Hybrid Intelligence (HI) at the University of Oulu in Finland is accelerating the co-evolution of humans and AI to enhance human agency.
Current approaches to artificial intelligence (AI) remain narrow and have yet to realize its full transformative potential. With hybrid intelligence (HI), researchers are looking toward a future where the relationship between humans and AI is synergistic rather than adversarial.
“Without HI, the danger is that AI may shape human behavior even more than humans shape AI, and not in a direction that fosters human development,” says Professor Sanna Järbela, leader of the Hybrid Intelligence (HI) program at the University of Oulu, which receives funding from the Finnish Research Council through the Profi7 initiative.
Hybrid intelligence: Positioning AI as a partner through interdisciplinary research
The HI program emphasizes a paradigm shift and positions AI as a partner to strengthen human agency.
“There is a growing recognition that future intelligence will not come from machines or humans alone, but from mutual understanding, adaptation and co-evolution,” Professor Jarbela continued.
Oulu’s unique position in shaping the human-AI symbiosis lies in its truly interdisciplinary approach. Artificial intelligence, learning sciences, health sciences, psychology, and ethics are at the heart of designing new technologies. Four strategic research themes ensure a comprehensive approach.
Data and Algorithm-Assisted AI: Advances in Multimodal Interpretation, Behavioral, Sentiment Analysis, and Explainability
The theme Data and Algorithm-Assisted AI establishes the data and algorithm infrastructure for a new generation of socially conscious AI systems. In addition to deciphering, models can also express emotion and cognition, and facilitate human-AI interaction and collaboration.
The large-scale, realistic datasets developed reflect how people feel and behave in real-world situations. These datasets help ensure that AI models learn from natural and diverse human experiences, as well as controlled laboratory environments such as SMG (Spontaneous Micro Gestures for Stress Analysis) and iMiGUE (Unidentified Micro Gestures at Real Press Conferences). These capture naturalistic human behavior and provide the first wild corpus for understanding microgestures and emotions.
Advances in large-scale multimodal models (such as AffectGPT and EmoCapCLIP) extend emotional understanding and narrative reasoning by learning from natural language captions and multimodal behavioral cues, moving beyond fixed emotional labels to more cognitively-based representations.
Professor Guoying Zhao, deputy leader of the HI program, said: “At its core, HI is focused on the co-evolution of human learning and computational models, advancing the data and algorithmic foundations needed for bidirectional understanding between humans and AI systems. We are developing a scalable, interpretable, and ecologically grounded framework that connects perception, emotion, and reasoning across modalities.”
Further advances come from open vocabulary multimodal emotion recognition that adds causal relationships and narrative explanations, aligning AI inference with human cognitive interpretation of emotions. The effort also prioritizes privacy (encryption of facial ID) and preference-tailored interfaces (advanced conversational avatars), providing interpretable narrative explanations that build trust, transparency, and co-adaptive interactions.
Human understanding in AI interactions: Studying learning and human agency in AI-rich environments
Under the theme “Human Understanding in AI Interactions,” hybrid intelligence enables humans and AI to work together in complementary ways, enhancing human agency, learning, and adaptability rather than replacing human capabilities.
“By providing transparent feedback, prompts, and just-in-time support, hybrid intelligence helps individuals maintain metacognitive control over cognitive, motivational, and emotional processes in complex environments. It scaffolds self-regulated learning skills and enables learners to monitor, adjust, and strategically adapt their learning in ways essential for lifelong learning, upskilling, and reskilling,” said Professor Hanna Jarvenoja, leader of the theme.
The team created operational hybrid intelligence systems designed to maintain and enhance human control, including metacognitive AI agents (MAIs), embodied GenAI agents (KAIs), and adaptive nudge tools that support learner agency. These systems are tested in immersive virtual reality and mixed reality environments, including medical simulations, to provide regulatory insights under realistic and demanding conditions.
Augmented Human Mind in Multi-Reality: Integrating Augmented Reality and Virtual Reality to Augment Cognition
With the theme of the human augmented mind in multi-reality, augmented reality (XR) technology allows humans to use and interact with digital guidance embedded in the physical environment, supporting the execution of complex tasks and enhancing real-time perception, memory, and decision-making. Advances in computer vision and spatial AI are making augmented reality systems increasingly adaptable, responsive, and context-aware in real-world environments. New neurobehavioral techniques make it possible to quantify and understand human perception and cognition in virtual worlds.
“Research on presence and human-virtual agent interaction is deepening our understanding of how cognition, embodiment and social cognition work within immersive virtual spaces,” explained Professor Janne Heikkila, leader of the subject.
Strengthening sustainable effects on quality of life: Applying HI to healthcare, education, and lifelong learning
The Enhanced Sustainable Impact on Quality of Life theme focuses on human-AI collaboration to maintain professional agency and health.
“Hybrid intelligence in medical education must prioritize human agency, ethical judgment, and empathy, and rather than accelerating performance pressures, AI and XR act as supporting partners that reduce cognitive load, enhance self-reflection, and enhance psychological safety,” said Theme Leader Professor Kristina Mikkonen.
One of the main research topics is adaptive XR learning to achieve sustainable capabilities and quality of life. Human-centered XR enables adaptive and inclusive learning pathways that respond to individual needs, cultural diversity, and career stage, improving the quality of life for both healthcare professionals and patients while supporting long-term well-being, resilience, and safe clinical practice.
Latest research: Trailblazers and creating social impact in Oulu
The program focuses on applications in education, healthcare, sustainability, and lifelong learning that create social impact. By combining technological innovation with data-driven and ethical design, Oulu HI aims to ensure that AI enhances human potential while preserving autonomy and trustworthiness.
Please note: This is a commercial profile
This article will also be published in the quarterly magazine issue 26.
Source link
