Programma

In recent years, the creation of sovereign national and European Large Language Models (LLMs) has accelerated significantly. With numerous initiatives underway, including major projects like OpenEuroLLM and GPT-NL, as well as community-driven efforts such as GEITje, various stakeholders are collaborating to develop LLMs that are trained on official EU languages, leverage national and EU infrastructure, and adhere to open and transparent data management practices.

This discussion will delve into the technical and policy challenges associated with developing these models, while also exploring the potential benefits and implications they may have for research and the public sector. Both for the presence as well as the future.

Join us in this open discussion and let's connect our minds!

Johannes Schleiss (SURF) will moderate this discussion.

The FAIR principles are pushed as a standard for a decade now, but implementation is lagging. In this Challenge session we will investigate and discuss with you, experts and your peers, how to improve FAIR practices in research.

How big is the gap between FAIR as a policy and FAIR as daily practice in your research environment? What can you do in your role to close that gap? What kind of support is needed to help you? Together, we will look for the best ways to move forward and implement the FAIR principles in practice.

Do you have good ideas for this or specific challenges you want to bring to the table? Join us in this open discussion and let's connect our minds!

Jeroen de Wijn (SURF) will moderate this discussion.

How can you reduce undesirable dependency on external digital platforms, tools and infra and take greater control of your research?
In this interactive Challenge Session, you’ll explore the future of digital autonomy in an open space together with researchers, experts and peers. This is not a passive talk, it’s a dynamic, thought-provoking discussion where your perspective, thoughts and experience matter. Together, we’ll explore the impact of technology, the role of policy, and the power of the research community.

Join us in this open discussion and let's connect our minds!

Annette Langedijk (SURF) will moderate this discussion.

In this Challenge session we will investigate and discuss with you, experts and your peers what it takes to create sustainable infrastructure to support open science practices.
We all know it. The sustainability paragraph of your Project proposal. Of course you want your research infrastructure/tools to last beyond the project period. But you also know, there are many stakeholders that determine whether or not that can be realised. Guarantees are hard to give, but how can you optimise the chance that your infrastructure will last?

Join us in this open discussion and let's connect our minds!

John Doove (SURF) will moderate this discussion.

In this Challenge session we will discuss how you share one or more infrastructures (data and compute) when you collaborate during your research project. In the session, we will prioritize the topics for further discussion and focus on the main challenges and potential solutions. 
Think of:

1) Data storage and Discovery: Where is the data stored and how can your collaborators find it? 
2) Data governance: where to implement data governance policies? 
3) Compute workflows: how is this embedded/automated in your compute workflows? 
4) Future-proofing data: what about the future storage challenges for re-use? 
5) Sensitive data handling: what about sensitive data needs? 
6) Cost and ownership: who should pay for what and where is the ownership?

If any of these challenges have crossed your paths, this challenge session is your opportunity to exchange ideas on how to solve these challenges. 
Join us in this open discussion and let's connect our minds!

Magriet Miedema (SURF) will moderate this discussion.

Do you also struggle to equip researchers with the right knowledge and digital skills? Rapid technological developments - such as AI, new digital research infrastructures, and evolving data practices - mean that training needs are constantly changing. Come to our discussion session to share the bottlenecks you experience.

The landscape of digital skills training is fragmented, as many organisations offer training to researchers on a variety of topics. Local Digital Competence Centers have the ambition to improve the organisation and delivery of training for researchers. One of the means to achieve this is through active collaboration. 

Through an open discussion with the audience, fueled by experts in the field, we want to collect input that we can use in further collaborations. The discussion will focus on practical questions: How to collaborate (within and across institutions) on developing and maintaining training materials? What are the main barriers that need to be overcome? And how can we ensure that these efforts are sustainable?

Two current examples that aim to tackle this challenge and find sustainable solutions for delivering digital skills training via regional collaborations are the projects - Programming CAFEs and Building Digital Capacity for NES (DC4NES). 
Programming CAFEs (Code Along, Feel Empowered) foster inclusive spaces where researchers can develop and share coding skills. DC4NES focuses on regional Training Hubs and Collaborative Lesson Development, enabling institutions to pool instructors, share responsibilities, and deliver workshops even with limited resources.

These examples, together with other initiatives put forward by experts from the audience, will serve to inspire the discussion. We are looking forward to hearing your perspective!

Eva Lekkerkerker from Netherlands eScience Center will host the open discussion.

