New AIPEX Publication – AI data labelling: a pathway or peril to refugee self-reliance?

Mariia Lesina (Lund University, extern at the AI Policy Lab)

Abstract

Humanitarian budgets are collapsing at the same moment that AI systems are generating major demand for high-quality human-labelled data. Considering flexibility and low entry requirements of AI data labelling, social enterprises have moved quickly to position refugees at this intersection, framing this digital work as a scalable, dignified pathway to economic self-reliance. Drawing on organisational documents, social enterprise impact reports, and first-hand worker testimonies, the policy brief examines whether data labelling constitutes a meaningful livelihood solution for displaced populations or whether it reproduces new forms of digital precarity within global AI supply chains.

The analysis focuses on four interrelated dynamics: the dual identity of social enterprises, which pitch their impact-oriented services to tech clients while selling the same work as empowerment to humanitarian donors; the self-reliance narrative promoted by social enterprises; and the lived realities behind this narrative, including both the positive aspects of giving refugees access to AI work and structural labour harms embedded in this labour. While social enterprises are providing hope and some insulation for refugees entering the labelling market, they cannot overcome structural harms of the market, leaving refugees to bear the cost of that failure. This paper recommends that UNHCR publish binding operational guidelines for social enterprises, drawing on the Fairwork Framework, to ensure that AI-related digital work can support refugee livelihoods without entrenching the very exploitation it claims to address.

Keywords: humanitarian support, refugees, AI data labelling, social enterprises, self-reliance

…

Read the full article here

New AIPEX Publication – Arbitrary cognitive offloading to GenAI: Does the current policy landscape account for the right to quality education of children and youth in the European Union?

Marit Brademann (University of Edinburgh, extern at the AI Policy Lab)

Abstract

General-purpose generative AI (GenAI) is increasingly used by young learners in the European Union, with 20% of individuals aged 15–29, approximately 15 million people, relying on it for educational tasks such as information retrieval, feedback, or full task delegation. While cognitive offloading is a natural part of human learning, GenAI’s ease of use risks disrupting efficient learning processes, potentially depriving children and youth of their right to quality education, including foundational skills like literacy, numeracy, and critical thinking.

This paper argues that the unmediated use of GenAI in educational settings undermines the right to quality education by interfering with cognitive development. It examines the role of cognitive load, metacognition, and cognitive offloading in learning, linking these concepts to GenAI’s potential to disrupt memory consolidation, problem-solving, and critical thinking. The analysis evaluates the current European policy landscape, focusing on the EU AI Act, OECD Digital Education Outlook 2026, and the European Parliament CULT Committee’s 2026 briefing to assess whether existing hard and soft policies address the cognitive risks of GenAI. While these policies recognize risks, they primarily target AI-enhanced educational technologies rather than general-purpose GenAI, leaving a regulatory gap.

The paper concludes that arbitrary cognitive offloading via GenAI jeopardizes the normative goals of quality education, as defined by frameworks like UNICEF’s and Bloom’s Revised Taxonomy. It calls for a revision of the EU’s normative approach to general-purpose GenAI, emphasizing the need to protect young learners’ right to cognitive development and autonomy.

Keywords: cognitive offloading, metacognition, AI in education, right to education, metacognitive laziness, OECD, EU AI Act, CULT, AI and Human Rights, Governance and Compliance, Human Well-being

…

Read the full article here

Our new piece on Tech Policy Press – The UN Scientific Panel on AI’s Preliminary Report Does Not Establish Its Independence

See our piece in Tech Policy Press on the importance of establishing the independence of the United Nations International Independent Scientific Panel on AI, and our recommendations for how they can do so 👇

Written with Virginia Dignum, Rachele Carli, Petter Ericson & Tatjana Titareva at the AI Policy Lab @Umeå University

About Tech Policy Press

Our goal is to provoke new ideas, debate and discussion at the intersection of technology, democracy and policy, with a particular focus on:
• Concentrations of power: the interaction of tech platforms, governments and the media and the future of the public sphere;
• Geopolitics of technology: how nation states approach technology in the pursuit of advantage;
• Technology and the economy: the relationship between markets, business, and labor;
• Racism, bigotry, violence oppression: how tech exacerbates or solves such challenges;
• Ethics of Technology: how technology should be viewed alongside existing democratic ethos, especially with regard to privacy, surveillance and personal freedoms;
• Election integrity participation: mechanisms of democracy, problems such as disinformation and how citizens come to consensus.

