From Detection to Visibility: Reclaiming Assessment Integrity with AI-Powered Learning Analytics

Wednesday, September 30, 2026 | 1:00PM–1:45PM MT
Session Type: Breakout Session
Delivery Format: Presentation
Higher education faces a growing crisis in assessment validity as traditional student outputs—essays, code, and problem sets—no longer reliably reflect original thinking in an AI-rich environment. Detection-based responses are pedagogically limited and fail to capture authentic learning processes. This proposal reframes learning analytics as “Learning Visibility Infrastructure,” shifting from AI detection and predictive risk scoring to the capture of “Reasoning Traces” and “Communicative Acts” through faculty-designed, pedagogically constrained AI tutors. By leveraging structured interaction logs, faculty gain interpretable evidence of student reasoning, revision patterns, and decision-making processes. The session demonstrates how process-level analytics can both restore assessment validity and generate publishable scholarship from the evolving landscape of AI-mediated learning.

Presenters

  • Meenakshi Baker

    Snr. Program Manager, AI Center of Excellence, New York University

  • Elizabeth McAlpin

    Director of Educational Technology Research, New York University