Advancing Innovative AI Applications from Concept to Deployment: Sandboxing, Piloting, Policy Experimentation and Becoming an “AI-First” Nation
Instructor: Steve Miller, Professor Emeritus of Information Systems, School of Computing and Information Systems, Singapore Management University (SMU).
This course provides guidance to decision-makers in public and private sector organizations who are already involved in or considering AI-based digital initiatives, and policymakers who are interested in the implications of these initiatives. The course will enable the participants to grasp lessons learned from both successful and unsuccessful past digital initiatives that have been launched in a diverse set of countries, from least developed countries to those with highly developed economies. The course will help participants better appreciate the issues of translating high-level objectives for AI-based systems into practice, and to better understand the process of introducing innovative and potentially impactful new AI-based systems through a structured series of phases. While the primary focus of the course is on advancing innovative uses of AI towards deployment, the course will also address important broader implications. This includes how organizational leaders and policymakers can build the capability for public and private sector AI, and advance towards becoming an “AI-first nation,” and the impact that AI deployment can have on jobs and employment.
Emerging products and services based on artificial intelligence (AI) bring tremendous opportunities for both the public and private sectors. AI can be transformative, improving the cost and effectiveness of an organization’s existing capabilities, and unleashing new ones. However, adoption of AI is challenging, especially in government agencies and large organizations, because the speed of AI capability development far outpaces the speed of traditional policymaking. This is compounded by the scale and reach of AI’s potential impacts, including both positive impacts such as productivity improvement and enablement of new and better services, and negative impacts such as amplifying existing risks and creating entirely new and unanticipated problems.
It is possible to benefit from rapidly changing technology through a multi-phased approach that includes technical testing, sandboxing, piloting as part of an overall process to validate and evaluate the performance and broader impacts of these systems in the context of real-world use. The goal is not just to determine whether the AI-based technology works from a technical perspective, but to ensure that the entire system meets the needs of its internal and external users, the needs of these with responsibility for accountability and broader outcomes, and other stakeholders, under real-world circumstances and settings.
An important topic that will be discussed in depth are the steps that need to occur between the very early stages of technical testing and the eventual desired stage of operational deployment at scale. This includes an in-depth treatment of sandboxing and piloting, and the distinctions between these two types of efforts. To varying degrees and with different types of constraints, sandboxing and piloting allows for the live testing of technologies, services, business models and user behaviors in the real use case setting with real users. Sandboxing in particular benefits from relaxed or flexible regulatory requirements, often at a smaller scale or limited timeframe and with safeguards in place. Sandboxing lays the groundwork for a more reliable and innovative pathway forward for subsequent phases of pilots of increasing scope, eventually leading to a more reliable widescale deployment and adoption while limiting risk.
The course will then use real-world examples to illustrate how one might seed and progressively expand AI capability development in ways that make it possible for an individual organization or a national ecosystem of organizations to level up their ability to understand AI and make use of AI in effective ways. This includes efforts to spur industry adoption and talent development, using approaches such as AI apprenticeship programs and use of an AI project management framework. The course will also include a discussion of how AI can affect employment, and the future of work, including how AI can both automate existing jobs and augment existing jobs, thereby reducing employment opportunities in some areas while simultaneously expanding employment opportunities in others. The course will be delivered in a highly interactive way to maximize participant engagement and contributions. This makes for an active learning environment and maximizes the amount of peer-to-peer learning from the other professionals participating in the course.