A landslide is a downward movement of slope materials that could be triggered by rainfall and earthquake. Having a landslide potential map of an area could be very useful information for geotechnical engineers to provide a risk evaluation of the area. The main types of landslide maps include landslide inventory map, landslide susceptibility map, and landslide hazard map. The landslide hazard map considering triggering factors must be conducted for the risk assessment.

The conventional landslide hazard mapping approach uses geotechnical engineering data to assess hazard areas. However, engineering properties of materials must be obtained to predict the landslide hazard area, which is costly and time-consuming. Recent studies indicate that improvements in analysis methods, especially Artificial Intelligence (AI) and Machine Learning (ML), can be applied to increase the reliability of landslide predictions. This paper discusses the review of the application of AI in landslide mapping.

Landslide mapping using AI could include the following applications: (1) using AI applications for predicting landslide inventory, (2) using AI to assess landslide susceptibility area, and (3) using AI to conduct the landslide hazard map. Research studies show that the most used AI technique in landslide mapping is ANN. Nowadays, using developed ML such as LSTM, MaxEnt, and GBM shows a high success rate. However, the prediction of landslide-prone areas currently depends on spatial environmental variables such as geology, topography, land use, land cover, etc., which are not the geotechnical engineering parameters. The future development of AI and ML in landslide-prone area mapping will focus on how to link between the geotechnical behaviors of geological materials and the spatial environmental parameters, which can help improve the accuracy of the landslide hazard mapping.