AI driven detection of soil nutrient deficiency and technogenic contamination: A Systematic Literature Review with emerging pathways for sustainable land use

Purpose: With millions of hectares of farmland in Vidarbha already affected by erosion, salinity and declining soil nutrition, the region faces a critical, worsening threat - technogenic contamination from electronic waste (e-waste) and other industrial waste adds further complexity. Nagpur is identified as one of India’s top 10 e-waste generating cities. Artificial Intelligence (AI) offers a transformative potential to identify and isolate contaminant chemicals while assessing soil nutrition, yet adoption among rural farming communities remains constrained by a persistent digital and knowledge divide. This review systematically examines literature on key soil contaminants, AI-driven detection of hazardous chemicals, and recommended mitigation pathways toward sustainable agricultural land use. Method: A systematic search across Scopus, Web of Science, IEEE Xplore, arXiv, Elsevier, and Google Scholar used terms spanning remote sensing, technology adoption, e-waste, soil contamination, digital agriculture and socio-cultural barriers. Of 169 records identified, screening against inclusion/ exclusion criteria focused on Indian rural contexts resulted in 124 studies for PRISMA- based thematic synthesis. Research: Vision-based AI/ML in soil nutrient grading, contamination profiling, remote sensing and soil health advisory does show strong technical promise; however, adoption is severely constrained by traditional resistance, high initial investment, low contamination - risk awareness and gaps in digital literacy. Partnerships through Farmer Producer Organizations, government initiatives and localized voice-interface tools emerged as effective adoption enablers. Conclusion: The review synthesizes literature-based recommendations and identifies systemic gaps – in policy enforcement, recycling infrastructure, as well as farmer-level digital literacy – that must be addressed to prevent further, compounding land degradation. The evidence points to decentralized, offline-capable AI tools deployed through trusted local intermediaries as the most consistently recommended pathway.