1st International Conference on Human Centric AI – ICHCAI 2026

Signe Riemer-Sørensen, Research Manager for Analytics and AI at SINTEF Digital

Title: Human-Centric AI: Rethinking What AI Should Deliver?

Abstract: AI is advancing fast, but our ability to use it for real decision making is not. Across sectors, we see high expectations, limited maturity, and a persistent gap between what AI can do and what our most critical decisions require. It is time to stop and reflect if we are focusing on the right technologies and the right problems. Examples from healthcare to energy systems demonstrate the need for shifting from pattern‑finding AI to decision‑relevant human-centric AI with integrated domain insight, uncertainty and safety.

Signe Riemer-Sørensen is Research Manager for Analytics and AI at SINTEF Digital, and Senior Research Scientist in machine learning for industrial applications including hybrid AI (data driven modelling of physical systems) and machine learning for decision processes. Her primary research focus is overcoming challenges for implementing machine learning and artificial intelligence in across diverse industrial settings where physics play a role and data is less than perfect. She is co-director for the Norwegian Center on AI for Decisions with 13 research partners and more than 60 financially supporting partners from industry and public sector.

Synnøve Olset, CEO & Founder of Mynder

Title: Humans in the Loop: Who Decides What AI Agents Do in Regulated Compliance Infrastructure.

Abstract: As AI agents increasingly operate within regulated infrastructures, the question is no longer whether they can act autonomously, but who defines, constrains, and defends their decisions. This talk explores the emerging role of humans in the loop within agent-native compliance systems through the case of Mynder, a European compliance and governance platform where AI agents perform tasks traditionally handled by consultants. The presentation introduces “Defensible AI agents” — agents designed not only to support organisational compliance, but also to remain explainable and auditable under regulatory scrutiny. Through practical lessons from deploying a Multi-Agent Control Framework (MACF) and an MCP-based agent platform, the talk examines how governance, autonomy boundaries, auditability, and stakeholder negotiation are operationalised in real-world AI systems. The session offers both conceptual insights and practical reflections on building trustworthy human-centered AI infrastructure for regulated environments.

Synnøve Olset is a technology entrepreneur, innovation leader, and CEO & founder of Mynder. With experience in technology and digital innovation since the late 1990s, she has worked across industries building solutions that simplify complexity and create human-centered impact. Her work spans company building, large-scale innovation programs, and pioneering initiatives such as Norway’s first healthcare chatbot. Today, she leads the development of AI-powered compliance infrastructure that automates and continuously monitors privacy, information security, and risk management for regulated organisations. Her work focuses on combining technological insight, creativity, and human imagination to build trustworthy AI systems for the future.

Cagri Erdem, postdoctoral fellow in the Department of Informatics at the University of Oslo.

Title: Humans <> Artifacts; Art <> Science: Objects Emerge from Relations

Abstract: Design has long been shaped by efforts to center the human, from human-centered to participatory approaches that emphasize needs, values, and oversight. Though essential, these approaches rely on a relatively stable distinction between humans and the systems they design and use; an assumption that is becoming increasingly difficult to sustain. Drawing on work in creative AI, particularly in sound and music computing, I argue that humans, artifacts, and environments take form through evolving relations. From this perspective, such distinctions cannot serve as fixed starting points; they are continuously drawn and reworked in practice. What we recognize as objects, then, can be understood as some coherent mental entities emerging from these relations. In this talk, I will develop this view through a series of systems and practices in which rhythmic structures, new interfaces for musical expression, and generative models are used not as fixed tools, but to expose, manipulate, and reflect on the relations through which they operate.

Çağrı Erdem (he/him) is a programmer–composer and improvisor. He develops much of his music alongside musical human-computer interaction research. His previous work focused on creating biosignal-based musical interfaces and explored various forms of human-AI collaboration in performance and composition. As a postdoctoral fellow in the Department of Informatics at the University of Oslo, he continues his research in multimodal interaction techniques, exploring the entanglement of relations, trajectories, and networks.