In 2017, the 3A Institute was created as the first of the innovation institutes at the Australian National University, intended as structures adjacent to established research and teaching programs to explore new education models and pathways towards the application of research.
3Ai focused on guiding and accelerating into existence a new branch of engineering centred on cyber-physical systems and artificial intelligence. We are building the skills and knowledge we need to help shape the future safely, sustainably and responsibly.
In 2021, the 3A Institute graduated from an ANU Innovation Institute to the ANU School of Cybernetics. The School of Cybernetics now houses the Master of Applied Cybernetics, which was established and piloted at the 3A Institute.
At the ANU School of Cybernetics we continue to explore new education models and pathways towards the application of research, and our goal remains to to help shape the future safely, sustainably and responsibly. Our way into this goal is cybernetics, and the work of the 3A Institute is what has led us to this framing.
In particular, at the ANU School of Cybernetics we continue to reference and refine the ways the the 3A Institute sought to ensure the safe, sustainable and responsible development of cyber-physical systems, including through 6 core themes that form the beginnings of a critical practice of engineering. These themes, and associated questions, look beyond the software and hardware in the lab, and contemplate what happens when technologies are let loose on the world – are taken up by many different organisations and governments, are connected to other intelligent systems, and go to other places in the world from where they were designed.
Founding themes and questions of the 3A Institute are elaborated below.
Questions about autonomy aren’t just technical questions like “how do we make the car brake when it needs to?” – they are also social, regulatory and public policy questions, which will be different for each use and context. Just because we can make a system autonomous, should we? How do we think about degrees of autonomy? How is an autonomous system different to an automated system? When we translate human processes into autonomous processes, what do we need to consider?
As the ability for machines to act independently of human oversight increases, questions need to be asked about how much agency we give cyber-physical systems. To what degree should the system be able to make decisions without human intervention? If a system learns, should we allow it to independently change its behaviour? And if so, does the system bear any responsibility for its actions? What is the override system and who has access?
Technological progress necessitates new regulatory tools and processes, as systems designed in one place are introduced into different settings around the world, with access to new data. How do we preserve our safety and values? What are the mechanisms for assessing and managing safety, security and policing concerns? How do we reassure users and the wider community that the system is safe? To what laws and regulations is the system subject to, and are they adequate?
3Ai was founded on these 3 questions-–-and we have since imagined more. And will likely imagine more still!
Interfaces: throughout the 20th century we used keyboards and screens as interfaces with computers. What interfaces will new CPS use? How will people know they are interacting with an AI system? Does the type of interface change our relationship with technology? What’s next beyond the field of Human-Computer Interaction?
Indicators: the world is changing. Commercial measures of success such as productivity and efficiency used without reference to safety, sustainability and responsibility rarely cut it anymore. What measures of performance and success should we use for CPS and AI? How do we account for unintended consequences? How might we change the intent of the system through our choice of metrics?
Intent: the last set of questions, but in some ways the first. A focus on safety, sustainability and responsibility means that we must deeply interrogate the intent – explicit and implicit – in building a system. Why does this CPS exist? Who does it serve? What is the power relationship between those who commission it, those who design and implement it and those whose lives are impacted by it?
As we stand on the brink of cyber-physical systems at scale, we start by asking the right questions.