South Korea's Ministry of Science and ICT said on March 11 it has secured grounds to 추진 AI-based research administration support services through the Personal Information Protection Commission-led "Public AI Transformation (AX) Innovation Support Help Desk".

The help desk is a channel through which the commission reviews and advises on privacy issues that may arise during public institutions' AX projects. When a public institution applies for a review through the help desk, it links the project to suitable systems and tools such as the prior adequacy review system, the regulatory waiver system (sandbox), one-stop pseudonymisation and proactive legal interpretation. It then jointly reviews the flow of personal information processing from the planning and development stages and co-designs necessary safeguards.

The commission held its fourth plenary meeting on March 11 and approved the results of a prior adequacy review for AI recommendation services for evaluators and collaborative researchers provided by the ministry's pan-government integrated research support system, IRIS, as the first help desk support case.

The ministry introduced AI in developing the service to overcome the limitations of a simple keyword-matching approach. AI that understands context and meaning like a person will analyse a large volume of materials accumulated in IRIS, including papers, patents and research reports. The ministry plans to provide a service that recommends evaluators with expertise best suited to the field and connects researchers with collaborative partners.

The commission reviewed whether the project was appropriate, considering the purpose and intent of the National Research and Development Innovation Act, the legal basis for establishing and operating IRIS, and that researcher data within IRIS were collected and built based on clear laws and regulations including Article 20 of the act.

To reduce privacy risks that may arise in introducing the AI recommendation service, it prepared safeguards with the ministry based on an AI privacy risk management model. Measures include establishing an internal management plan, strengthening transparency and setting up a channel for objections related to recommendations.

The commission will expand support so that public AX moves forward on the foundation of privacy protection. It also plans to promote the help desk and improve on-the-ground accessibility.