Sovereign AI and human oversight key to secure digital government

To build trustworthy, inclusive digital public services, Manon Lafitte emphasizes the importance of genuinely engaging citizens in decision-making. This process can be challenging, especially as nations like Vietnam accelerate their digital transformation. However, both citizens and public officials must participate in the decision-making process and the digital transformation journey.

Public officials should be involved in determining needs and expressing concerns to integrate into digital projects effectively. Transparency is another crucial principle, encompassing both the government's digital strategy and the algorithms used in the digital transformation process. Transparency can be achieved through the open sharing of data, enabling effective AI strategies and fostering the development of new services.

Open data has proven successful in France, providing a strong foundation for AI development and economic growth. By making public data open, government agencies can unlock economic value and encourage the creation of new services. For instance, data on weather, mobility, and transportation can be repurposed to develop innovative solutions.

To ensure digital transformation remains inclusive, governments and businesses should carefully evaluate their AI usage. A review should identify who is using AI, the purposes for which it is employed, and whether the tools are sovereign AI systems or unregulated commercial AI tools. This evaluation will help establish a management map of AI usage.

Developing in-house sovereign AI models can replace the indiscriminate use of external platforms, reducing risks associated with information security and data confidentiality. Stricter oversight mechanisms are essential to prevent AI hallucinations, which can generate inaccurate responses or contain prejudice and bias. France has a dedicated body that audits AI solutions and algorithms to ensure transparency, explainability, and traceability before deployment.

Maintaining human review, oversight, and approval of AI models is vital to guaranteeing full control over information systems. To promote inclusivity in digital transformation, establish common standards and promote a shared ethical framework among citizens. Public officials should provide clear guidelines and accessible charters for everyone.

Additionally, making public websites accessible to people with disabilities, supported by legislation on online public services accessibility, ensures that digital public services are accessible to those who need to use them.