HCC Is All You Need: Alignment-The Sensible Kind Anyway-Is Just Human-Centered Computing

TL;DR

AI alignment is reframed as a problem in Human-Centered Computing (HCC) rather than a standalone field. The paper analyzes alignment through HCC lenses, drawing on STS, Ethics, and design research, and enumerates core questions about stakeholders, preferences, design choices, evaluation, and anti-oppression. It proposes using the established HCC toolkit and themes instead of creating new terminologies, and formalizes the idea with an equation that situates alignment as optimizing a loss function over a space of human preferences. This reframing aims to connect AI safety with practical, participatory, and ethically grounded design, enabling cross-disciplinary insights and leveraging existing HCC structures (journals, conferences, and NSF programs).

Abstract

This article argues that AI Alignment is a type of Human-Centered Computing.