From Self-Goals to AI-Coaching: Evaluating Large Language Models as Productivity Advisors in Gig Work Environments
Keywords:
large language models, gig work, productivity advisors, AI coaching, governance, fairness, leakage-safe evaluation, algorithmic managementAbstract
The proliferation of gig work has intensified the need for effective productivity support systems that can operate within the highly individualized and volatile context of platform-mediated labor. While self-set goals have long been a cornerstone of worker motivation and performance management, the emergence of large language models presents novel possibilities for AI-driven coaching that can offer real-time, context-aware advice. This paper examines the transition from self-goals to AI-coaching in gig work environments, evaluating large language models as productivity advisors from a systems perspective. We analyze architectural trade-offs inherent in deploying such advisors, including the balance between personalized guidance and standardization, the integration of privacy-preserving mechanisms, and the demands of low-latency inference across heterogeneous devices. Governance and infrastructure considerations are explored, focusing on data ownership, algorithmic accountability, and the sustainability of model updates in rapidly changing labor markets. The paper further addresses fairness and robustness challenges, such as the risk of reinforcing platform biases, the vulnerability of advisory outputs to adversarial manipulation, and the need for evaluation benchmarks that are resilient to information leakage. A leakage-safe benchmark design for market-stress early warning is discussed as a critical methodological contribution to credible assessment. Cross-domain comparisons with AI coaching in healthcare and finance illuminate structural parallels and divergences. The conclusion synthesizes policy implications, advocating for participatory design, transparent auditing, and regulatory frameworks that preserve worker autonomy while harnessing the productivity benefits of AI-assisted goal formation. This work aims to provide a comprehensive foundation for researchers, platform designers, and policymakers navigating the socio-technical frontier of intelligent productivity support.
References
1. Autor, D. H. (2015). Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives, 29(3), 3–30.
2. Min, X., Chi, W., Hu, X., & Ye, Q. (2024). Set a goal for yourself? A model and field experiment with gig workers. Production and Operations Management, 33(1), 205-224.
3. Liu, T. (2026). Beyond volatility: A leakage-safe residual-stress signal for drawdown risk monitoring. Available at SSRN 6503179.
4. Wood, A. J., Graham, M., Lehdonvirta, V., & Hjorth, I. (2019). Good gig, bad gig: Autonomy and algorithmic control in the global gig economy. Work, Employment and Society, 33(1), 56–75.
5. Kellogg, K. C., Valentine, M. A., & Christin, A. (2020). Algorithms at work: The new contested terrain of control. Academy of Management Annals, 14(1), 366–410.
6. Lee, M. K., Kusbit, D., Metsky, E., & Dabbish, L. (2015). Working with machines: The impact of algorithmic and data-driven management on human workers. In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (pp. 1603–1612). ACM.
7. Liu, T. (2026). Leakage-Safe Benchmark Design for Market-Stress Early Warning: An Economically Credible Evaluation.
8. Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., ... & Liang, P. (2021). On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258.
9. Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610–623). ACM.
10. Weidinger, L., Mellor, J., Rauh, M., Griffin, C., Uesato, J., Huang, P. S., ... & Gabriel, I. (2021). Ethical and social risks of harm from language models. arXiv preprint arXiv:2112.04359.
11. Liang, P., Bommasani, R., Lee, T., Tsipras, D., Soylu, D., Yasunaga, M., ... & Hashimoto, T. (2022). Holistic evaluation of language models. arXiv preprint arXiv:2211.09110.
12. Srivastava, A., Rastogi, A., Rao, A., Shoeybi, M., & Abbeel, P. (2023). Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. In Advances in Neural Information Processing Systems (Vol. 36).
13. Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., ... & Amodei, D. (2020). Language models are few-shot learners. In Advances in Neural Information Processing Systems (Vol. 33).
14. Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and task motivation: A 35-year odyssey. American Psychologist, 57(9), 705–717.
15. Acemoglu, D., & Restrepo, P. (2019). Automation and new tasks: How technology displaces and reinstates labor. Journal of Economic Perspectives, 33(2), 3–30.
16. Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280.
17. Shapiro, A. (2018). Between autonomy and control: The effects of algorithmic management on platform workers. New Media & Society, 20(10), 3789–3808.
18. Rosenblat, A., & Stark, L. (2016). Algorithmic labor and information asymmetries: A case study of Uber’s drivers. International Journal of Communication, 10, 3758–3784.
19. Irani, L. (2015). The cultural work of microwork. New Media & Society, 17(5), 720–739.
20. Sundararajan, A. (2016). The sharing economy: The end of employment and the rise of crowd-based capitalism. MIT Press.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Global Financial Analytics Research Review
This work is licensed under a Creative Commons Attribution 4.0 International License.
This article is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
