Evaluating HSI with AI-Enabled Systems: What should you consider in a TEMP?
Abstract:
The Institute for Defense Analyses and the Joint AI Center (JAIC) are developing a framework to support testers as they evaluate Human Systems Integration (HSI) with AI-enabled systems. This product is a short briefing that gives an overview of the framework, conveys a sense of what the fuller framework looks like, and ends with an example of how increased autonomy in aviation led to the loss of Air France 447. We are creating this framework because HSI expertise in the DOD Test and Evaluation workforce is rare and increasingly important with more automation in systems. This briefing focuses on the portion of the framework which introduces non-experts to core HSI concerns to help testers write and assess TEMPs. This briefing covers a subset of HSI topics included in the framework, including mental models, trust, workload, and function allocation. HSI will continue to be a vital component of creating and adopting effective, suitable, and trustworthy AI-enabled systems.