Does your research rely on interviews, qualitative fieldwork, or creative methods — or do you build digital infrastructure for researchers who do? Either way, you are likely hitting the same gap from opposite sides.

While national and European data strategies race to create infrastructures to securely and efficiently manage "Big Data" (e.g., satellite imagery, sensors, IoT streams), a quieter crisis is unfolding for "Thick Data" (e.g., qualitative insights, narratives, observations, visuals), as it lacks appropriate infrastructure for it to thrive. Thick data is experiential, contextual, and meaning-rich, yet often unstructured, which makes it challenging for existing infrastructure to enable FAIR (Findable, Accessible, Interoperable, Reusable) use. Furthermore, thick data workflow is never neutral - it shapes whose experiences become data, who interprets them, and who gets to reuse them.

In this open discussion, you will explore an interactive dialogue on:

• Infrastructure gaps: What gaps do you face when collecting, storing, analysing, and sharing thick data, and where does the current infrastructure fall short?
• FAIR for thick data: How FAIR principles can be applied to thick data? You will debate how FAIR applies when your data is a sketch, an interview transcript - not a spreadsheet.
• AI tools: how emerging AI tools for transcription, thematic analysis, and pattern recognition are opening new possibilities for thick data research and what risks of bias or loss of nuance come with them.

Join us to take first steps toward a thick data infrastructure — one that treats human experience with the same rigor as numerical data.

Do you still see researchers downloading large datasets to their laptops before starting analysis? Despite the availability of cloud platforms and large online datasets, this remains common practice in many research fields. But what is holding us back from more efficient, cloud-native approaches?

In this open discussion you will exchange experiences with others who support research infrastructures and digital research practices in the Netherlands. Together you will explore what is needed to make cloud-native data access and processing easier and more widely adopted across research domains.

You will discuss questions such as:

• What technical, organisational or skills barriers still prevent researchers from using cloud-native workflows?
• Where are the biggest gaps in infrastructure, training, standards or coordination?
• How can community-driven initiatives be supported by institutions and national infrastructures?

You will work together to identify concrete challenges, needs and opportunities. The goal is to produce a shared list of priorities and possible next steps for the Dutch research ecosystem.

The discussion builds on experiences from recent national initiatives exploring cloud-native research infrastructures, but the focus will be broader: connecting insights across disciplines and institutions.

If you work on research support, digital infrastructure, data services or policy, this session offers a chance to compare perspectives and help shape future approaches for more efficient research workflows.

Do you also spend months arranging access to data, models or computing power from other institutions before your research can even begin? Many collaborative projects run into delays because infrastructure, agreements and access are arranged ad-hoc every time.
The DigiLab Toegepaste Kennis virtual research environment aims to change this. This initiative is being developed and explores how knowledge institutions can securily connect data, models and computing power so researchers can collaborate much faster across organisations.

In this session you will learn about our ambitions and ideas in realizing this facility. You will also learn about the first steps we took in this 5-year project funded by the Ministry of Economic Affairs with the help of specialists SURF and TNO ISP (ICT, Strategy and Policy). This involves challenges in architecture and security, governance and cost coverage, community building, implementation and adoption. The community building involves not only researchers, but also developers, ICT specialists, clients and stakeholders.

We look forward to reaching out to you in this session to connect with us and starting sharing knowledge and experiences, and hopefully data, model and compute power in the future.
This session is also an invitation. The environment will only succeed if it reflects the needs of the research community. If you collaborate across institutions or expect to in the future you can help shape how this facility should work and how researchers like you could use it.

Erik Kentie (SURF) will moderate this discussion.

Unsure how to make your data, software, or workflows truly interoperable? Do standards such as file formats, metadata schemas, and shared vocabularies feel difficult to apply in practice ? And how do these challenges relate to integrating and collaborating across fragmented research-IT infrastructures?

In this open and informal session, we explore what it really takes to make interoperability work in practice. You share your experiences, hear how others tackle similar challenges, and uncover where researchers often encounter difficulties when choosing or applying standards. You also exchange ideas on how to train and support researchers in making choices that empower interoperability.

You will hear how the Dutch Interoperability Network (DIN) connects communities across domains to address shared challenges, stimulate collaboration and co-develop training. You will also gain insight into SURF’s Innovation Zones (IZ), where collaborative efforts are already working toward a more integrated and user-friendly research-IT ecosystem in the Netherlands by improving access, workflows, and use of infrastructure.

