Designing the Trustworthy Principles of Artificial Intelligence : case: Finnish Police
Lassila, Sanna (2021)
2021
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021110919576
https://urn.fi/URN:NBN:fi:amk-2021110919576
Tiivistelmä
The objective of this thesis was to produce ethical principles of Artificial Intelligence (AI) for the Finnish police. There is a constant pressure to digitalize public sector processes. Due to the increasing amount of digital information and requirements for efficiency, digital processes with AI aided solutions need to be taken in use.
Ethical principles of AI lead the way to the upcoming EU regulation on AI solutions. The principles are needed prior to the regulation to maintain the high level of trust that the Finnish police enjoys among the citizens.
The development task was to produce ethical principles that would support the needs of the police by taking into account the operating environment with applicable national laws and current regulations, while also considering the upcoming EU regulation.
Service design methods were used to design principles that would fit into the organizational processes and to the operative environment of the police. The methods included prototyping, in-depth interviews, and cooperative workshops. The development was done in phases. The first draft was sketched based on principles and requirements presented in the Ethics guidelines of trustworthy AI by European Commission. The first version evolved to second version by self-iteration, but the following versions got enhanced by interactions between different stakeholders.
The key findings of this thesis were that the service design methods used apply well in this kind of development of processes concerning a reasonably new topic. A tangible product, a prototype, serves as a basis for discussion and gives the interviewees or attendees of workshops a vision of the outcome, something to build on.
As the discussion of ethical use of AI is ongoing, the Finnish police wanted to act as a pioneer in law enforcement and publish their ethical principles of AI. The outcome of the work is a description of trustworthy principles of AI. The principles describe what is meant with AI in the police and how AI systems are used and what kind of measures implement the ethical application of artificial intelligence systems. Besides the ethical aspects, the trustworthy principles consider compliance with law and reflect the requirements of robustness.
AI systems are not ethical or unethical themselves, the issue concerns the use of the systems. Ethical principles guide the development, implementation and use of the systems. To apply the principles in practice, their requirements need to be communicated inside the organization and implemented to organizational processes. The implementation of such a small organizational change in practice is implemented and monitored in accordance with suitable change management processes.
Ethical principles of AI lead the way to the upcoming EU regulation on AI solutions. The principles are needed prior to the regulation to maintain the high level of trust that the Finnish police enjoys among the citizens.
The development task was to produce ethical principles that would support the needs of the police by taking into account the operating environment with applicable national laws and current regulations, while also considering the upcoming EU regulation.
Service design methods were used to design principles that would fit into the organizational processes and to the operative environment of the police. The methods included prototyping, in-depth interviews, and cooperative workshops. The development was done in phases. The first draft was sketched based on principles and requirements presented in the Ethics guidelines of trustworthy AI by European Commission. The first version evolved to second version by self-iteration, but the following versions got enhanced by interactions between different stakeholders.
The key findings of this thesis were that the service design methods used apply well in this kind of development of processes concerning a reasonably new topic. A tangible product, a prototype, serves as a basis for discussion and gives the interviewees or attendees of workshops a vision of the outcome, something to build on.
As the discussion of ethical use of AI is ongoing, the Finnish police wanted to act as a pioneer in law enforcement and publish their ethical principles of AI. The outcome of the work is a description of trustworthy principles of AI. The principles describe what is meant with AI in the police and how AI systems are used and what kind of measures implement the ethical application of artificial intelligence systems. Besides the ethical aspects, the trustworthy principles consider compliance with law and reflect the requirements of robustness.
AI systems are not ethical or unethical themselves, the issue concerns the use of the systems. Ethical principles guide the development, implementation and use of the systems. To apply the principles in practice, their requirements need to be communicated inside the organization and implemented to organizational processes. The implementation of such a small organizational change in practice is implemented and monitored in accordance with suitable change management processes.