AI and Higher Ed Predictions

As many of us head into a potential summer “break” I thought we might want to hear predictions on how some [experts] believe we might integrate instructional design with 2025 AI innovations. 

1. Agentic AI and Cognitive Apprenticeship

Agentic AI refers to systems that can reason, plan, and act autonomously, often breaking down complex problems into manageable subgoals. This directly aligns with cognitive apprenticeship, where learners develop expertise through guided experience on real-world tasks (Collins, Brown, & Newman, 1989). AI agents can model expert problem-solving behavior, offer coaching, and provide scaffolding that fades as learners gain proficiency—mirroring the zone of proximal development (Vygotsky, 1978). In instructional design, agentic AI can serve as a digital mentor, supporting PBL and complex inquiry tasks while dynamically adjusting guidance levels (Reiser, 2004).

2. Inference Time Compute and Information Processing Theory

Inference time computation introduces dynamic, real-time reasoning based on the complexity of a user’s query. This capability can be interpreted through information processing theory, which emphasizes attention, working memory, and long-term memory (Atkinson & Shiffrin, 1968). Educationally, this supports adaptive instruction, where the AI adjusts the depth and complexity of its responses based on learner input and cognitive load. Such responsiveness mirrors principles from cognitive load theory (Sweller, 2011), suggesting that variable reasoning depth can prevent overload and optimize schema acquisition in complex learning environments.

3. Large Models, Constructivism, and the TPACK Framework

The rise of trillion-parameter models capable of deep contextual understanding aligns with constructivist learning theory, which posits that learners construct knowledge through interaction with rich, meaningful environments (Fosnot & Perry, 2005). These large models can serve as epistemic partners in knowledge construction, co-generating insights with learners. Instructional design models like TPACK (Technological Pedagogical Content Knowledge) emphasize the intersection of technology, pedagogy, and content (Mishra & Koehler, 2006), underscoring the importance of pedagogically sound integration of large models that can enhance inquiry, discussion, and metacognitive dialogue.

4. Very Small Models and the SAMR Model

The proliferation of highly efficient, task-specific small models running on local devices supports the SAMR model of technology integration (Puentedura, 2010). These lightweight agents allow modification and redefinition of instructional tasks—enabling personalized learning experiences on mobile devices without requiring cloud connectivity. Their availability democratizes AI-enhanced learning by making it feasible in low-resource or offline environments, consistent with the universal design for learning principle of providing access to tools for all learners.

5. Advanced Use Cases, Connectivism, and Authentic Assessment

AI's ability to handle complex real-world tasks—such as optimizing networks or defending against cybersecurity threats—invites alignment with connectivism, which emphasizes learning as a process of network formation (Siemens, 2005). As AI augments problem-solving across domains, instructional designers should embed authentic assessments that require learners to engage with realistic, open-ended scenarios in partnership with AI tools (Herrington, Reeves, & Oliver, 2014). This positions AI as both a collaborator and evaluator in demonstrating applied competencies in digital contexts.

6. Near-Infinite Memory and Situated Cognition

The concept of near-infinite memory—where AI agents maintain persistent, contextual knowledge of user interactions—supports situated cognition theory, which posits that knowledge is inherently tied to the context of its acquisition and use (Brown, Collins, & Duguid, 1989). AI tools with long memory windows can contextualize support based on past learner behavior, goals, and misconceptions. Instructional design can consider longitudinal learning analytics that capitalize on these persistent interactions to guide personalization, formative feedback, and metacognitive reflection over time.

7. Human-in-the-Loop Augmentation and Distributed Cognition

The paradox of underperformance in human-AI collaboration highlights the need for better interfaces that support distributed cognition, where cognitive processes are shared across people and tools (Salomon, 1993). Effective augmentation requires designing scaffolded user interfaces that help professionals interact with AI agents without needing expert prompt engineering skills. Instructional design strategies such as cognitive task analysis and guided prompting frameworks can support productive synergy, ensuring that human-AI teams outperform either component alone (Norman, 1993).

References

Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. In The psychology of learning and motivation. Academic Press.

Brown, J. S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18(1), 32–42. https://doi.org/10.3102/0013189X018001032 

Ertmer, P. A., & Newby, T. J. (2013). Behaviorism, cognitivism, constructivism: Comparing critical features from an instructional design perspective. Performance Improvement Quarterly, 26(2), 43–71. https://doi.org/10.1002/piq.21143 

Fosnot, C. T., & Perry, R. S. (2005). Constructivism: A psychological theory of learning. In C. T. Fosnot (Ed.), Constructivism: Theory, perspectives, and practice. Teachers College Press.

Mishra, P., & Koehler, M. J. (2006). Technological pedagogical content knowledge: A framework for teacher knowledge. Teachers College Record, 108(6), 1017–1054.

Norman, D. A. (1993). Things that make us smart: Defending human attributes in the age of the machine. Addison-Wesley.

Puentedura, R. (2010). SAMR and TPCK. Retrieved at http://www.hippasus.com/rrpweblog 

Salomon, G. (Ed.). (1993). Distributed cognitions: Psychological and educational considerations. Cambridge University Press.

Siemens, G. (2005). Connectivism: A learning theory for the digital age. International Journal of Instructional Technology and Distance Learning, 2(1), 3–10.

Sweller, J. (2011). Cognitive load theory. Psychology of Learning and Motivation, 55, 37–76. https://doi.org/10.1016/B978-0-12-387691-1.00002-8 

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.