Interoperability is not something you can fix on your own. It takes shared understanding, coordination, and community effort. Join the session to connect with peers, explore collaboration opportunities, and help identify concrete focus areas and actions that DIN and the SURF IZ can take forward. Your input will directly influence what comes next.

Are you looking for a practical way to make your organization ready for the European Health Data Space (EHDS) and ensure that internal datasets are consistently registered and FAIR? Many organizations struggle to keep track of which datasets exist, how they relate to each other, and how to make them transparently available. In this session, you will learn how a structured metadata collection and validation process improves data findability and reusability, which in turn increasing research efficiency and strengthening both internal and external research collaborations.

You will explore how an open‑source workflow based on Data Stewardship Wizard (DSW) guides researchers step‑by‑step through essential metadata elements. You will also see how this process becomes truly effective once it is embedded in internal policy and connected to a findability platform for searchable metadata.

After this session, you will understand:

• How to design a future‑proof metadata registration process that supports your organization’s EHDS readiness.
• Which core elements make datasets consistently discoverable, FAIR, and usable across projects.
• How to combine tooling, workflows, and internal policy into one scalable and transferable approach.

In summary, this session offers a practical blueprint you can use to enhance data discoverability, boost research efficiency, and accelerate your path towards the EHDS era.

Do you wonder how all the different research data management tools and initiatives in the Netherlands actually fit together? Are waiting to see how innovation projects can lead to sustainable research services? Many institutions and funders invest heavily in research data management, yet the landscape still consists of many parallel services, platforms and collaborations. Innovation funding often adds new solutions rather than streamlining this already crowded ecosystem.

In the first part of this session, you will contribute to an open discussion how local RDM initiatives, institutional strategies and national infrastructure relate to each other. You reflect on how the current landscape of tools, consortia and platforms has emerged — and whether it is moving toward a coherent national ecosystem or remaining a collection of individual efforts. In the second part, a panel of representatives at different levels – institutional, national, European – will present their ideas on a national RDM strategy and reflect on input from the audience.
Using examples from national collaborations such as Yoda and Digilab, we try to answer an urgent question: how do we align fixed-term innovation investments with the long-term sustainability of the services researchers depend on?

After this session you will:
• Gain a clearer picture of the current Dutch RDM landscape and the collaborations shaping it.
• Recognise the challenges that arise when innovative data services grow into essential infrastructure.
• Exchange perspectives with peers on how institutions and national partners can better align innovation, coordination and long-term sustainability.

Do you also struggle with missing ORCIDs, incomplete affiliations or conflicting open access metadata?
These gaps affect reporting, evaluation and discovery of research outputs. But where exactly do the problems occur in the research information chain, and what can we do about them together?

In this interactive workshop you will work with the first data-driven monitoring results from the ORI Monitoring Framework, based on use cases collected from Dutch universities and research funders. These results compare metadata quality across systems such as CRIS platforms, OpenAlex and OpenAIRE.

Instead of just presenting results, you will actively help identify interventions to improve open research information.

During the workshop you will:

• explore example dashboards showing gaps in PIDs, affiliations and differences of metadata records within different sources.
• discuss with peers where problems originate in the research information chain
• propose concrete interventions that institutions, infrastructures and publishers could implement

The goal is simple: move from monitoring problems to improving the ecosystem together.
Your input will help shape the next steps in ORI monitoring, dashboard development and community improvement actions for an open research information commons.

Whether you want to optimize computing power, data storage, research information or open scholarly communication, we achieve this using scarce resources (infrastructure, expertise, support, etc.).

During an interactive workshop, we will explore together what these scarce resources actually are, how we can make the best possible use of the scarce resources we have, and what role each of us plays in this. How does an IT manager, a researcher, a research support staff member, see their roels, how do funders of research think, and how does SURF? And what does this actually mean for your role?

Research often involves sensitive data. Ranging from (pseudonymised) medical records to interviews with threatened or persecuted individuals.

During collection and usage, but also after use, we want to store these records securely: protected against unauthorised access from outside, but also within the organisation. Even against sysadmins and dataset access managers. Recent examples of data-breaches at Clinical Diagnostics and ODIDO stress the urge for more solid security and privacy measures.

To achieve this at new levels of privacy and security, University of Amsterdam is developing a new digital archive, using the open source PEP Repository software developed at Radboud University in Nijmegen. The PEP Repository software has been developed and in use since 2016 for several large longitudinal studies on sensitive data, and was designed for this purpose by a group of privacy experts, academic researchers and digital security experts with novel cryptographic privacy enhancing techniques (PETs).

