Rethinking AI in Education: The CoRe-3 Model

education, the rise of generative AI is a double-edged sword. It makes getting answers easy but understanding the process hard. Schools still largely focus on unaided performance. Yet, the reality is that today's real task is to harness AI effectively. What does that involve? Framing a task, judging AI output, and steering the model for optimal results.

The CoRe-3 Approach

Introducing CoRe-3, or Co-Reasoning, a competency model that seeks to redefine how we measure productive AI use in educational settings. The model breaks down AI use into three distinct skills: Framing, Judging, and Steering, abbreviated as FJS. Framing involves defining an ill-structured task before invoking AI. Judging is about evaluating AI outputs for errors and underlying assumptions. Steering is the iterative process of redirecting the AI model.

Here's what the benchmarks actually show: by separating pre-generation Framing from post-generation Steering, with Judging as the intermediary, CoRe-3 provides a more nuanced understanding of AI interaction. Schools could benefit from this approach, yet they rarely assess these skills independently. Could this be why AI use in classrooms often fails to hit the mark?

The Testing Ground

The CoRe-3 model has been instantiated in CoReasoningLab. This open platform presents users with flawed AI outputs and scores each response independently. Over simulations involving different AI models, each skill tracks its own competence without overlap. When a single competence is shared across the FJS skills, the grades align, demonstrating both convergent and discriminant validity.

Notably, CoRe-3 isn't just theoretical. Human-rater agreement and outcomes are next on the agenda. The model isn't just for academics. it's an actionable framework backed by data, grounding the lofty ambitions of AI education in practical reality.

Why It Matters

Strip away the marketing and you get a critical look at how we engage with AI in learning contexts. The architecture matters more than the parameter count. Are we equipping students to use AI effectively, or merely letting them offload cognitive effort? The CoRe-3 model challenges educators to rethink their methods and implement a more comprehensive evaluation of AI skills.

In an era where AI is ubiquitous, the need for nuanced skills assessment is important. If educational systems continue to measure old skills without adapting to new technology, they'll be left behind. The CoRe-3 model offers a blueprint for educators willing to embrace this challenge. It not only aims to improve AI literacy but also to ensure students are prepared for a future dominated by artificial intelligence.