Website

New AIPEX Publication – Question Zero: Why Responsible AI Begins Before AI Adoption

Tatjana Titareva (AI Policy Lab, Umeå University), Jason Tucker (Institute for Futures Studies & AI Policy Lab, Umeå University), Rachele Carli (AI Policy Lab, Umeå University), Viktoriia Movchan (AI Policy Lab, Umeå University), Virginia Dignum (AI Policy Lab, Umeå University)

Abstract

The Question Zero (Q0) Self-Assessment Tool for Responsible AI, developed by the AI Policy Lab at Umeå University, supports organisations in posing foundational questions before adopting AI. Grounded in the concept of Question Zero: “Under what conditions should an AI system be adopted, if at all?”, the tool offers support for cross-functional team discussions covering motivation, stakeholder mapping, system type, adoption process and infrastructure. It is designed for public institutions, civil society organisations, policymakers and other actors looking for responsible AI decision-making approaches. With the help of the tool, we argue that purposeful, problem-led assessment must precede any procurement and deployment decisions.

Keywords: Question Zero, Q0, Responsible AI, AI Self-Assessment, AI Governance, EU AI First Strategy, AI Policy, AI Adoption Motivation, Stakeholder Mapping, Explainability, AI Policy Lab, Umeå University

The Problem with Starting from the Answer

Public and private organisations face growing pressure to adopt artificial intelligence (AI). Governments across the EU and beyond are committing vast public resources to AI acceleration. Organisations, large and small, feel urged to integrate AI into their operations for fear of being perceived as falling behind. The European Commission’s Apply AI Strategy (European Commission, 2025) has formalised this pressure into policy direction, framing AI as the default first response to organisational and societal challenges. The same trend can be seen at a national level, as well as across sectors and organisations.

…

Read the full article here

Book Club discussion on “The AI Paradox: How to Make Sense of a Complex Future”

On Friday at AI Policy Lab @Umeå University we had a book club discussion on “The AI Paradox: How to Make Sense of a Complex Future” by Professor Virginia Dignum.

We explored the book’s eight AI paradoxes and reflected on what they mean for education, governance and society.

🧠 One discussion focused on the Intelligence Paradox: The more AI can do, the more it highlights the irreplaceable nature of human intelligence. Rather than asking whether AI is intelligent, we considered a Competence Paradox. What is a system actually capable of? How robust, trustworthy and reliable is it?

🎓 Education sparked one of the longest discussions.
Generative AI creates a new educational grey zone. It can support learning and reduce unnecessary workload. It can also create an illusion of competence, shallow learning and what one participant called metacognitive laziness.
A key idea was productive friction. Learning should not be effortless. The challenge is not to remove struggle, but to ensure that struggle leads to understanding.

This raises a difficult question: should education move away from assessing products and focus more on assessing processes, reasoning and reflection?

👩‍🎓 Another important point was that students should be part of the solution. They often understand how these tools are used in practice better than institutions assume.

⚖️ The Solution Paradox: Solving problems with technology often creates more problems, lead to an interesting conversation. New tools alone do not transform education. Smartboards did not. Computers did not. AI will not either.
Technology without training, support, strategy and institutional change risks becoming an expensive distraction.

🏗️ We also touched upon the material side of AI. Discussions about responsible AI often focus on models andc ompaniex, and pay less attention to energy use, exploitation of labour and governance.
Responsible AI cannot be only about models and companies. It must also address the systems, power structures and resources that make AI possible.

