(including funding and service delivery).231 The proposed predictive risk modelling programme for the child protection and welfare sector is an example of a specific policy initiative which uses algorithmic techniques to identify risk and target service interventions accordingly. The use of Big Data in this way has significant implications for not only the right to privacy, but also the right to freedom from discrimination due to the inherent nature of the algorithmic approach. The Special Rapporteur on the right to privacy has highlighted a number of well documented risks associated with algorithmic techniques, including that they: • are value laden, Peter Gluckman. 6.4 Artificial Intelligence The development of Big Data and Artificial Intelligence (AI) technologies has occurred hand in hand in recent years. While AI has existed for over sixty years, its development and application over the last ten years has rapidly accelerated due to “better algorithms, increases in networked computer power, and the tech industry’s ability to capture and store massive amounts of data.”234 A recent report by AI Now, a US research institute which examines the social implications of AI, describes how AI has infiltrated every aspect of our lives: • may be based on imperfect or selective data, • may be used for profiling, and • are opaque and unaccountable – it is difficult to attribute responsibility or accountability for harm caused by algorithmic processing.232 The Special Rapporteur further observes that: Recommendations and decisions that result from algorithmic processing appear to spring from an inscrutable and unknowable black box, a kind of twenty-first century Delphic oracle that seemingly makes unchallengeable and authoritative pronouncements divorced from human agency.233 As with surveillance activities by intelligence and security agencies, operational and procedural safeguards in the social sector are an essential bulwark against the risk of human rights breaches occurring and becoming normalised in relation to personal data. The New Zealand Government’s proposed Privacy, Human Rights and Ethics Framework therefore will be a critical frontend procedural safeguard for ensuring that the deployment of algorithmic Big Data techniques in the social sector conform with human rights obligations and accordingly are carried out with the necessary “social licence” identified by Sir 231 Sir Peter Gluckman, Using Evidence to Inform Social Policy: the role of citizen-based analytics. A Discussion Paper, Office of the Prime Minister’s Chief Science Adviser (19 June 2017) p 10. 232 Report of SR on privacy (24 November 2017) paras 57-70. 233 Ibid para 55. AI systems are already integrated in everyday technologies like smartphones and personal assistants, making predictions and determinations that help personalize experiences and advertise products. Beyond the familiar, these systems are also being introduced in critical areas like law, finance, policing and the workplace, where they are increasingly used to predict everything from our taste in music to our likelihood of committing a crime to our fitness for a job or an educational opportunity.235 As discussed earlier in the paper, AI is increasingly used in the criminal justice system. For example, by the police to target resources or highrisk individuals, by the courts to predict the likelihood of re-offending and prisons in targeting restorative justice. The use of algorithmic risk assessments in sentencing was recently challenged in the United States in the case of State v. Loomis in the Wisconsin Supreme Court.236 The Court sentenced the defendant based in part on a tool called COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) that was developed by a private company and purports to predict a defendant’s risk of committing another crime. The defendant appealed the ruling on the grounds that the court’s reliance on COMPAS 234 AI Now 2017 Report https://ainowinstitute.org/AI_Now_2017_Report.pdf. 235 AI Now 2017 Report, p 3. 236 For an overview and analysis of the decision see https://harvardlawreview.org/2017/03/state-v-loomis/ 44