Government Use Of Artificial Intelligence In New Zealand PDF Download

"This is the first major report from the Artificial Intelligence and the Law Project. The overall focus of the report is on the regulatory issues surrounding uses of artificial intelligence (AI) in New Zealand. There are many types of AI systems, and many spheres within which AI systems are used (in New Zealand and beyond). Phase 1 of the project focuses on regulatory issues surrounding the use of predictive AI models in New Zealand government departments. As discussed in the report, while there are many types of AI model, the concept of a “predictive model” picks out a reasonably well-defined class of models that share certain commonalities and are fairly well characterisable as a regulatory target. The report specifically focuses on the use of predictive models in the public sector because the researchers want to begin by discussing regulatory options in a sphere where the New Zealand Government can readily take action. New Zealand’s Government can relatively easily effect changes in the way its own departments and public institutions operate. The report identifies and discusses a number of primary concerns: Accuracy, Human control, Transparency and a right to reasons/explanations, Bias, fairness and discrimination, Privacy. Individual rights are vital for any democracy but exclusive reliance should not be placed on individual rights models that depend on affected parties holding predictive algorithms to account. Often, individuals will lack the resources to do so. Furthermore, individual rights models might offer limited efficacy in monitoring group harms. With regard to oversight and regulation, one of the key recommendations of the report is that Government should consider the establishment of a regulatory/oversight agency. Several possible models for the new regulatory agency are proposed in the report. The new regulator could serve a range of other functions, including: Producing best practice guidelines; Maintaining a register of algorithms used in government; Producing an annual public report on such uses; Conducting ongoing monitoring on the effects of these tools. The report indicates preference for a relatively “hard-edged” regulatory agency, with the authority to demand information and answers, and to deny permission for certain proposals. However, even a light-touch regulatory agency could serve an important function. The researchers stress the need for consultation with a wide range of stakeholders across New Zealand society, especially with populations likely to be affected by algorithmic decisions, and with those likely to be under-represented in construction and training. This is likely to include those in lower socio-economic classes, and Māori and Pacific Island populations. Quite simply, they are likely to have insights, concerns and perspectives that will not be available to even the most well-intentioned of outside observers."--Publisher's website.

"This is the first major report from the Artificial Intelligence and the Law Project. The overall focus of the report is on the regulatory issues surrounding uses of artificial intelligence (AI) in New Zealand. There are many types of AI systems, and many spheres within which AI systems are used (in New Zealand and beyond). Phase 1 of the project focuses on regulatory issues surrounding the use of predictive AI models in New Zealand government departments. As discussed in the report, while there are many types of AI model, the concept of a “predictive model” picks out a reasonably well-defined class of models that share certain commonalities and are fairly well characterisable as a regulatory target. The report specifically focuses on the use of predictive models in the public sector because the researchers want to begin by discussing regulatory options in a sphere where the New Zealand Government can readily take action. New Zealand’s Government can relatively easily effect changes in the way its own departments and public institutions operate. The report identifies and discusses a number of primary concerns: Accuracy, Human control, Transparency and a right to reasons/explanations, Bias, fairness and discrimination, Privacy. Individual rights are vital for any democracy but exclusive reliance should not be placed on individual rights models that depend on affected parties holding predictive algorithms to account. Often, individuals will lack the resources to do so. Furthermore, individual rights models might offer limited efficacy in monitoring group harms. With regard to oversight and regulation, one of the key recommendations of the report is that Government should consider the establishment of a regulatory/oversight agency. Several possible models for the new regulatory agency are proposed in the report. The new regulator could serve a range of other functions, including: Producing best practice guidelines; Maintaining a register of algorithms used in government; Producing an annual public report on such uses; Conducting ongoing monitoring on the effects of these tools. The report indicates preference for a relatively “hard-edged” regulatory agency, with the authority to demand information and answers, and to deny permission for certain proposals. However, even a light-touch regulatory agency could serve an important function. The researchers stress the need for consultation with a wide range of stakeholders across New Zealand society, especially with populations likely to be affected by algorithmic decisions, and with those likely to be under-represented in construction and training. This is likely to include those in lower socio-economic classes, and Māori and Pacific Island populations. Quite simply, they are likely to have insights, concerns and perspectives that will not be available to even the most well-intentioned of outside observers."--Publisher's website.