A multi-stage fuzzy inference system (FIS), a symbolic knowledge-based artificial intelligence technique, is used to delineate exploration targets for rare earth elements (REEs) associated with carbonatite-alkaline complexes in NE India. A conceptual REE mineral systems model was used to identify the following targeting criteria for REE deposits. The multi-stage FIS was structured based on the mineral systems model. The first stage of the multi-stage FIS comprised of three individual FIS to represent (1) plume-metasomatised SCLM in an extensional regime that make up fertile source regions for REE-bearing fluids and favourable geodynamic settings; (2) trans-lithospheric structures that provide favourable lithospheric architecture for the transportation of REE-enriched alkaline-carbonatite magma and (3) near-surface higher-order structures that make up a shallow crustal architecture facilitating emplacement of alkaline-carbonatite complexes. The targeting criteria were represented by their spatial proxies in the form of GIS layers derived using spatial analyses and geoprocessing tools for inputting to the FIS. The outputs of the FIS were mapped to generate prospectivity maps that were analysed to identify exploration targets for REE in the study area. The uncertainties in the outputs of the FIS were quantified using Monte-Carlo-based simulations. Exploration targets at low uncertainty levels were delineated around Sung valley and Jasra carbonatite-alkaline-complexes. Areas around the carbonatite-alkaline complex around Swangkre and to the south of the Nongstoin town were identified as high-uncertainty targets. It is recommended that ground follow-up exploration should be carried out in the former targets, and more data should be collected to increase confidence in the latter targets.