Use of artificial intelligence (AI) technologies like machine learning in the public sector is set to explode around the world as governments invest heavily in digital transformation efforts prioritising automated decision-making technologies. South Korea, the European Union (EU) and most recently the USA have announced various funding and grants packages worth billions of dollars as a means to stimulate and futureproof their economies and public services.
NSW’s version of this stimulus, the $1.6bn Digital Restart Fund, will likely also involve greater procurement of technologies that rely on AI. To complement this, the NSW Government’s Artificial Intelligence (AI) Strategy (NSW Strategy) has now been released after being delayed from earlier this year due to the impact of COVID-19.
The NSW Strategy is the first of its kind for an Australian jurisdiction and was preceded by the creation of an AI Ethics Policy outlining principles for use, a User Guide to support agency implementation of AI systems, and the national Australian AI Roadmap developed by the former Department of Industry, Innovation and Science and CSIRO’s Data61.
While the AI Roadmap deals with research and development, building technological capability and potential whole-of-economy benefits, the NSW Strategy is focussed on use of AI in the public sector, particularly to support decision-making in policy and service delivery.
Artificial intelligence features prominently in the cluster-specific ‘Strategy in Action’ directions listed in the state’s Beyond Digital strategy, such as use in transport planning and future use of autonomous vehicles as well as supporting records management systems and providing personalised services.
The NSW Strategy notes the value automating processes could have to improve efficiency of services and free up more time for frontline staff. It intends to balance these benefits with the technology’s potential pitfalls, noting that “AI will not be used to make unilateral decisions that impact our citizens or their human rights”.
Agencies will be held accountable for decisions that make use of the technology, with “quick and efficient” review processes for decisions, and highlighting the need to make decisions understandable to citizens (that is, removing the ‘black box’) are key elements of the NSW Strategy.
These sentiments are common in overseas jurisdictions that have developed strategies and frameworks for the use of AI in public sector decision-making. The April 2019 Canadian Directive on Automated Decision-Making calls for keeping humans “in the loop” for automated decision-making, with assessments that could have a negative impact on customers required to be accountable to individuals or government panels.
Given the high-risk of AI technologies to negatively impact people’s lives and the considerable number of unknowns in further implementation of automation in the public sector, the NSW Strategy will be subject to constant reviews processes as adoption progresses. Again, this is in line with moves made by other jurisdictions such as the views put forth in the European Union Commission’s Whitepaper on Artificial Intelligence.
Based on stakeholder consultation, the NSW Strategy identifies five key themes with associated timelines for related actions to guide the use of AI:
- Building public trust;
- Digital uplift;
- Building data capability;
- Innovation and collaboration; and
“Building public trust” will seek to address ethical concerns, with plans to deliver an online platform to field questions from the public on the technology’s use in Q4 2020. An AI Advisory Council chaired by NSW Chief Data Scientist Ian Oppermann will provide expert advice to agencies and a “repository of reusable patterns” will be kept by the Data and Analytics Centre (DAC) to aid agencies in explaining automated decisions.
AI technologies like machine learning rely on large and accurate datasets to make useable and fair decisions. The NSW Strategy’s “building data capability” actions will further involve the DAC to monitor data governance and policy as government datasets are increasingly subject to automation.
As an emerging field of public sector technology use, the NSW Strategy calls for action to build “innovation and collaboration” with industry and academia, particularly to independently review use of automation in government. A partnership with the University of NSW has already commenced and a second AI Summit is planned in Q4 2021. The first Summit held in December 2019 brought together data scientists and public sector workers to lay the foundations of the inaugural strategy.
Procurement actions include continuing development of the AI User Guide. A potential panel for AI assurance services will be explored as will identifying AI-enabled technologies on the buy.nsw online procurement marketplace, with timeframes ending in Q2 and Q3 of 2021.