⏳ Another theme was speed. In AI, faster is often assumed to be better. But is it?
AI forces urgent decisions, yet responsible decisions often require time. Perhaps moving more slowly can sometimes lead to better and more democratic outcomes.

🌍 Some of the takeaways were:
– AI is not something that is happening to us. It is something we are actively building, shaping and governing.
– That means responsibility sits with researchers, educators, institutions, companies and policymakers alike.
– Who gets access to powerful AI systems? Who benefits? Who bears the costs? And who takes responsibility for shaping the future of technology and society?

🙏 Thank you to everyone who joined the discussion and shared their perspectives. Interdisciplinary conversations like these remind us that the future of AI is not only a technical challenge. It is a human one.

Virginia Dignum receives 2026 Nordic DAIR Awards Lifetime Achievement Award in AI

On May 7, 2026, Virginia Dignum, Director of the AI Policy Lab and Professor of Responsible Artificial Intelligence at Umeå University, has been named the 2026 Nordic DAIR Awards Lifetime Achievement winner in AI!

The DAIR Awards, Data and AI Readiness Awards, recognize achievements in data, analytics and AI across the Nordic region. In 2026, the awards are integrated into the Data Innovation Summit in Stockholm, bringing recognized work in AI and data directly into one of the region’s major meeting places for practitioners, leaders and innovators.

This year’s awards focus on maturity and real-world impact in AI and data.

Against this backdrop, Virginia’s recognition highlights her long-standing contribution to responsible AI, AI ethics and AI policy. Her work has helped shape international discussions on how AI can be developed and governed in ways that place human values, accountability and societal benefit at the center.

In its award citation, DAIR writes:
“There are few individuals whose work has shaped the ethical and technical landscape of AI as profoundly as Virginia Dignum. As a world-renowned researcher and a leading voice in Responsible AI, Virginia has spent her career ensuring that as we build more powerful systems, we do so with human values at the center.”

At the AI Policy Lab, we are proud to see Virginia’s work recognized in this way. Her leadership continues to inspire researchers, policymakers, students and partners working toward responsible and trustworthy AI.

Warm congratulations, Virginia!

Could not load the image

1 / 2

About DAIR

The DAIR Awards (Data and AI Readiness Awards) recognize organizations that lead the way in using data, analytics, and AI to drive measurable business and societal impact. Focused on organization achievements, the awards highlight companies that demonstrate strategic vision, innovation, mpact and maturity in their data, analytics and AI practices. Through the recognition of real-world success stories, the DAIR Awards aim to accelerate the adoption of data-driven technologies, inspire others to follow best practices, and benchmark progress across the Nordic region’s most advanced organizations.

Learn more

Workshop on Question Zero: Beyond the ‘AI First’ Hype

On March 12, 2026, the AI Policy Lab at Umeå University team conducted the workshop “Question Zero: Beyond the ‘AI First’ Hype” during the Winter School on Ethical, Legal, and Societal (ELS) aspects of AI and ASat Umeå University.

Before you adopt AI, ask the right question first. Not “Which AI should we use?” But: “Under what conditions should an AI system be adopted, if at all? “That is Question Zero (Q0). We live in an era of AI hype. Governments are pouring huge resources into AI acceleration. Organisations are rushing to adopt. But speed is not a strategy. And technology is not destiny.

Winter School on Ethical, Legal, and Societal (ELS) aspects of AI and AS

Could not load the image

1 / 4

The Dutch childcare benefit scandal shows what happens when we skip Question Zero. An algorithm accused tens of thousands of innocent parents of fraud, destroying jobs, families and lives. Q0 was never asked.
Next up: figuring out what situated AI looks like not just as critique, but as practice, that is, research that is itself accountable to the communities it studies.

Q0 is a practical, free assessment tool developed at the AI Policy Lab at Umeå University with five categories of questions. Below we list some of the questions under each category (see the full version of the tool – below).

WHY? Motivation

Why do you plan to adopt an AI system?
What problem(s) is your organisation trying to solve with a new AI system?
What are the available alternatives, incl. human, other technical non-AI solutions, etc.?

