AI-Driven Inclusive Assessment Tool for Education

Abstract

Traditional exam systems usually depend on fixed question sets and manual grading. Such methods fail to address differences in students’ skills or accessibility needs. In this work, an AI-based assessment platform is presented that focuses on flexibility and inclusivity. The system uses Natural Language Processing (NLP) techniques to automate the evaluation of descriptive answers. It applies adaptive logic to pick multiple-choice questions (MCQs) that match a learner’s level of understanding. Handwritten responses are checked with Optical Character Recognition (OCR) to keep the marking process accurate and fair. For better accessibility, Speech-to-Text (STT) and Text-to-Speech (TTS) tools are added, helping students who face visual or motor challenges take the test comfortably. The overall system follows a modular architecture, connecting React.js for the user interface, Node.js for backend operations, and Python-based AI modules for question generation and scoring. This paper focuses on the system design, methodology, and prototype-level feasibility analysis of an inclusive AI-driven assessment framework.