Human-Centered AI Language Technology (HCAILT): An Empathetic Design Framework for Reliable, Safe and Trustworthy Multilingual Communication
Year: 2026
Type: article
Abstract: Language technologies are increasingly ubiquitous and now translate emergency bulletins, draft clinical notes and mediate everyday conversations, yet their impressive fluency can be misleading–masking limited reliability, unpredictable errors and uneven performance across different user groups and languages. Building on Shneiderman’s human-centered AI (HCAI) paradigm, this article introduces the Human-Centered AI Language-Technology (HCAILT) model, a domain-specific framework that binds reliability, safety culture and trustworthiness to the full language-technology pipeline. HCAILT couples technical guardrails (such as retrieval-augmented generation and quality estimation) with organizational practices (like bias audits and incident-report loops), together with user-facing features that maintain meaningful human control. Two blueprint use cases–in multilingual healthcare and crisis communication–illustrate how the HCAILT model guides system architecture, deployment practices and evaluation. A demo system demonstrates immediate feasibility on public large language models. By translating HCAI principles into actionable design levers, HCAILT provides scholars, developers and policymakers with a pragmatic path from ethical aspiration to deployable practice. The paper concludes with a research agenda for empirical validation in real-world settings and invites multidisciplinary collaboration to ensure that next-generation language technologies are not merely powerful, but demonstrably reliable, safe and worthy of public trust.
Authors: Vicent Briva-Iglesias, Sharon O’Brien
Institution: Dublin City University
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7127900910