Abstract The application of Artificial Intelligence (AI) in the automotive industry can dramatically reshape the industry. In past decades, many Original Equipment Manufacturers (OEMs) applied neural network and pattern recognition technologies to powertrain calibration, emission prediction and virtual sensor development. The AI application is mostly focused on reducing product development and validation cost. AI technologies in these applications demonstrate certain cost-saving benefits, but are far from disruptive. A disruptive impact can be realized when AI applications finally bring cost-saving benefits directly to end users (e.g., automation of a vehicle or machine operation could dramatically improve the efficiency). However, there is still a gap between current technologies and those that can fully give a vehicle or machine intelligence, including reasoning, knowledge, planning and self-learning. Since a vehicle or machine can be used at different places, routes and terrains, the scope of prediction is most challenging for AI applications in the automotive industry. However, if a machine performs a substantially repetitive work cycle during its operation life, the challenge of prediction can be easily solved by partitioning a work cycle to many discrete segments. Track-type excavators, wheeled excavators, dragline excavators, wheel loaders, wheeled scrapers, and front shovels are all repetitive- work machines. If off-road trucks or articulate trucks only travel between certain locations along fixed routes in a mining site, they can also be regarded as repetitive-work machines. The task of repetitive- work machines requires machines to follow certain operation patterns regardless of terrain. Whenever a machine cycle can be recognized by engine or machine controllers, the operation cost for end users or clients can be dramatically reduced if the AI application strategies are focused on the following four areas: 1. Automation of the current segment of a work cycle 2. Adaptive adjustment according to future events 3. Global or system-level optimization 4. W ork site or fleet management improvement through work group cooperation This paper first reviews the traditional AI applications, and then further explores AI application in the four areas by analyzing the related innovations and technology trends. The AI application strategy that can save the greatest operation cost for end users is illustrated as well. Introduction Many Artificial Intelligence (AI) methodologies have been utilized in a machine system control, such as engine control, transmission control, hydraulic subsystem control and implement subsystem control. The advanced control technologies improve the machine performance and reduce the total life cycle costs of the machine. S. Prabhu has summarized the Neural Network and Fuzzy Logic control applications in an earthmoving machine [ 1]. Among many successful AI applications in machine control, two types of application focus on cost reduction: 1. Neural Network empirical model for powertrain calibration, especially for engine calibration to meet emission regulations. Calibration is a critical process in determining an engine's performance. In New Product Introduction (NPI) projects, calibration is the final step before mass manufacturing. With stringent emission regulations, modern engines have to utilize aftertreatment to abate emissions. With the added subsystems, more Engine Control Module (ECM) parameters are introduced to the engine calibration process. In order to minimize the engine tests that increase exponentially with the ECM inputs, empirical models have been widely adopted in the automotive industry. I. Brahma, et al., applied the neural network models in engine steady-state calibration. [ 2] W. Kurniawan, et al., describes the applicability and capability of the neural network as “an artificial intelligence tool to determine the performance and emissions in a compressed T