Abstract

Executive functioning enables learners to organize work, monitor progress, and decide when a response needs revision. For some students with special educational needs, difficulties in these areas can restrict participation even when they understand the subject matter. This critical narrative review considers whether artificial intelligence can reduce such barriers and, more importantly, what kind of support it actually provides. Research on executive functions, self-regulated learning, artificial intelligence in education, and special education was brought together to examine intelligent tutoring systems, adaptive platforms, planning tools, learning analytics, conversational agents, and generative AI. Across this literature, a recurring distinction emerges. Some applications compensate for an executive demand, others scaffold performance while a task is being completed, and a smaller group are designed to foster strategies that learners can later use without the tool. These outcomes should not be treated as equivalent. Completing a task successfully with automated prompts does not, by itself, demonstrate lasting improvement in executive functioning. AI may nevertheless be useful when support is tied to a clearly defined learning goal, interpreted by a teacher, and adjusted or faded in response to the learner’s progress. Its use also raises practical concerns about cognitive offloading, inaccurate feedback, privacy, bias, and unequal access. The review proposes a three-function framework - compensation, scaffolding, and development - for planning and evaluating AI-supported practice.