How to end hunger and achieve food security within the next decade? To address this question, top key players in the United Nations’ Sustainable Development Goals initiative (SDGs) have recently teamed up in a joint research effort, summarising the conclusions of over 50.000 datasets of grey literature on agricultural practices and development interventions, aimed at reducing hunger worldwide. Unlike previous initiatives of the same kind, though, this particular research project has seen for the first time a massive use of artificial intelligence (“AI”) technologies, in particular, natural language processing (“NLP”) and machine learning.
In what turned out as a coming together of two separate projects, on the one hand, the “Ceres2030: Sustainable Solutions to End Hunger” initiative led by The Cornell University, the International Food Policy Research Institute (“IFPRI”) and the International Institute for Sustainable Development (“IISD”), and, on the other hand, an independent research from the Center for Development Research (“ZEF”) and the Food and Agricultural Organization of the United Nations (“FAO”), new methods of discovery were deployed into function to overcome the painstaking process of reviewing literature and account for the inefficiency of keyword search, against the bulk of information available.
The results gathered from a sample of bibliographic records of assorted journals and reports from 2008-2018 on rural development, which the Ceres2030 group subsequently made available on the Airtable platform, were nothing short of staggering: having digested this input, and recognised patterns in the relationships between ways to describe similar contents, the AI indicated that humanity has indeed a chance to truly put worldwide hunger to an end, if interventions on agricultural R&D, technology and innovation are properly targeted and scaled up to be cost effective. To do so, however, the world needs to reportedly kick in $14 billion per year during the next decade, this being an amount of money which is roughly double to what the world currently spends on aid for food security and nutrition.
The AI has not limited itself to merely forecast monetary figures or passing along to some generic advice: instead, it has provided researchers with specific direction on how to deal practically with the issues at stake: as an example, to help farmers in developing countries breaking the vicious cycle of low-intensity, subsistence-oriented farming, the machine learning analysis suggested specifically to invest in livestock, in order to increase overall productivity, and to improve access to mobile phone data networks, so as to minimise runoff and waste, vis-à-vis sudden climate and weather variations. In a press statement by the FAO of 12 October 2020, Maximo Torero, Chief Economist at the organisation, was evidently eager to embrace all these findings and indications, thus urging not only a call to action to rich countries to double their aid commitments, but also identifying clear targets for change to eradicate hunger and improve worldwide nutrition:
“If rich countries double their aid commitments and help poor countries to prioritize, properly target and scale up cost effective interventions on agricultural R&D, technology, innovation, education, social protection and on trade facilitation, we can end hunger by 2030.”
According to Ceres2030, the machine learning process took about a whole week to pare down the numerous dataset of grey literature, to those who were actually of any use; one problem slowing down the process reportedly lied in the inefficiency of systems in use by a number of organisations, which regrettably lacked a number of basic features and accessibility functions, such as that, e.g., of selecting and downloading multiple citations. Having gotten my hands dirty with grey literature at the IAEA, in the past, leading technological innovation in the Systems Development and Systems Group (“SDSG”) to leverage the International Nuclear Information System (“INIS”), I certainly admit that a lot of turnover remains to be done, both at the technological level as well as the managerial one, so as to streamline a modern vision and approach to grey information, which is able not only to ensure access to modern-day machine learning techniques and methods, but also to successfully overcome the rigors of organisational resistance to change.
In conclusion, this story remains not only as a witness to the ongoing dedication of a number of key players to United Nations’ Sustainable Development Goal 2, aimed at achieving “zero hunger”, but also as a witness of the readiness of computer assisted technologies at offering a variety of pattern recognition methods, which can be ultimately applied immediately, and practically, in a number of heterogeneous fields, including that, for example, of grey literature, which traditionally deals with high-dimensional data with a large search radius. In the future, the ability of entrepreneurs at seamlessly incorporating these technologies in the daily operations of a company could ultimately mean making or breaking a business; but for nonprofit organisations such as the FAO, it could actually mean much more, as much as changing the lives of millions and, thus, the world itself.