Governments and tech companies are pouring money into artificial intelligence without knowing whether the research actually addresses what people need. This project will map the real-world trajectory of AI development—using agriculture as a test case—to reveal where priorities and public value diverge. The problem is a blind spot. Current discussions about “responsible AI” lack evidence about what AI research is actually doing, making it hard for funders and regulators to steer it toward societal benefit. The researchers will combine publication analysis, expert surveys, and text mining to trace how AI is reshaping agricultural R&D, then compare that against the needs of groups in the Global South often excluded from rich countries’ AI race. If successful, the work could give governments and international bodies a practical tool for spotting mismatches between research investment and social needs—not just in agriculture, but across AI more broadly. It would help funders decide where to put money and regulators see where oversight is needed. The project is methodological and diagnostic, not a technology demonstration; its impact lies in making an opaque system legible.
View original technical description
As the momentum behind AI builds, discussions about “Responsible AI” and “AI for good” are being taken increasingly seriously. We are seeing new governance structures and policy priorities. And yet we know little about current trajectories of AI research and innovation, making it hard to anticipate funding or regulation needs. The promises of AI have not yet been systematically tested against societal needs. There is a shortage of evidence that might connect AI with questions of public value (Bozeman and Sarewitz 2011). Recent research (e.g. Ciarli and Rafols 2019) has shown the potential for innovation studies to illuminate gaps between research priorities and societal needs. Our project will explore and explain the potential of mapping emerging AI trajectories. Our analysis will allow us to assess the feasibility of some strategies for understanding the values currently implicit in AI research and development and the gaps with public values. AI is hard to define, hard to see and hard to hold to account, which presents substantial methodological challenges. Our project will explore the potential for new ways of seeing AI as an emerging technology, with a view to informing government funding and regulation. It will involve a collaboration between researchers in the UK and the Netherlands, linking four centres of metascience research - UCL, Warwick, CWTS Leiden and UNU MERIT in Maastricht. Our case study of AI for agriculture will allow us to explore the alignment of R&D and social needs with greater focus. We will study how AI is affecting agricultural R&D and reveal the overlaps or gaps with social needs among groups in the Global South that are often left out of rich countries’ priorities as they discuss the need to compete in a ‘race’ to develop AI technologies. Our research will use scientometrics, Delphi surveys and cutting-edge text analysis, complemented by qualitative methods. We will draw upon, connect with and inform major UKRI AI investments such as the Responsible AI programme and the Generative AI Hub (via Jack Stilgoe). We will also collaborate with Google Deepmind and draw on our expert advisers, including Juan Mateos-Garcia and Alondra Nelson. As we develop our case study, we will use our networks of collaborators in India and international bodies such as the Food and Agricultural Organisation and United Nations Development Programme to sharpen our research questions.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know