Protecting Privacy in Indian Schools: Regulating AI-based Technologies' Design, Development and Deployment

Abstract

Education is one of the priority areas for the Indian government, where Artificial Intelligence (AI) technologies are touted to bring digital transformation. Several Indian states have also started deploying facial recognition-enabled CCTV cameras, emotion recognition technologies, fingerprint scanners, and Radio frequency identification tags in their schools to provide personalised recommendations, ensure student security, and predict the drop-out rate of students but also provide 360-degree information of a student. Further, Integrating Aadhaar (digital identity card that works on biometric data) across AI technologies and learning and management systems (LMS) renders schools a ‘panopticon’.

Certain technologies or systems like Aadhaar, CCTV cameras, GPS Systems, RFID tags, and learning management systems are used primarily for continuous data collection, storage, and retention purposes. Though they cannot be termed AI technologies per se, they are fundamental for designing and developing AI systems like facial, fingerprint, and emotion recognition technologies. The large amount of student data collected speedily through the former technologies is used to create an algorithm for the latter-stated AI systems. Once algorithms are processed using machine learning (ML) techniques, they learn correlations between multiple datasets predicting each student’s identity, decisions, grades, learning growth, tendency to drop out, and other behavioural characteristics. Such autonomous and repetitive collection, processing, storage, and retention of student data without effective data protection legislation endangers student privacy.

The algorithmic predictions by AI technologies are an avatar of the data fed into the system. An AI technology is as good as the person collecting the data, processing it for a relevant and valuable output, and regularly evaluating the inputs going inside an AI model. An AI model can produce inaccurate predictions if the person overlooks any relevant data. However, the state, school administrations and parents’ belief in AI technologies as a panacea to student security and educational development overlooks the context in which ‘data practices’ are conducted. A right to privacy in an AI age is inextricably connected to data practices where data gets ‘cooked’. Thus, data protection legislation operating without understanding and regulating such data practices will remain ineffective in safeguarding privacy.

The thesis undergoes interdisciplinary research that enables a better understanding of the interplay of data practices of AI technologies with social practices of an Indian school, which the present Indian data protection legislation overlooks, endangering students’ privacy from designing and developing to deploying stages of an AI model. The thesis recommends the Indian legislature frame better legislation equipped for the AI/ML age and the Indian judiciary on evaluating the legality and reasonability of designing, developing, and deploying such technologies in schools.