The Johns Hopkins Institute for Assured Autonomy (IAA), in partnership with the Johns Hopkins University Applied Physics Laboratory (APL) and Institute of Electrical and Electronics Engineers (IEEE), is launching the IEEE International Conference on Assured Autonomy (ICAA). The virtual conference will take place March 22-23.
Advances in machine learning and artificial intelligence (AI) have shown promise in automating complex decision-making across multiple domains, including transportation, critical infrastructure and cyber infrastructure. As the algorithms developed for these purposes are integrated into real-world systems, they will require significant support in terms of systems engineering and policy solutions to address the safety, security and privacy issues that will arise.
The ICAA will provide a forum for experts to address the gap between theory and actual implementation of these systems.
Lanier Watkins, the conference chair and a senior cybersecurity research scientist at APL – which leads the IAA jointly with the Johns Hopkins University Whiting School of Engineering – said it can be useful to think of autonomy along a spectrum. A level one system requires direct manual human intervention, and on level four, an autonomous agent creates and approves its own rules.
“Right now, autonomy is seen as a disruptive technology, and everyone wants to be at level four right away,” Watkins said. “But we need to be prepared for what happens when we reach that level – how can you be sure that the machine is really doing what you want it to do, that it’s actually doing something good, that it’s working on behalf of the user?”