WHO? Stakeholders and inclusion

Which stakeholders could benefit if the AI system is adopted and how?
Which stakeholders could potentially experience any risks/harms after adopting the AI system and how?
Does the AI system offer an opt out option for all impacted stakeholders?

WHAT? Type of AI system

What type of AI system are you planning to adopt?
How does this choice match the specific problem your organisation aims to solve?

HOW? Adoption and governance

How do you plan to monitor/analyse the new AI system’s outputs and performance?
How will you ensure the security of your organisation’s and your clients’ data?

WHERE? Infrastructure and control

Where does the training data originate from?
Where will the AI system run and data be stored?
Where is the AI system’s provider based?

During the workshop, participants worked in groups and applied the Q0 assessment tool to realistic AI adoption scenarios, including:

an AI system for emotional music personalisation on streaming platforms
automated hiring screening systems used in recruitment
workplace analytics tools analysing employee activity and productivity
AI systems for prioritising drug discovery in pharmaceutical research

Q0 is not anti-AI. It is pro-thinking. Technology is a human endeavour. We create it. We shape it. We can choose differently.

Download Q0 Assessment Tool v3

Draft – March 2026

Download PDF

If you have questions or comments about the Q0 tool, feel free to reach out to us via: contact@aipolicylab.se.

Yearly Research Retreat with AI Policy Lab @Umeå University and the Responsible AI group

Dates: 17-20 March 2026
Format: Responsible AI Retreat

Just back from our yearly research retreat with AI Policy Lab @Umeå University, the Responsible AI group at Department of Computing Science and colleagues from different places.

Our theme this year was Situated AI: grounding AI research in place, community, and lived knowledge rather than a view from nowhere that mascarades as objectivity.
We talked about solarpunk visions for AI at community scale: whose resilience, whose future, built on whose knowledge? We sat with the uncomfortable truth that participation can be co-opted, that inviting more voices into a process doesn’t redistribute power, and can even become a new form of data extraction.

And we turned the lens on ourselves. The publish-or-perish pressure of academia doesn’t just shape what gets said, it shapes who gets to say it, and on what timeline. The incentive structures of academic AI research can reproduce the very dynamics we critique from the outside.

Our working conclusion, borrowed from Donna Haraway: “stay with the trouble”. Sometimes not resolving tensions prematurely, but staying in them long enough, is the most honest thing we can do.

Next up: figuring out what situated AI looks like not just as critique, but as practice, that is, research that is itself accountable to the communities it studies.

Could not load the image

1 / 8

AI Policy Lab Day 2025: Highlights and Reflections (Recording Available)

Date & Location: November 19, 2025, Umeå University, Västerbotten, Sweden

The AI Policy Lab Day 2025 was rich in insight and exchange.

Sennay Ghebreab delivered a keynote that grounded Question Zero in lived experience, reminding us that the decision to use, or not use AI is never a static checkpoint. He urged us to think in terms of Question Infinity: a continuous, reflective process in which risks and opportunities are held in tension rather than framed as opposites.

Daniel McQuillan‘s talk added a powerful systemic lens. By framing contemporary AI as a product of deeper structural failures, he challenged us to confront the material and social realities beneath technological optimism. His proposal of decomputing, a combination of degrowth, conviviality, and care, called us to imagine responses that prioritise collective well-being over speed or scale.

Our researchers’ posters reflected a striking level of maturity. Their work is rigorous, thoughtful, and already influencing wider debates on responsible AI. It was encouraging to see how confidently they engaged with participants and how deeply their projects connected to real societal needs (Rachele Carli, Petter Ericson, Jason Tucker, Tatjana Titareva, Themis-Dimitra Xanthopoulou, PhD Mattias Brännström).

Throughout the afternoon, participants brought curiosity, openness, and an eagerness to engage in discussions and informal exchanges between sessions.

