top of page
  • Writer's pictureJuno

Can the next great thing come from taking great care of millions of small details?

Updated: Aug 8, 2023

A new report by Juno sets to find out

The Juno Evidence Alliance is creating an ambitious State of the Field of Agricultural Research evidence synthesis report. Image generated by AI

How can we provide impartial and data-driven advice on priorities for the future in a way that helps all stakeholders—governments, funders, researchers, and even farmers—understand and learn from what’s happened in the past?

A new landscape report produced by Juno – the State of the Field for Agricultural Research – is an initial step towards helping us answer that question.

The report will use artificial intelligence to assess the contributions of millions of scientific and development papers across the agrifood system.

The report aims to lay the groundwork for opportunities to evaluate science and development’s contribution towards global goals on a scale that we we have previously been unable to assess, using reproducible and transparent methods that will allow the report to be updated and improved upon over time, with open protocols and search strategies.

A pilot study using similar approaches was conducted in 2022 on more than one million papers focused on small-scale farmers across countries located in Sub-Saharan Africa. The study found that there are significant gaps in the research on interventions that focus on outcomes that would improve women’s empowerment, agency and inclusion, and work focused on nutrition-sensitive agriculture.

A machine modelled mapping of interventions and outcomes using 1.2 million summaries on agricultural research

The State of the Field Report will use a significantly larger dataset. Until now, such ambitious evaluations have been outside of our grasp because they require immense computational power, and the availability of massive amounts of high-quality, representative data.

By combining expert and machine intelligence – and in particular large language models (LLMs) – Juno will analyze vast amounts of text-based data quickly, efficiently, and in the proper context.

Schematic of how the Juno Evidence Alliance uses natural language processing (NLP) along with human-AI collaboration to inform evidence-based policy decisions

AI-driven assessment of a representative, global dataset, paired with a unified framework informed by an advisory panel of experts, will form the foundation of this “State of the Field” report, as well as a variety of real-world evidence assessments commissioned by the UK Foreign, Commonwealth & Development Office (FCDO).

We look forward to sharing the findings and using them to help inform global goals and agendas.

165 views0 comments

Recent Posts

See All
bottom of page