Does AI Need Governance? – The Potential Roles of a ‘Responsible Innovation Organisation’ in Australia (Submission to the Australian Human Rights Commissioner on the White Paper Artificial Intelligence: Governance and Leadership)

Abstract

The purpose of this submission is to address the list of questions posed by the Australian Human Rights Commissioner on the White Paper Artificial Intelligence: Governance and Leadership (2019). It is an interim submission, and a final submission will be made later. Some main points of the interim submission, stated in summary, are: 1) The main goals of government regulation in the area of artificial intelligence include: a. ‘Regulation’ must not be confused with economic enablement and stimulation. Both may be legitimate in relation to AI, but the same organisation must not be given responsibility for both, as they involve an inherent potential for conflict of interests. b. Attempts to ‘pick winners’ or to encourage governments to do so should be avoided. Focus on principles, not products. c. Encourage Australian input into the global development of standards for, ethics of, and treaties concerning, AI development and use. d. Legislate where necessary to prevent the most harmful uses of AI. e. It should be assumed that some regulation (possibly including co-regulation) will be necessary. f. The general objectives of the regulation of AI are primarily the protection of human rights and the protecting of competition. 2) There is no Australian policy-oriented organisation with a permanent high level of expertise in AI-related technologies, and certainly no such organisation which does not have potentially conflicting industry development objectives. 3) There would be value in Australia creating such a new policy-oriented organisation, with a limited life-span, but ‘Responsible Innovation Organisation’ (RIO) is a completely inappropriate title, as it suggests that the body has a role in promoting innovation (ie industry development), either by itself or as part of some larger economically-oriented organisation, which will inevitably dominate and distort its views. We should consider approaches taken by other national governments. 4) Many Australian organisations would find it valuable to have an Australian policy-oriented organisation recommending international standards for Australian adoption, or assisting in the developing of Australian standards. 5) Strictly econometric measures of the ‘business case’ for such policy-oriented organisations are not possible or desirable. 6) The desirable goals for such an organisation are set out in (i) above. a. It should primarily encourage and assist other Australian organisations to provide inputs into development of standards and ethics (internationally and in Australia), including by funding assistance where necessary. It should have the capacity to make such inputs itself, when needed. b. A board of part-time Commissioners, with one full-time Commissioner, is desirable. It should be a stand-alone organisation, as it would not fit comfortably within other organisations c. Public funding is more desirable, as the costs will be modest. Joint funding by industry prevents very grave risks of regulatory capture. d. It should provide an annual report to Parliament, and communicate its work between reports via a variety of Internet channels.