The evening screening of Humans in the Loop added an emotional and narrative dimension that tied the day together. The dramatized story, rooted in the real experiences of data workers in India, wove together the daily realities of annotation labour with local culture, personal aspiration, and the power of lived experience. It captured the invisibility of this global workforce while honouring their agency and resilience. The discussion that followed made clear how crucial these perspectives are for any serious conversation on responsible AI.

A full day of insight, critical dialogue, and shared commitment.

Recordings

Slides

Could not load the image

1 / 2

Film Screening and Discussion: Humans in the loop

Close out AI Policy Lab Day with a special screening of the acclaimed independent documentary Humans in the Loop, a powerful portrait of a young data annotator navigating the rapidly shifting AI industry in India.

Date: Wednesday, November 19
Time: 19:30 to 21:30 CET
Location: Rotundan, Universum, Umeå University

Participation is free but registration is required and places are limited.

Film Screening

A groundbreaking 72-minute Hindi-Kurukh film follows Nehma, an Adivasi woman from Jharkhand’s Oraon tribe who trains AI systems as a data labeller. Director Aranya Sahay (FTII) was inspired by journalist Karishma Mehrotra’s exposé, revealing how over 70,000 Indians – mostly rural women – form AI’s invisible workforce.

A striking, human-centered view of AI from the ground up.

The 72-minute film will be followed by a discussion on the hidden role of data workers in AI.

Snacks and warm drinks will be available!

? Watch trailer & learn more

This film screening is a part of the AI Policy Lab Day programme.

Webinar: How can we soften the blow for the public sector when the Gen-AI bubble bursts?

Date: 8 December 2025 | 17:00-18:30 CET
Format: Online panel featuring short talks, followed by an open discussion

About the Workshop

With significant public investment and political capital currently riding on AI, particularly generative AI, the socio-economic and political consequences of the hype bubble bursting will be profound. This would be a fork in the road for states, and state authorities who have been championing and adopting GenAI. These actors can either change course, and seek new ways to tackle societal challenges, or continue to implement sub optimal and potentially harmful applications using GenAI. Given that many states have aligned with the techno-solutionist discourses and have framed AI adoption in terms of geopolitical positioning, the latter is more likely.

To prepare for this, and mitigate its potential harms, the workshop will focus on the organisational, technical, and social tools we can develop in advance to cushion the societal impacts of the GenAI bubble bursting. In doing so, we aim to preserve institutional legitimacy, redirect existing AI investments toward salvaging public benefit, and maintain old, and open new, avenues for AI development that aligns with the public interest. We will do so by focusing on a range of scales, from the geopolitical to the local.

We invite participants to reflect on how a range of stakeholders, such as governments, civil society, and academia, can respond to the decline of GenAI in ways that promote resilience, accountability, and long-term public value.

Panel discussion between

Virginia Dignum, Professor in Responsible AI, Director Policy Lab, Department of Computing Science, Umeå University.
Gary Marcus, Scientist, author and entrepreneur, known as a leading voice in AI. Six books including The Algebraic Mind, Rebooting AI, and Taming Silicon Valley; NYU Professor Emeritus.
Wendy Hall, Regius Professor of Computer Science at the University of Southampton and Director of the Web Science Institute. A pioneer in AI policy and web science, she co-chaired the UK Government’s AI Review and now serves on the UN’s High-Level Advisory Body on Artificial Intelligence
Gry Hasselbalch, Danish author and scholar specialising in the politics and power dynamics of technology, with a focus on data, AI ethics, and the historical forces shaping technological development.
Joshua Gans, Professor of Strategic Management, at the University of Toronto; economist who studies innovation, entrepreneurship, and business strategy and author of The Prediction Machine.
Frank Dignum, Professor in socially-aware AI, Department of Computing Science, Umeå University, Director of Umeå University’s research center on Transdisciplinary AI for the Good of All (TAIGA).

Moderator:
Jason Tucker
Adjunct Associate Professor at the AI Policy Lab, Umeå University and Researcher at the Institute for Futures Studies.

