Name in original language
Guía de prompting para la gestión pública
Initiative overview
The Prompting Guide for Public Administration was developed to support professionals across Argentina’s National Public Administration as the use of generative AI tools rapidly expanded within government. As public servants increasingly rely on large language models for drafting, analysis and information processing, the guide provides a shared institutional framework to ensure that prompting practices are safe, effective and legally compliant.
The guide addresses both skills development and governance. It introduces a practical, structured methodology for prompt design—most notably the ROCLET framework (Role, Objective, Context, Limitations, Style and Output type)—to improve the quality, clarity and reproducibility of AI‑assisted work. At the same time, it establishes clear safeguards covering data protection, confidentiality, intellectual property, transparency and bias mitigation. AI‑generated outputs are explicitly defined as preliminary inputs that support, but do not replace, human judgment or legal, technical and administrative responsibility.
Although the guide has not yet been formally published, its expected impact is clear. It is designed to strengthen institutional AI literacy, standardise prompting practices across agencies and reduce risks such as data exposure, hallucinations and overreliance on automated outputs. By embedding mandatory human supervision and traceability into AI‑assisted workflows, the initiative aims to increase confidence, accountability and compliance with Argentina’s legal and administrative framework.
Key challenges emerged during the design phase. Balancing innovation with strict legal and security requirements—particularly around data protection and institutional responsibility—required careful framing. Another challenge was ensuring that the guide discourages dependency on AI outputs while promoting their responsible use. The main lesson learned is that practical, operational methodologies are more effective than abstract principles alone. Clear boundaries, institutional endorsement, targeted training and regular updates are essential for sustainable adoption.
The central purpose of the guide is to professionalise how public servants interact with AI systems, enabling consistent, transparent and responsible use across government. Following formal publication, it is expected to be institutionalised through internal dissemination, training programmes and integration into broader digital transformation and AI‑governance initiatives. Over time, the guide may be updated to reflect technological advances, regulatory changes and feedback from practical implementation.