During this presentation, Tim Fierens from UvA’s ICTS will discuss the challenges in implementing the digital archive in a responsible manner. Joep Bos-Coenraad, project leader and co-developer of the PEP Repository, will explain how the privacy-by-design approach enables what was previously ethically impossible.

Do you collaborate daily in tools like Microsoft 365 or Google Workspace? And do you ever wonder how much control the research sector still has over the digital environments where research happens?

In this session you discover how Dutch research and education institutions are exploring an alternative: a sector-wide collaboration environment based on open-source technology. Instead of relying entirely on commercial ecosystems, institutions are experimenting with a shared platform that combines file sharing, document editing, communication and collaboration in one environment.

More than thirty institutions are currently participating in a pilot to test how such an open collaboration platform can work in practice. The goal is not to replace existing tools overnight, but to explore how the sector can gain more strategic choice and influence over the digital infrastructure used for research and education.

During this session you will discover:

• what digital autonomy means in practice for research institutions
• what it takes to build a shared collaboration environment across organisations
• what lessons we are learning from the current multi-institution pilot

Curious whether a sector-driven collaboration platform can really work? Join this session and explore what the future of digital collaboration in research could look like.

Do you use language models in your research or support, but sometimes doubt whether the answer is correct? In this interactive session, you will discover how to quickly and practically check AI results using the SURF AI Hub. 
 

You will start in a secure, sovereign AI environment that runs in the SURF data center and is built on public values such as autonomy and fairness. In a short live demo, you will see how convincing AI can sound when it is wrong: hallucinations, fabrications, and false certainties. You will then practice with simple checks that you can immediately apply in your own work. 
 

You will also discover how small choices in your prompt—such as framing or bias—can have a major impact on what the AI makes seem “true.”

Curious about what SURF does? Join us as we showcase our compute and data services. We offer solutions for processing, storing, managing, sharing, and publishing data, along with high-performance and cloud computing. We'll discuss our portfolio, future vision, current challenges, and upcoming roadmap.

In this session you will first learn about the activities of the FAIR4ChemNL and VHP4Safety projects which have been actively working on making chemistry research output more FAIR.

You will discover how we are exploring solutions from the international chemistry communities, e.g. IUPAC, NFDI4Cat, NFDI4Chem, and from the Dutch research community, like SURF’s Taxila and PeerTube platforms for sharing training events and educational material.
FAIR4ChemNL focuses on chemical catalysis and the VHP4Safety platform focuses on chemical safety assessment.

The session will continue with an open discussion on how IT supporters can best help researchers to make their chemistry data more FAIR.

In this presentation you will learn about practical use cases of applying AI in the domain of HAS green academy.

HAS green academy, with 3.000+ students the biggest HBO green school in The Netherlands, started last year with a new research group. The name of this research group is "Data Intelligence for Sustainable Transitions". The group consists of +/- 10 researchers with a solid background in applied biology, geography, ecology, food technology and agriculture.

With this new group, HAS green academy aims to provide a powerful impulse to research on how spatial data intelligence can deliver added value to the sustainable transitions of our living environment.  These transitions contribute significantly to climate adaptation, energy transition, and biodiversity in both urban and rural areas.
The research group focuses on the core areas Knowledge Modeling, Remote Sensing, Digital Twins, AI and Data & Ethics.

In this presentation, we will explain why high quality research data is extremely important for us. This explains the title “Sowing with Data, Harvesting with AI”. We will also pay attention to the AI Roadmap of HAS green academy + the connection with SURF.

The presentation of Harm and Koen will be interactive and includes a practical demo of an AI use case we recently implemented.

How can researchers working at institutions spanning the globe use the tools and have seamless access to the science outputs of a large and complex radio telescope that generates petabytes of data annually?

In this session the team that has enabled SURF Research Access Management (SRAM) integration and operation for the Low Frequency Array (LOFAR) telescope system will take you on a journey of discovery that will show you how the end-goal of enabling user access transparency of a complicated research infrastructure system was achieved.

You will learn about the the path we took and the decisions we made leading to the current implementation, about our experience with SRAM as well as our day-to-day operations and maintenance of the system.

We will outline some challenges that we have faced interacting with the world-wide scientific community, which is our primary user base.

Finally, you will see what we envision to achieve in the future as we open more and more services to the world.

By the end of the session you will learn what it takes to implement the integration of SRAM within a system like LOFAR and what to expect when you open it up to the public.

