Malihe Alikhani
Northeastern University
The Alignment Gap: When AI Agrees Too Much and Institutions Assume Too Little
Abstract
What does it mean for an AI system to be aligned, and aligned with whom? In this talk, I treat alignment not as a fixed technical property, but as a narrative that spans machine learning research, system design, and public policy. I begin with my work on sycophancy and friction in language models, showing how alignment to user intent can collapse into over agreement, misplaced confidence, and reduced epistemic resilience. I then connect these dynamics to applied systems, including sign language technologies and code generation tools, where alignment choices shape access, productivity, and who ultimately benefits. Finally, I reflect on how these patterns surface in policy contexts, drawing on my experience working in Congress and with international partners, where alignment is often assumed rather than examined. I argue that alignment without friction is fragile, and that closing the alignment gap requires institutions to take responsibility for how alignment choices shape whose judgment is mirrored, whose access is granted, and whose agency is preserved.
About
Malihe Alikhani is an assistant professor in the Khoury College of Computer Sciences at Northeastern University, based in Boston. In an era of increasingly present AI, Alikhani is both enthused and wary of its transformative power. She believes in the ability of language technologies and AI to bolster critical education, health, and social justice efforts, but she also studies the ways in which they become biased. Her vision is to design inclusive and equitable language technologies that communicate effectively with diverse populations. By integrating insights from cognitive science, social sciences, and machine learning, these models can capture diversity of interpretation and benefit underserved communities.
Alikhani practices in the classroom what she preaches in her research, and strives to prepare a diverse generation of students to deploy the transformative power of AI. After three years as a professor at the University of Pittsburgh, during which time she garnered a handful of “best paper” honors, she aims to build new collaborations within Northeastern’s Ethics Institute, the Institute for Experiential AI, and the Network Science Institute. Her work has been supported by DARPA, the NIH, Google, and Amazon, and her team is among the finalists of the second Amazon Alexa TaskBot Challenge.