Participation

The workshop will run for 90 minutes, combining short expert talks with an open discussion.
Participation is open to anyone interested in the societal and policy implications of AI, whether you work in government, academia, civil society, or simply want to join the conversation.

Register here to reserve your place.

AI Policy Lab Day 2025

Date: Wednesday, November 19
Day Programme: 12:30 to 16:45 CET (free lunch and fika included)
Evening Programme: 19:30 to 21:30 CET
Location: Galaxen and Rotundan, Universum, Umeå University (see location in mazemap)

Join us on Wednesday, November 19 for AI Policy Lab Day 2025 – an interactive afternoon showcasing the lab’s work on responsible AI and engaging participants in real, practical conversations. The event features highlights from the AIPL’s projects, a keynote on Question Zero in AI and a hands-on clinic where we invite your toughest AI policy questions.

Whether you’re a policymaker, researcher, student, or practitioner, this is a space to learn, challenge ideas, and share insights across sectors.

Keynote speaker: Prof. Sennay Ghebreab (Amsterdam University)

Participation is free but registration is required for logistics and Zoom access: Registration Form

Zoom participation is available for the keynote, talk and AI Policy Lab (AIPL) introduction.

Agenda

12.30 – Lunch & networking

13.00 – Welcome & AI Policy Lab presentation
Updates & insights from the Lab’s ongoing work by Virginia Dignum, Director of the AI Policy Lab

13.15 – Keynote: Prof. Sennay Ghebreab (University of Amsterdam)

Title:

Abstract:

Rethinking Question Zero in AI

Question Zero – the question of whether AI is the right answer to a problem – has become more prominent in discussions of Responsible AI. This is encouraging, since AI is often adopted as a quick fix, sometimes with harmful consequences for people and the environment. At the same time, there is a growing risk that organizations and governments may misuse Question Zero as a reason not to apply AI in cases where it benefits people and the environment. In this talk, I will explore cases from the Dutch context that highlight different ways of engaging with Question Zero and discuss why it is worth rethinking how we approach this question in AI.

14.30 – Poster Presentations

15:15 – Talk by Dan McQuillan

Title:

Abstract:

16.00 – AI Policy Café – Drop-in Clinic

Explore research from Lab members Topics: AI and human rights, healthcare, education, anti-capitalist perspectives, transparency & explainability

Decomputing

This talk will characterise contemporary AI as the broken product of an already-broken system. While AI sucks more of everything into its cycle of simulated solutions, it diverts us from the underlying structural and environmental crises. By focusing on energy as the conjunction of materiality and hype at the heart of the AI question, the talk will outline ‘decomputing’ as possible response to technocratic nihilism. Decomputing combines degrowth and conviviality into a policy for post-collapse liveability, where the cybernetic may still find a place in support of the common good.

Bring your real-world AI policy or ethics cases, questions, or dilemmas.
Our team will offer ideas, support, and resources in an open, informal format. Think: a cross between a helpdesk, repair café, and fika table. No preparation is needed.

16:45 – Wrap-Up

Announcing the AI Policy Lab’s Drop-in Clinic:

A space to discuss real-life cases in AI policy and ethics.

Got an AI policy or ethics challenge or practical implementation questions? Our team will provide pointers, support, and fresh ideas to help you move forward.

Who this event is for: Policymakers, practitioners, researchers, students, or anyone facing AI-related challenges
What to bring: Any question, case, or challenge related to AI governance, ethics, or implementation
What you’ll get: Practical advice, connections to resources, and space to explore solutions together

Evening Program: Film Screening

Film Humans in the Loop
Rotundan, Universum, Umeå University
19:30-21:30 | Wednesday, November 19

A striking, human-centered view of AI from the ground up.

The 72-minute film will be followed by a discussion on the hidden role of data workers in AI.

Snacks and warm drinks will be available!

Registration is required and places are limited.

? Watch trailer & learn more

We look forward to welcoming you to the AI Policy Lab Day at Umeå University!