Aditi Chugh

Impacts of personalized picture-based crop advisories: Experimental evidence from India and Kenya

The rise of artificial intelligence (AI) has heightened interest in digital models to strengthen agricultural extension. Such tools could help provide personalized advisories tailored to a farmer’s unique conditions at scale and at a low cost. This study evaluates the fundamental assumption that personalized crop advisories are more effective than generic ones. By means of a large-scale randomized controlled trial (RCT), we assess the impact of personalized picture-based advisories on farmers’ perceptions, knowledge and adoption of recommended inputs and practices, and other downstream outcomes. We find that personalizing advisories does not significantly improve agricultural outcomes compared to generic ones. While farmers who engage relatively more with advisories (i.e., those who receive and read a substantial number of messages based on self-reports) tend to achieve better outcomes, this is irrespective of whether the advisories they receive are tailored to their specific situation or not. We conclude that investments in digital extension tools should aim to enhance engagement with advisories rather than focusing solely on personalization.