LLM Alignment should go beyond Harmlessness–Helpfulness and incorporate Human Agency
Cognitive Computation · 2026-12 · 18 (1)
DOI: 10.1007/s12559-026-10568-9 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cognitive Computation, online erschienen am 13.03.2026. Verfasst von Usman Naseem und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Macquarie University.
Abstract (Originalsprache)
Large Language Models are transforming communication, research, and decision-making, but misalignment – when models diverge from human values, safety requirements, or user intent – poses serious risks. In this position paper, we argue that many alignment failures stem from operational choices in training and deployment. We posit that alignment should shift from static, post-training constraints toward dynamic, participatory approaches that safeguard pluralism, autonomy, and human flourishing. We outline forward-looking directions, including pluralistic evaluation, transparency, and the Flourishing–Justice–Autonomy (FJA) framework, and present a roadmap for advancing alignment research and practice.
Förderung
- Macquarie University
Zitieren
APA
Naseem, U., Chakraborty, T., Chang, K., Dras, M., Nakov, P., Peng, N., & Poria, S. (2026). LLM Alignment should go beyond Harmlessness–Helpfulness and incorporate Human Agency. Cognitive Computation, 18(1). https://doi.org/10.1007/s12559-026-10568-9
Vancouver
Naseem U, Chakraborty T, Chang K, Dras M, Nakov P, Peng N, et al. LLM Alignment should go beyond Harmlessness–Helpfulness and incorporate Human Agency. Cognitive Computation. 2026;18(1). doi:10.1007/s12559-026-10568-9
BibTeX
@article{Naseem2026,
author = {Naseem, Usman and Chakraborty, Tanmoy and Chang, Kai-Wei and Dras, Mark and Nakov, Preslav and Peng, Nanyun and Poria, Soujanya},
title = {LLM Alignment should go beyond Harmlessness–Helpfulness and incorporate Human Agency},
journal = {Cognitive Computation},
year = {2026},
volume = {18},
number = {1},
doi = {10.1007/s12559-026-10568-9}
}
EUVIMED