open-world before, during, and after autonomous operations.

MMAIS - Realizing ethical AI responsibly

Do you ever wonder how AI systems can be developed responsibly and ethically? We want our AI systems to be developed in line with our societal values, such as privacy, fairness, and safety. During this workshop, you will learn how this can be tackled. You will experience how to consider and specify ethical implications using concrete methods developed by TNO. Methods that can also be applied within your organization.

SNOW - Teaming up with flexible autonomous systems

In this workshop we will guide you through the developments within the SNOW project. The project aims to close the gap between research and operational application. We will share our vision for the future of the effective and safe deployment of human-machine teams in highly complex environments and use-cases. Starting from the early development of the core-components towards flexible delegation of tasks between human and robot. To top it off we will give you a sneak peek into the mind of our robots to examine the knowledge representations that are used to capture and understand the open-world before, during, and after autonomous operations.

SEAMLESS - Challenges and Solutions for Systems Engineering and Lifecycle Management of Automated and Autonomous Systems

With a bit of system science, we illustrate how things become uncertain once AI is an integral system part, re-evaluate architecting and design processes, and show how adaptive AI invalidates stability assumptions, which hinders Verification, Validation, and Preventive Maintenance.

Based on this problem understanding, we discuss, with you, where the biggest challenges are and confer possible system engineering approaches to solve them. Interestingly, such solutions do not only need to tackle future, AI-based systems: in all likelihood, they will also have some AI inside – and that means change.

So, we will explore with you, a topic at the heart of Future System Engineering:  What do you consider applicable, feasible, and helpful to engineer trustworthy systems with AI inside?