ECAI Model — Yanina Torres Research

Research · Higher education · Artificial intelligence

ECAI Model

Assessment in Coexistence with Artificial Agents

The ECAI Model is a conceptual proposal for rethinking assessment in higher education in contexts where students coexist with generative artificial intelligence agents. It proposes assessing both the knowledge and abilities students can draw on independently and the intellectual operations they perform when interacting with those agents.

The proposal addresses a problem: restrictions on artificial intelligence, together with the persistence of assessment practices focused exclusively on individual performance, may leave out forms of learning and knowledge production that take place through interaction with technological tools.

Conceptual foundations

ECAI brings together the perspective of distributed cognition (Salomon and Pea), the distinction between person-solo and person-plus (Perkins), coexistence with artificial agents (Floridi), and the view of technologies as intellectual partners in learning activities (Jonassen).

Person-solo and person-plus

Person-solo

Assessment examines the knowledge and abilities students have internalized and can draw on without assistance from artificial intelligence.

Person-plus

Assessment examines the operations performed with artificial agents and other resources, the construction or revision of knowledge, and the outcome of the interaction.

ECAI proposes combining both modes according to course objectives, disciplinary characteristics, and professional requirements. It does not prescribe a universal proportion between them.

The interstice

The interstice is understood as a space for constructing and discussing knowledge that emerges through coexistence between a human agent and an artificial agent. Meaningful intellectual operations can take place there, such as selecting, comparing, investigating, arguing, justifying, revising, or rejecting contributions.

Interaction may involve collaboration, exploration, expansion, or conceptual reorganization; it may also involve tension or friction. Friction is one possible dynamic through which the interstice is constituted, not a necessary condition. Not every conversation with AI constitutes, in itself, an interstice of meaningful learning.

How the model works

  1. Instructional design. The instructor defines disciplinary objectives, professional competencies, and conditions for interaction.
  2. Human–AI interaction. Activities take place in which knowledge can be constructed, discussed, or revised.
  3. Process documentation. Evidence is collected of the intellectual operations and decisions made during the interaction.
  4. Assessment. The process, the final product, and, where appropriate, the knowledge and abilities demonstrated in person-solo assessment are examined.

Assessment dimensions and evidence

The model considers learning with AI (disciplinary), learning about AI (instrumental), learning that can be demonstrated individually, the interstitial product, executive function as expressed in observable decisions, and the ethical aspects of interaction.

Possible evidence includes records of exchanges, justifications for choices, comparison and validation of responses and sources, individual work, the final product, and disclosure of the use of artificial agents. Specific criteria must be adapted to each course.

Scope and limitations

ECAI is a conceptual model whose implementation and empirical evaluation require further research. Interaction records allow observable operations to be examined, but do not provide direct access to all of a student’s cognitive processes. Its application also depends on access conditions, teacher training, confidentiality, and data protection.

The proposal does not assume that every interaction with artificial intelligence produces meaningful learning, nor that the use of artificial agents replaces the assessment of disciplinary knowledge.

Academic presentation

2026 — 8th Meeting on Scientific and Academic Communication in Education (Santiago, Chile). Presentation: “From Person-Solo to Person-Plus: Human Agency and University Assessment in the Age of Artificial Intelligence”.

Academic presentation of the ECAI Model · October 2026.

Download Presentation (Spanish only)
Download Abstract