Mark Ollila

My research focuses on Play, particularly through the MIRANDA framework, which uses knowledge graphs derived from gameplay to understand learning as an emergent property of play. MIRANDA captures and interprets player interactions, applying principles from James Paul Gee’s learning theory and Bloom’s taxonomy to reveal how cognitive, social, and collaborative skills evolve during gameplay. By representing learning as a dynamic, interpretable network, MIRANDA provides a structural alternative to the opaque representations of large language models—transforming play data into actionable insight for educators and learners alike. This work explores how games can serve as transparent, measurable environments for demonstrating and credentialing complex human skills.

The Make dimension of my research focuses on the process of building games collaboratively in large, community-based teams as a powerful mode of learning. Rather than treating game creation as an isolated act of design or coding, this approach emphasizes the social, organizational, and creative dynamics that emerge when diverse individuals work together to build a shared world. Within these teams, participants develop vital skills across domains—narrative design, art, 3D modeling, audio production, systems thinking, and project management—while simultaneously learning to communicate, negotiate, and co-create effectively. This “Make-to-Learn” model transforms the game development process itself into a living classroom, where the act of producing a game mirrors the collaborative complexity of real-world innovation.

By situating learning inside the collective act of creation, these large-scale game-making communities also become laboratories for exploring new forms of production. AI tools increasingly act as collaborators, assisting in scripting, design iteration, asset generation, and quality assurance, allowing teams to focus on creativity and coordination rather than repetitive tasks. This interplay between human and machine contribution reveals emerging patterns of distributed creativity and adaptive learning. The result is a new paradigm of game making as social learning infrastructure—one that not only produces games but cultivates the next generation of creators who understand how to learn, lead, and build together in complex, evolving systems.

The Earn component aligns with Realm 5, a concept for infinitely scalable and infinitely affordable learning ecosystems. This research envisions a future where learning—facilitated by AI, play, and peer-to-peer ecosystems—can occur continuously and globally, independent of traditional institutional constraints. Realm 5 treats learning as a living economy of creativity and contribution, where skills demonstrated through systems like MIRANDA can lead to microcredentials, income, and opportunity. By linking demonstrable competence in play and creation to value generation, this model proposes a new social contract for education: one that rewards lifelong learning as a form of productive play.

Finally, my work in Games and AI explores how generative systems can create entirely new game frameworks, patterns, and genres. One such project is the Unified Scene Atom System (USAS)—a standalone, GPU-accelerated video-to-3D reconstruction pipeline capable of producing photorealistic 3D models from ordinary video. USAS represents a frontier in AI-driven world creation, advancing toward the concept of “Scene Atoms,” which unify geometry, radiance, and temporal dynamics into a single adaptive primitive. By merging AI, 3D graphics, and learning science, I aim to create environments where both the world and the learner evolve together—games that don’t just teach, but that learn in return.