RECONCEPTUALISING EFL ASSESSMENT IN THE AGE OF GENERATIVE AI: DEVELOPING A HUMAN-CENTRED AND AI-RESILIENT ASSESSMENT FRAMEWORK FOR EFL LEARNERS
Annotatsiya
The rapid development of generative artificial intelligence (GenAI) has challenged the assumptions underlying conventional English as a Foreign Language (EFL) assessment. Assessment practices that rely predominantly on final written products increasingly face difficulties in establishing whether submitted work represents learners’ own linguistic knowledge, reasoning, and decision-making. This conceptual study examines how EFL assessment can be reconceptualised in response to the growing presence of GenAI and proposes a Human-Centred AI-Resilient EFL Assessment Framework. The study employs a structured conceptual synthesis of recent scholarship published primarily between 2024 and 2026 on generative AI, language assessment, EFL writing, assessment validity, teacher agency, and AI-resilient assessment. The analysis indicates that detection-oriented approaches alone provide an insufficient response to the assessment validity problem. Instead, assessment should generate multiple forms of evidence of learning through authentic task design, process documentation, learner reflection, oral verification, transparent AI-use disclosure, and professional teacher judgement. Based on this synthesis, the proposed framework integrates six dimensions: authenticity, process evidence, human agency, oral verification, responsible AI use, and multidimensional judgement. The framework shifts the central assessment question from whether AI has been used to whether the learner can demonstrate meaningful linguistic competence, reasoning, and ownership of learning. The study argues that AI-resilient assessment should not seek to eliminate technology from educational environments but should redesign assessment so that evidence of learning remains attributable to the learner. The framework provides a conceptual basis for future empirical validation in secondary-school EFL contexts.