[Chodong Perspective] Understanding Civil Servants' Mindsets Is Key to Opening AI Data

'Public Data,' the Key to Fostering the AI Industry
Civil Servants Avoid Disclosure Over "Personal Information Concerns"
Incentives Needed to Motivate 'Nakjibudong' Bureaucrats

'The number of patients who received duplicate prescriptions for dementia medications and the number of deaths, as well as the average age of the deceased, from 2010 to 2011.'

Is this data, held by the National Health Insurance Service, subject to disclosure under the Public Data Act? Since the request does not involve the prescription records of specific individuals, it seems obvious that such data should be accessible. However, in reality, the agency refused to submit the data. The reason was not the absence of the data. The agency explained that "in order to prevent identification, the data would need to be processed, which would require significant time and effort, making it difficult to provide." The requester, unable to tolerate this, eventually filed a lawsuit at the Seoul Administrative Court in August last year.

This case raises concerns because the opening of public data plays a decisive role in fostering the domestic artificial intelligence (AI) industry. In Korea, where the population is relatively small and securing large-scale datasets is challenging, utilizing public data is essential to develop advanced industries such as AI-based healthcare and autonomous driving. This understanding has been shared since the Park Geun-hye administration promoted the concept of a "creative economy." Since the enactment of the Public Data Act in 2013, even after three changes of administration, the initiative for data disclosure still remains with the providers?government officials?rather than the users. Ministries occasionally release performance reports highlighting achievements such as "exceeding public data disclosure targets" or "surpassing tens of thousands of downloads," but when it comes to data, quality is more important than quantity. The fact that the Digital Platform Government Committee admitted at its launch in 2022 that "the disclosure rate of key public data is only 10%" illustrates this point.

No administration has made significant progress, so the opening of public data remains a key issue for the current government as well. At the Ulsan Data Center launch event last month, industry leaders voiced their concerns: "Please support data in specialized fields such as healthcare" (Seo Bumseok, CEO of Lunit), and "There is a severe lack of national-level datasets" (Chung Shina, CEO of Kakao). Currently, the Presidential Office, led by AI Chief Ha Jungwoo, is preparing technical support measures.

However, public data cannot be easily opened based solely on technological readiness. It is necessary to examine why, despite pressure from as many as three presidents over more than a decade, government officials have resisted disclosure. The aspect they are most sensitive about is public data containing personal information. According to a survey by the National Information Society Agency (NIA), about half of the cases of refusal to provide data were due to the Personal Information Protection Act. From the perspective of civil servants, no matter how much personal information is processed, the potential legal liability that may arise in the future is the greatest burden.

One high-ranking official explained, "Imagine I am in charge of insurance at the Korea Post. There are millions of policyholders, and if an AI company requests subscriber information to develop an insurance-related algorithm? Of course, the law requires disclosure, but if I refuse citing personal information, that's the end of it. There is no benefit to providing the data, and if problems arise later, I would have to face an audit by the Board of Audit and Inspection and even legal penalties. Every civil servant in every ministry likely feels the same way."

'Nakjibudong.' This is a neologism that goes a step further than 'Bokjibudong.' It satirizes bureaucratic organizations that, like an octopus clinging to the floor with its suckers, refuse to budge. To encourage risk-averse civil servants to take the initiative in opening public data, it is important to provide motivation. First, the structure in which the information officer?usually a non-mainstream position in most ministries?handles public data exclusively must be changed. It is essential to establish a system where the public AI officer, as proposed by Chief Ha, leads public data tasks and proactive civil servants are rewarded for their efforts. A situation in which government officials do everything possible to withhold data, while users fight to obtain it in court, is a waste of social resources. A government aspiring to become a global AI powerhouse cannot afford to squander time on such unproductive conflicts.