📄 Future of Work in the Age of LLMs

Summary

This is a two-page tutorial abstract, not an empirical paper, so it presents no new data, experiments, or results. Its subject is how large language models are reshaping human labor. LLMs can now carry out elaborate instructions and operate software, so they can support or even fully take over job tasks once seen as human-only. The authors note that this is already affecting labor markets and raising fears about lost jobs, weaker human control, and excessive dependence on automation. Three questions guide the tutorial: how occupations and their task demands will shift, what role LLM-based systems will play as partners or independent workers, and what technical and data infrastructure human–AI teamwork requires. The planned sequence starts with the economics of work and NLP-driven automation. It then covers ways to build LLMs that augment workers and recent LLM agents, followed by workplace-focused evaluation (datasets, benchmarks, metrics). It closes with open technical, human, and societal questions. Its main value is connecting NLP research with economic and human-centered views of work for the ACL audience.

Key points

• Venue: ACL 2026 Proceedings, Volume 5 (Tutorial Abstracts), pages 1–2. Materials are hosted at future-of-work-llm-tutorial.github.io.
• Authors: Zora Wang (CMU, Language Technologies Institute), Yijia Shao and Diyi Yang (Stanford Computer Science), and David Nguyen (Stanford Digital Economy Lab at HAI; ESCoE).
• The tutorial has five parts: (1) the economic landscape and automation, (2) LLMs that augment labor, (3) LLM-based agents, (4) evaluation in workplace settings, and (5) open questions.
• The abstract names no specific datasets, benchmarks, or numerical findings. It cites Hoffmann et al. (2024) and Demirci et al. (2025) on labor effects, and Hazra et al. (2025) on risks such as overreliance.
• The tutorial builds on Yang's earlier tutorials on language agents (EMNLP 2024), social intelligence (NAACL 2025), and human–AI collaboration (ACL 2025).