Framing, Judging, Steering: An Assessable Competency Model for Teach-ing Students to Reason With Generative AI

2026-06-06 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

The CoRe-3 (Co-Reasoning) competency model, submitted on June 4, 2026, factors productive generative AI use into three assessable skills: Framing, Judging, and Steering (FJS). Framing involves specifying an ill-defined task before invoking AI, while Judging evaluates AI output for errors and unstated assumptions. Steering then iteratively redirects the model toward a better result. A distinguishing claim of CoRe-3 is the clear separation of pre-generation Framing from post-generation Steering, with Judging acting as the critical gate between them. The model is instantiated in CoReasoningLab, an open platform designed to present flawed AI output and independently score these skills. Over simulated learners, the skills demonstrate dissociation, tracking manipulated competence while remaining flat in others, and grades correlate when competence is shared, showing convergent and discriminant validity across grader backends from two providers. The authors have released the instrument, data, and protocol.

Key takeaway

For AI educators designing curricula, you should integrate the CoRe-3 competency model to teach students effective generative AI co-reasoning. This framework helps you move beyond simple "prompting" scores by independently assessing Framing, Judging, and Steering skills. Implementing CoReasoningLab can provide a structured platform for diagnosing specific skill deficiencies, ensuring students develop a critical, iterative approach to AI interaction rather than uncritical cognitive offloading.

Key insights

Productive generative AI use requires distinct skills: Framing, Judging, and Steering, which can be independently assessed.

Principles

• AI use involves pre-generation Framing.
• Judging output is a critical gate.
• Steering refines AI-generated results.

Method

The CoRe-3 model (FJS) assesses AI co-reasoning by separating Framing (pre-AI task specification), Judging (output evaluation), and Steering (iterative model redirection). This is implemented in CoReasoningLab.

In practice

• Use CoRe-3 to diagnose AI skill gaps.
• Implement CoReasoningLab for student assessment.
• Separate pre- and post-generation AI tasks.

Topics

• Generative AI Competencies
• AI Co-Reasoning
• Educational Assessment
• CoRe-3 Model
• CoReasoningLab
• AI Literacy

Best for: Research Scientist, AI Scientist, AI Student, AI Ethicist

Related on AIssential

• Framing, Judging, Steering: An Assessable Competency Model for Teach-ing Students to Reason With Generative AI — CoRe-3 defines AI co-reasoning as three distinct, assessable skills: Framing, Judging, and Steering, to counter cognitive offloading.
• Reasoning Core: A Scalable Procedural Data Generation Suite for Symbolic Pre-training and Post-Training — Reasoning Core generates verifiable symbolic data to enhance language model reasoning and can be used for pre-training and post-training.
• Uneven Evolution of Cognition Across Generations of Generative AI Models — Generative AI models exhibit uneven cognitive development, excelling in language but struggling with visual-perceptual reasoning.
• Towards developing future-ready skills with generative AI — Generative AI can effectively assess complex "future-ready" skills through simulated conversational environments.
• A writing professor’s new task in the age of AI: Teaching students when to struggle — Effective AI integration in education requires fostering critical thinking and discernment, not just tool usage.
• Can AI tools assess coding assignments? — Generative AI can assist in coding assignment assessment when integrated thoughtfully with human expertise and context.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.