Lumberjack Tradeoff Model: The Degree of Imperfect Automation

A taxonomy of automation support for human performance is presented that describes support at each of four stages of human information processing at different levels of automation authority. These two dimensions (stages and levels) combining to define the degree of automation. The effect of degree of automation is modeled to show that, mediated by loss of situation awareness, the higher degree of automation, the better performance of the human-AI team when all goes well, but the more problematic is the human response when automation fails. Hence the Lumberjack model: the higher the tree, the harder it falls. Data from a meta-analysis, and from an air traffic control and a NASA robotic arm simulation are presented to support the model, and to show how automation transparency can mitigate the fall of the higher tree.