The Royal Civil Service Commission and the GovTech Agency recently published a guideline on the use of generative artificial intelligence (AI) for civil workers in an effort to encourage the ethical and efficient use of AI by civil servants. The guideline also emphasizes how critical it is to handle possible hazards that could jeopardize public confidence and the integrity of governmental procedures, such as bias, misinformation, and privacy issues. The guideline seeks to protect the national interest, ethical norms, private information, and data privacy while providing government servants with the skills and information they need to fully benefit from generative AI. Bhutan’s government has issued an interim guideline for Generative AI usage, referencing AI guidelines from Canada, the UK, Switzerland, USA, UAE, and the European Union Act.
The guideline advises civil servants to use AI tools responsibly, ensuring data is accurate, and not to use them as search engines without verifiable sources. Three main techniques for de-identification are described in the guideline: changing or replacing identifiable information with information that cannot be linked to a specific person; replacing data, like phone numbers or email addresses, with fictitious but format-consistent placeholders; and turning off the “Improve the model for everyone” option in data control settings to stop AI platforms from using the content for training. When employing generative AI in crucial decision-making processes, the guideline also counsels civil servants to exert human oversight, especially in areas like financial management, promotion planning, hiring human resources, and assessing student achievement in educational institutions.