<p>Can artificial intelligence help us decide whether digitalization is ultimately a climate asset or a liability? This paper tackles that question for MENA countries using an explainable, fuzzy-based AI approach. We build a country–year panel from the World Development Indicators (2000 to the latest available year for each economy) and construct a Fuzzy Net-Sustainability Gain (FNSG) index that weighs what ICTs enable against what they cost. On the “enabler” side, the index captures efficiency gains, inclusion and dematerialization; on the “footprint” side, it reflects energy use, material intensity, e-waste, and the associated pressures on natural capital through resource extraction and waste externalities. A transparent fuzzy inference system produces the FNSG scores, while an interpretable machine-learning layer is used to pick up nonlinearities and interactions in the data. Feature-importance and partial-dependence tools then help adjust membership functions, refine the rule base and identify distinct regimes characterised by electricity mix and access conditions. The evidence is clear: ICTs do not have a uniform effect on sustainability. In country–years where power systems are cleaner and access to electricity and digital services is broad and reliable, ICTs are associated with sizeable positive FNSG values. Where grids remain carbon-intensive or access is constrained, much of this potential is dissipated. A Tunisia case study applies the same machinery to actual levels, recent trajectories and simple counterfactual scenarios; joint gains in fixed broadband and the renewable share of electricity deliver the strongest improvements. Taken together, the results show how standard WDI indicators can be reorganised into an interpretable, policy-oriented tool for “greening the digital stack” while making better use of the enabling role of ICTs, and for linking digital transitions to the preservation of natural capital.</p>