Do you wonder which skills are essential for providing hands-on support to researchers during software development, or how to grow those skills in yourself or your team?
This session takes you through the experience of the Digital Competence Centre at TU Delft, where a team of Research Software Engineers (RSEs) and Research Data Engineers (RDEs) has been supporting research software development across faculties for the past five years. You will learn which skills and mastery levels they consider essential for managing research software projects, how the team structures support through a project-based model, and what lessons and recommendations you can take back to your own context.
After this session, you will have a clearer picture of what professional profiles make a support team successful in delivering research software services, which skills matter most for roles like RSE, RDE, and software, and how to think about professional development within teams that provide hands-on research support.
Whether you lead a support team, work in one, or are exploring these roles, you are welcome to join and share your perspective.

Across the Social Sciences and Humanities, many of the most pressing research data challenges are best addressed through collaboration - in the space between infrastructures, research practices, and support communities. The Thematic Digital Competence Centre for the Social Sciences & Humanities (TDCC SSH) was created to surface shared challenges and support collaborative solutions.
Since 2023, the programme has facilitated access to NWO funding, enabling a growing portfolio of projects addressing topics such as synthetic data, audiovisual workflows, sensitive data reuse, multimodal research data, and the responsible use of online public material. What these projects share is not only their technical ambition, but their collaborative nature: they bring together researchers, data professionals, infrastructures, and other research-performing organisations to address issues that no single actor can solve alone.
This session will highlight a selection of projects from the TDCC SSH portfolio, focusing on lessons learned from developing and implementing cross-institutional solutions. Speakers will reflect on how community needs can be translated into practical tools, workflows, and guidance, and how collaboration across institutions and domains can accelerate progress in FAIR data and software practices.
The session will also look ahead. Drawing on experiences from the project portfolio, we will invite the audience to reflect on how these lessons might shape future directions for collaborative initiatives in the SSH domain.
By connecting experiences across projects and institutions, the session aims to spark discussion, inspire new collaborations, and highlight how collective efforts can strengthen the future of SSH research.

Almost 1.5 year into building a national training and community platform for data professionals, this session will get you up to speed on the work done so far, on how it will help you grow as a data professional or strengthen your organisation’s training strategy. You will also hear how you can be involved.

Data professionals are instrumental in the transition to Open Science and the increasing demand for high-quality data. Their role in Dutch research performing organisations (RPOs) is developing rapidly and with it, the competencies required to fulfil the job. To meet the growing demand for well-trained data professionals, Research Data Netherlands (RDNL) develops a national training and community platform. The project to make this happen is funded by Open Science NL and will be running until 2028.

Key components of this platform will be a national curriculum and competency framework for data stewards, relevant training courses offered by various training providers, including a newly developed Train-the-trainer course, learning paths and community profiles to help you identify peers to connect with. But: it only works if it fits your needs, as a data professional, training provider or RPO, that’s why we want to achieve this in co-creation with you.

Struggling to connect the right datasets, models, and computing resources for your AI projects? Discover how we are building a national infrastructure for AI-driven research! We aim to seamlessly link open data, powerful compute, and diverse communities to turn collaboration into something much greater than the sum of its parts.

We present Project Hippocampus, an NWO Open Science project to build a unified Dutch infrastructure for AI. At the core of this project is OpenML—a popular open-source platform where researchers globally share machine learning datasets, algorithms, and experimental results.

While OpenML excels at sharing knowledge, advancing modern AI requires connecting minds across different tools and computing environments. By integrating OpenML with major infrastructures like SURF, DANS, and Hugging Face, we are bridging silos to create a seamless ecosystem. This integrated Dutch platform acts as a shared "collective memory" for AI-driven science, making research highly transparent, reproducible, and accessible.
After this session, you will better understand:

• What OpenML is and how to use it to discover and share AI datasets, pipelines, and experiments.
• How we aim to connect these resources seamlessly with high-performance computational power at SURF.
• How to get involved and collaborate to grow the open science community around OpenML to address future research challenges together.

Target Audience: Researchers, data stewards, and research supporters passionate about collaborative open AI infrastructure.

Data transfer orchestration: any help needed?

Do you need help in copying data, especially large data sets, from a data repository to a compute environment?
Do you want to share data with other researchers and you do not find an obvious way to do it?
Do you experience frustration when you need to first download those data to your laptop and then copy them somewhere else?
Then this session is for you.
You will hear about a new data transfer tool, Neptune, which allows you to orchestrate large data transfers between different storages and data repositories.
It comes with an intuitive Web UI and an API that allows you to easily integrate with other tools.
Are you familiar with SURFfilesender?
Just as SURFfilesender helps you to upload data from your laptop and share them with other users, Neptune helps you to copy data between different remote storages and to share those data with other users.
We will show you how to do all this during an interactive session.

