Agent theory · October 18, 2026
The I-PACE Framework and AI Tutors in the Classroom
This essay examines how the I-PACE framework explains the tension between focused immersion and AI dependency when students use tutoring agents. It is written for teachers and researchers designing or evaluating AI tools in educational settings.
The I-PACE model offers a psychological lens for understanding how students interact with AI tutoring agents, balancing deep engagement against the risk of cognitive outsourcing. This essay is for teachers and researchers who want to understand why an AI tool that captivates a student might simultaneously erode their capacity for independent thought.
When a large language model acts as a tutor, it does not merely deliver information; it alters the psychological environment of the learner. Educators often evaluate these tools by looking at test scores or completion rates, but those metrics miss the internal mechanics of how a student relates to the machine. To understand this relationship, we can look to the I-PACE framework. Originally developed to explain problematic internet use, I-PACE stands for Interaction of Person-Affect-Cognition-Execution. It maps how a person’s individual traits, emotional states, and cognitive biases drive their actual behavior when using digital technology. Recently, researchers have applied a partial I-PACE framework specifically to artificial intelligence, revealing a double-edged dynamic that every teacher deploying an AI agent should recognize.

The Pull of Focused Immersion
In a traditional classroom, capturing a student's attention is half the battle. AI tutoring agents are remarkably effective at winning this battle because they respond instantly, adapt to the user's tone, and never show fatigue. Within the I-PACE framework, this speaks directly to the affective and cognitive components of the interaction. When a student perceives an AI agent as highly intelligent and responsive, they are more likely to enter a state of focused immersion. This is a cognitive state where attention narrows, distractions fade, and the learner becomes deeply absorbed in the task at hand.
For a teacher coordinating work in a busy classroom, focused immersion is a valuable asset. If an AI assistant can keep a struggling reader engaged with a text for twenty uninterrupted minutes, the immediate educational benefit is clear. The student is practicing, reading, and interacting rather than disengaging. The cognition component of I-PACE helps explain why this happens: the AI reduces the friction of the learning task, making the execution phase—actually doing the work—feel smoother and more rewarding. The affective response is often positive, characterized by curiosity or satisfaction, which reinforces the desire to keep working with the agent.
However, the I-PACE model reminds us that execution is driven by the interaction of all these variables, not just the immediate feeling of flow. A student might feel immersed, but the nature of that immersion matters. Are they immersed in the process of thinking, or are they immersed in the process of prompting?

The Trap of AI Dependency
The same perceived intelligence that facilitates focused immersion introduces a significant risk: AI dependency. According to research applying the partial I-PACE framework to AI interactions, when students believe the AI is highly capable, they are more likely to outsource their critical thinking to it. This is where the execution component of the model begins to falter from an educational standpoint. Instead of executing a cognitive strategy—like analyzing a historical document or debugging a line of code—the student executes a behavioral strategy: asking the AI to do it for them.
This dependency undermines the very critical thinking skills that education aims to build. The I-PACE framework shows that this is not simply a matter of laziness. It is a predictable psychological outcome. The person variable (a student who may already lack confidence in their analytical skills) interacts with the cognition variable (the belief that the AI is smarter and faster) and the affect variable (the relief of avoiding frustration). Together, these produce an execution pattern centered on reliance rather than effort. Over time, the student learns that the most efficient way to complete a task is to bypass their own reasoning entirely.
For teachers, this presents a difficult design challenge. An AI agent that is too helpful can become a crutch. If the system is optimized purely for user satisfaction or task completion, it will naturally encourage dependency. The student feels good, the assignment gets done, but the underlying cognitive architecture remains undeveloped.
Designing Ethical Execution
Understanding the I-PACE framework is not just an academic exercise; it is a practical necessity for ethical implementation. As outlined in broader discussions on the ethics of AI in education, the design of AI tutoring systems must be contextualized within socio-technical frameworks that account for both affective and cognitive impacts on learners. We cannot separate the emotional experience of using an AI from the cognitive consequences of relying on it.
If we accept the I-PACE model’s premise—that a student’s background, emotions, and beliefs about the AI dictate how they use it—then educators and developers must intervene at the level of system design. An AI agent should not be built solely to maximize focused immersion if that immersion comes at the cost of critical thinking. Instead, the agent needs to be calibrated to disrupt dependency before it solidifies into a habit.
Practically, this means the AI should occasionally introduce productive friction. Rather than providing a seamless path to the answer, the agent might ask a clarifying question that forces the student to articulate their own reasoning. It might withhold the final step of a math problem, requiring the student to execute the conclusion themselves. These interventions target the cognition and execution phases of the I-PACE model, gently shifting the student away from passive reliance and back toward active thought.
Teachers also play a vital role in shaping the "Person" variable in the I-PACE equation. By explicitly discussing how AI works, what its limitations are, and why human judgment remains essential, teachers can alter the cognitive biases students bring to the interaction. If a student understands that the AI is a statistical engine rather than an infallible oracle, their perception of its intelligence changes. That shift in perception can reduce the likelihood of blind dependency while preserving the benefits of focused immersion.
Ultimately, placing an AI agent in a classroom is not just a technological upgrade; it is a psychological intervention. The I-PACE framework provides a structured way to anticipate how students will react to that intervention. By recognizing the tension between the deep focus an AI can inspire and the critical thinking it can quietly replace, teachers and researchers can build environments where technology serves the mind without supplanting it.