독일에서 공법상 인공지능(AI) 활용의 억제와 육성 논의

The global debate over the curbing and fostering of AI is at a fever pitch, as evidenced by the recent firing of Sam Altman, the father of ChatGPT, by the OpenAI board, only to have him reinstated five days later. This is due to a combination of anxiety and excitement about the future of AI, which is currently unclear. However, there is no stopping the development of AI, as it will bring enormous wealth in the near future. Just as the amendments to the AI Law adopted by the EU Parliament are more focused on restraint, so too is Germany, with Article 22 of the GDPR, Article 35a of the Federal Administrative Procedure Act, and state laws in the states of Bayern and Schleswig-Holstein to minimize concerns about violations of individual rights. One of the main arguments for fostering AI in public administration is the fairness and predictability of judgment. Of course, AI is also at risk of errors, such as discrimination. However, this can be prevented through algorithm design, which is better than human subjective judgment, which cannot be controlled and guided by the right algorithm. The fairness of judgment is ultimately a matter of value judgment, so it comes down to a matter of discretion, which is ultimately the same as judicial judgment. As a prerequisite for AI judgment, "hyperconnectivity of information" and "transparency of AI utilization" are necessary. The former has a big problem of conflicts with data protection. The latter encompasses the disclosure of the facts of AI utilization to the relevant parties and the explanation of the AI algorithm protocols that led to the final conclusion. These two challenges can only be solved by "technological progress". Effective national support is needed, such as the EU AI law's “AI labs” (which will also serve as a kind of AI regulatory sandbox), which will allow countries to provide a controlled environment for a limited period of time to develop, test, and validate innovative AI systems before they go to market.