In Norway, public institutions, academia, and industry should actively work to promote diversity and increase the visibility of underrepresented groups in AI and technology. This year’s event, “High-Impact AI: When Algorithms Shape People’s Lives” brings together diverse perspectives on how AI is influencing education, healthcare, communication, and society.
The program will address inclusive and ethical uses of generative AI in early childhood education, the development of culturally and linguistically adapted AI for Indigenous and minority communities, and the challenges of fairness, bias, accountability, and governance in healthcare and other AI ecosystems. It will also explore how assistive AI can expand communication and participation while raising important questions about autonomy, authorship, consent, and human agency.
Through these perspectives, the event will examine both the opportunities and risks of high-impact AI and discuss how AI can be developed to support equity, inclusion, and meaningful participation.
About "Diversity in AI"
The Diversity in AI Event is organized as a side event of the Northern Lights Deep Learning Conference (NLDL). NLDL is organized by SFI Visual Intelligence and hosted by UiT The Arctic University of Norway. The Diversity in AI Event is organized by PhD fellows from the Visual Intelligence Graduate School, in collaboration with NORA.ai (Norwegian Artificial Intelligence Research Consortium).
Jianying Hu
IBM
Lars Ailo Bongo
UiT The Arctic University of Norway
Maria Dardanou
UiT The Arctic University of Norway
LaVonne Roberts
Scott-Morgan Foundation
& SMF Labs
Date: January 12th 2027
Time: 15:15 - 17:15 CET
Place: UiT, The Arctic University of Norway, Auditorium 1 in Teorifagsbygget Hus 1 Floor U1.
Program:
15:15 - 16:15: Presentations
16:15 - 17:15: Panel discussion
Jianying Hu, IBM Fellow, Director HCLS Research; Global Sci Leader, AI for Health; Exec. Partner for Cleveland Clinic
AI fairness discussions have traditionally focused on bias in predictive models and unequal performance across demographic groups. However, the emergence of foundation models, AI agents, and scientific copilots is expanding the challenge beyond individual algorithms.
This talk explores how bias is evolving from a model-centric problem to an ecosystem-level concern that spans human-AI workflows, healthcare systems, and societal contexts, and examines new approaches to evaluation, accountability, and governance.
Maria Dardanou, Professor at UiT The Arctic University of Norway
Generative AI can shape young children’s experiences through educators’ choices about stories, images, and learning activities. This raises questions about whose experiences are represented, which biases are reproduced, and how children’s voices and agency are protected.
This talk introduces GENIAL, a European Erasmus+ project coordinated by UiT that supports ethical, inclusive and pedagogically meaningful approaches to generative AI in early childhood teacher education.
Drawing on the project’s ongoing development of an ethical and pedagogical framework and practical learning resources, it explores opportunities for creativity and inclusion alongside concerns about bias, privacy, and professional judgement. It asks what future educators need to know—and be able to question—when deciding whether and how to use AI.
LaVonne Roberts, CEO of the Scott-Morgan Foundation and Co-founder of SMF Labs
Can AI make us better thinkers without quietly deciding what we think?
AI can allow someone whose speech or movement has been severely restricted to communicate, work, and participate in ways that weren’t previously possible. But when AI proposes the words, it can also shape which thoughts are easiest to express.
Imagine someone intending to say, “No, I disagree,” and being offered three polished responses that never actually say no. Choosing the closest one may be the only practical way to remain in the conversation. But what does that choice tell us about authorship?
Drawing on our work at the Scott-Morgan Foundation and my essays at SMF Labs, I’ll examine authorship drift, the distinction between assistance and manipulation, and how AI can gradually shape the way we express ourselves.
I’m also interested in a larger question: Does better performance with AI mean we’re becoming better thinkers, or simply getting better results from machines? The goal isn’t to have AI do less. It’s to design systems that expand our capacity to communicate, learn, reason, and disagree without quietly substituting the machine’s version of what we mean.