SDL for Innovation: Learning in the FSU Innovation Hub
Intensive Innovation Events as Informal Learning Environments for Self-Directed Learning
To understand how self-directed learning (SDL) operates within fast-paced, informal environments, our research inquiry investigates how learners tackle complex, undefined “wicked challenges” during intensive innovation events such as 24-hour design sprints hosted at the FSU Innoation Hub. Grounded in a six-step interpretation of Malcolm Knowles’ SDL definition, spanning initiative, needs assessment, goal setting, resource identification, strategy implementation, and progress evaluation, we are investigating participants’ SDL through post-event surveys, interviews, focus groups, activity logs, and artifact analysis.
Initial findings are showing behavioral patterns in problem-based environments. Although participants exhibit high initial curiosity and openness, they frequently struggle to diagnose their precise learning needs at the outset. Furthermore, when confronting problem-solving roadblocks under tight time constraints, participants demonstrate preference for human supports, prioritizing teammates and expert mentors over digital tools, static manuals, or generative AI, which seems to functions as a supplemental accelerator rather than a primary learning support.
As we continue to investigate in these areas, we expect insights to inform the design of informal learning environments and intensive events such as intensive innovation events. It seems likely that by actively scaffolding problem-scoping early in an event and prioritizing human-centric collaboration to cultivate the SDL capabilities required in a time of accelerating technological change.
Teachers’ SDL: International Research Partnership with North-West University, South Africa
Pre-Service Teachers’ Self-Directed Learning (SDL) for Professional Development
The teaching profession requires continuous, agile learning to navigate evolving curricula and classroom demands, prompting educators to rely heavily on informal self-directed learning (SDL) to solve immediate problems of practice. Historically, teachers have navigated these challenges through established human networks—such as mentors, peers, and professional learning communities—strategically reserving complex “core” instructional problems for trusted colleagues while using online tools for peripheral “background” tasks. However, the rapid proliferation of Generative AI (GenAI) introduces an instantaneous, highly personalized form of just-in-time professional development that disrupts these established help-seeking patterns.
The purpose of this research project is to explore how pre-service teachers engage in self-directed learning (SDL) for professional development and the role generative AI tools play in this process.
This research project investigates this topic during two distinct career-entry stages: 1st-year (new) and 4th-year (internship-phase) PSTs. Specifically, the study investigates how PSTs perceive SDL, what targeted training opportunities (e.g., design sprints and workshops) best support their professional growth, and the precise stages of the SDL process at which they turn to GenAI for support. By grounding this investigation in the tension between cognitive offloading and the OECD definition of student agency (i.e., the capacity to set goals, reflect, and act responsibly to effect change rather than being passively shaped), this project evaluates whether GenAI empowers PSTs as active agents or fosters over-reliance. Furthermore, the study examines how AI shifts the social nature of PSTs’ SDL processes, strategies, and self-evaluation practices as they transition from initial teacher preparation into active classroom internships.
SDL Analytics: Context-Aware SDL + AI Model Intelligence Reliability (CASTMIR)
Context-Aware AI Performance Intelligence: A Self-Directed Learning System for Institutional AI Accuracy Tracking
Grounded in Malcolm Knowles’ (1975) definition of self-directed learning (SDL), where individuals take the initiative to diagnose their learning needs, select resources, and evaluate their own outcomes, the CASTMIR project empowers users and institutions to autonomously monitor, refine, and optimize their technological engagement as part of their larger SDL activity. This is of vital importance as institutions (and individuals) today are adopting and utiliizing more AI platforms without a unified performance layer to detect silent model degradation, evaluate output accuracy, or identify compounding technical debt. That is, everyone is adopting new resources faster than they can update strategies or evaluate effectiveness.
To close this gap, the CASTMIR platform introduces a novel approach that integrates four agents for performance intelligence: a multi-metric performance monitor tracking quality and cost; a three-way degradation diagnostician identifying model, prompt, and context drift; an autonomous recommendation engine that coaches users with in-context prompt improvements and suggests routing optimizations; and a reporting engine, an interactive, college-level dashboard for real-time benchmarking across departments and sub-units. By moving beyond passive business intelligence reporting to autonomous correction, we aim to develop this tool-agnostic architecture delivers the first closed-loop AI performance system scalable across multi-platform institutional and enterprise environments.