← Back to Publications
Why AI in Agriculture Is Harder Than It Looks
Published
Dr. Saleh Alwer · ISDI Article 2026/46
Summary

Artificial intelligence holds real promise for Jordanian agriculture — smarter irrigation, earlier disease detection, better yield forecasting, and more efficient resource use. But the gap between what AI can do and what Jordanian farms can actually use is wide, and it isn't a technology problem so much as a structural one. Three barriers dominate. First, Jordan's farms are small, so the fixed costs of sensors, controllers, and software fall on producers who lack the volume to spread them out — meaning even a technically sound system can fail to pay for itself. Second, reliable local agricultural data is scarce: irrigation, soil, pest, and yield records often don't exist in usable digital form, and models built on European, North American, or Chinese datasets don't transfer cleanly to Jordanian crops, soils, and climate without local retraining. Third, every new layer of technology brings a new layer of upkeep — installation, calibration, connectivity, monitoring — that small and mid-sized farms typically can't staff internally, pushing them toward costly outside support. These problems compound each other: thin margins limit investment in data infrastructure, weak data limits localisation, and localisation demands more technical support than most farms can absorb. None of this is a dead end. The piece argues for spreading fixed costs through cooperatives and shared infrastructure (regional weather networks, shared sensors, public satellite data); closing the data gap by fine-tuning existing models with smaller local datasets rather than building from scratch, backed by ministry- and university-led data collection; adopting technology incrementally to limit risk and build trust; and closing the skills gap through extension services and local technicians who turn maintenance into a nearby service rather than an imported speciality. The bottom line: AI's success in Jordanian agriculture won't be decided by model accuracy alone. It depends on building an affordable, locally adapted technical ecosystem around the model — one that delivers real economic value and puts these tools within reach of the small farms that make up most of the sector.

Subscribe to our newsletter
Stay up to date with ISDI's latest research, publications, and events — delivered straight to your inbox.
We respect your privacy — no spam, unsubscribe anytime.