From Algorithmic Adherence to Cognitive Autonomy: Mitigating Automation Bias in Student Decision-Making Workflows

This paper investigates the critical intersection of generative Artificial Intelligence (AI) integration and cognitive development within the Indian higher education ecosystem under the National Education Policy (NEP) 2020 framework. While the rapid democratization of Large Language Models (LLMs) across metropolitan hubs and Tier-2 cities has drastically eliminated resource accessibility gaps, it has simultaneously introduced severe cognitive vulnerabilities among young learners. This study examines how the exceptional textual fluency and lack of linguistic hedging in AI outputs function as a psychological camouflage, actively inducing automation bias the systemic tendency to uncritically accept automated suggestions even when they defy empirical facts or human logic. Utilizing a mixed-methods systematic literature review comprising high-impact academic sources, the research analyzes the phenomenon of "cognitive offloading," wherein students delegate core analytical tasks like literature synthesis, error detection, and data interpretation to conversational agents. This offloading fundamentally bypasses the essential phase of "productive failure," leading to an "epistemic squeeze" and the systematic erosion of metacognitive and independent judgment skills. In the context of Indian universities, these vulnerabilities are severely magnified by institutional pressures, such as high student-to-faculty ratios, exam-intensive curricula, a premium on grade scores for campus placements, and the reliance on language tools by regional-language speakers, which inadvertently facilitates epistemic colonization. To counteract this accelerating algorithmic conformism, this paper proposes a systemic shift toward Structured Epistemic Scaffolding. Actionable recommendations include redesigning evaluation architectures from output to process, integrating "friction-by-design" into early-stage curricula, institutionalizing mandatory epistemic literacy courses, and developing localized, bilingual scaffolding tools to preserve cognitive autonomy and foster AI-critical graduates.