Transformasi Paradigma Evaluasi Pembelajaran di Era AI: Dari Assessment of Learning Menuju Intelligent, Humanistic, and Personalized Assessment (IHPA)

Abstract

The diffusion of artificial intelligence into educational practice has outpaced the conceptual frameworks available for evaluating learning. Recent work has advanced automated scoring, adaptive testing, and generative feedback, yet the frameworks proposed so far tend to treat the technology either as an instructional and administrative aid or as a risk to be governed, and few specify the point at which algorithmic analysis must give way to pedagogical judgement. This article addresses that gap by reconstructing the paradigm of learning assessment and proposing a conceptual model termed Intelligent, Humanistic, and Personalized Assessment (IHPA). The study employs library research with an analytical-conceptual approach; forty-two sources retrieved from international and national databases under explicit inclusion criteria were examined through content analysis, thematic coding, cross-theme comparison, and conceptual synthesis. The discussion traces the movement from assessment of learning towards assessment for and as learning, and further towards intelligent assessment driven by learning analytics and adaptive testing. IHPA rests on humanism, constructivism, and human-centred artificial intelligence, and operates through seven stages: learner profiling, learning data collection, artificial intelligence analysis, adaptive assessment, humanistic decision making, personalized feedback, and continuous improvement. Its specific contribution lies in the fifth stage, which makes teacher interpretation a mandatory filter between algorithmic output and assessment decisions, and in extending the framework to the moral and spiritual dimensions central to Islamic education. The model remains conceptual, and empirical testing of its validity and practicality is an agenda for further research.