Generative AI has significantly transformed research workflows. Beyond the direct use of LLMs for programming, writing, and reviewing tasks, agentic frameworks can build on these capabilities to support a broader range of research tasks. Such systems are particularly promising for addressing tedious and time-consuming aspects of  scientific work.

Many agentic frameworks and tools have been proposed for both general and domain-specific tasks. However, building such systems from scratch still requires substantial effort. This motivates the development of a flexible template that allows researchers to easily experiment with different designs or build their own tools.

In this talk, we introduce MAST4Sci (Multi-Agent-System Template for Science), a template that provides core components including support for multiple LLM providers (including SURF AI Hub), planning modules, tool integration, and memory management. The template is designed to be easy to implement, enabling researchers to prototype and evaluate new task-specific agentic frameworks or build practical tools without having to implement the underlying infrastructure from scratch.

To demonstrate its capabilities, we present several implementations built on top of this template, including a metadata-from-dataset tool for research data management (developed in collaboration with SURF) and a meta-analysis tool for agricultural data extraction from scientific literature. These examples illustrate how the template supports the rapid development of diverse agentic applications across research domains.

At University of Applied Sciences Utrecht (HU), we are exploring the use of Whisper, an open-source speech-to-text model from OpenAI, for automatically transcribing research interviews and other audio sources. A key advantage of Whisper is that it can be hosted within our own infrastructure, eliminating the need to share sensitive research data with external commercial parties. Integration with SURF Research Drive also allows transcription data to be stored centrally and securely, without duplication.

In this interactive session, we will share our practical experiences setting up a Whisper-based transcription solution within the SURF Research Cloud. At the end of last year, we assessed the need for a transcription tool for researchers within HU University of Applied Sciences. There was a need. In the spring of 2026, we conducted a short proof of concept.

To avoid having to host the model ourselves, we subsequently joined the SURF AI-hub pilot, a potential new SURF service that makes open-source AI models available via an API. Our case study within this pilot is transcribing audio with Whisper.

During the session, we will demonstrate our technical setup, the choices we made, and the lessons learned. We will also discuss the next steps: a broader experiment with researchers.

The session is intended to share experiences and explore together how AI transcription can be deployed safely and practically within practice-oriented research environments.

Latin America has been pioneering a publishing model that could reshape how science is shared — without paywalls, article processing charges and commercial publishers. If you're curious about what that means for your own research practice, this session is for you!

You'll discover how Diamond Open Access, or community-driven open access, grew organically in the Global South out of necessity, and why this path looks very different from that in the Global North. You'll also explore how the publish-review-curate (and its many flavours) is providing alternative pathways to recognition and scholarly communication — and what that could mean for early-career researchers seeking meaningful ways to learn and contribute.

By the end of this session, you will:

• Understand what Diamond Open Access and publish-review-curate actually look like in practice
• Have space to reflect on what it would take — for you, your discipline, or your institution — to make these models the norm

This is a hands-on session. After a short introduction, you'll get to share your own perspective. What would need to change in your field? What's holding you back? Your input will directly help map where the Dutch research community stands on these shifts — and where the real opportunities lie.

AI Technology is infiltrating the full tech stack of research workflows. Research is being done on AI, with AI, by AI, within existing software, through the creation of new software, on existing hardware and on new kinds of infrastructure.

Researchers are expected to and wish to act responsibly, however clear guidance is very hard to give in this very rapidly evolving landscape of AI technology.

We have developed a presentation that looks at responsible AI within the context of pushing research boundaries through the metaphor of “Free soloing”, i.e. rock climbing without the typical safety protections provided by climbing gear.

This presentation is being used within the university to facilitate open discussions on AI literacy, AI infrastructure, and specifically responsibility and risk awareness. Targeting questions such as: what are existing and emerging risks in the AI tech stack, whose responsibility is it to be aware of these risks and whose responsibility is it to establish or provide the mitigations in different kinds of research (and educational) environments?

At Surf Research Day we wish to bring some archetypical cases that have emerged from these sessions within the university. We will present these for an open discussion on responsible AI, the role of the researcher in taking responsibility and the role of service and AI infrastructure providers to take co-responsibility in providing (components of) the tech stack to do research on AI and with AI, or even by AI.