Openness on Artificial Intelligence and Algorithms (NO0063)

What problem for the general public does the commitment address? In the future, the public sector will use artificial intelligence to deliver more targeted and user-adapted services, enhance the social benefit of its own activities, rationalise operations and work processes, and reduce risk.

What is the commitment? a. Develop guidance on responsible use of artificial intelligence in the public administration. b. Establish a regulatory sandbox for data protection under the Data Protection Authority to improve the entities’ understanding of the regulatory requirements for data protection and reduce the time from development and testing to roll-out of AI solutions in the market. c. Stipulate that new solutions in the public administration that use artificial intelligence must be transparent and verifiable. d. Amend the Public Administration Act and other relevant regulations to also cover use of automated case processing and artificial intelligence. Develop an analytical platform that can provide better opportunities for analysis of data from Norwegian health registers, whilst still ensuring protection of personal data.

How will the commitment contribute to solving the problem? The use of artificial intelligence in public administration may seem demanding, especially if the organisation does not have adequate in-house expertise. Agencies may be reticent to start using AI systems due to inadequate knowledge among caseworkers about the regulatory frameworks and how to handle personal data, coordination with the Public Administration Act, and assessing when it is acceptable to use artificial intelligence. This can undermine the public sector’s ability to improve its services or rationalise its case processing. Guidance on responsible use of artificial intelligence will make public agencies aware of ethical dilemmas; for example, linked to when it is right and reasonable to use artificial intelligence for verification purposes. The process can be made transparent and verifiable through the development of regulations; for example, by amending the Public Administration Act, which regulates case processing in the public sector, including automated case processing, ensuring compliance with important public administration principles such as data protection and rule of law A regulatory sandbox on artificial intelligence and data protection will provide both public and private organisations with an understanding of the regulatory requirements that apply in respect of data protection, and provide the authorities with greater understanding of new technological solutions and the risks associated with them. Society will benefit from the responsible development of new and innovative solutions.

How is this commitment relevant to the OGP’s fundamental values? Guidance on and regulation of the use of artificial intelligence in the public sector shall ensure transparency in public solutions and decision-making processes that use artificial intelligence. The work on guidance and development of regulations builds on practice and guidelines that ensure the involvement of relevant stakeholders and the general public throughout the entire process. The work on legislation follows the standard rules on public consultation.

Milestones Start date End date Milestone 1) Develop guides Not set Milestone 2) Regulatory sandbox 1 December 2020 . > Milestone 3) Development of regulations, including the Public Administration Act March 2019 – Official Norwegian Report (NOU) delivered . > Milestone 4) Health analysis platform August 2020 (start-up of development project) . >

Commitment 6B. Openness on artificial intelligence and algorithms

· Verifiable: Yes

· Relevant: Access to information

· Potential impact: Minor

· Completion: No evidence available

· Did it open government? No early results to report yet

This commitment was introduced in the amended action plan and focused on the development of guidance on responsible use of artificial intelligence (AI) in public administration; the establishment of a safe technological space for trialling new products (a “regulatory sandbox”) for data protection to better understand regulatory requirements and reduce the time from development and testing to roll-out of AI solutions in the market; establishment of an agreement that AI-based solutions in public administrations must be transparent and verifiable; and an amendment to the Public Administration Act and other relevant regulation to cover the use of AI in automated case processing. The commitment further included the development of an analytical platform to provide opportunities for analysis of data from Norwegian health registers while ensuring compliance with data-protection regulations. The commitment forms part of the existing Norwegian National AI strategy. [39] The activities related to making AI solutions in public administration more transparent through regulatory changes ensure that the commitment is relevant to the OGP value of access to information. The IRM received no response after contacting representatives of the KDD to request additional information about the design and implementation of this commitment. The IRM was unable to find information about implementation.