UNESCO flying high at United Nations’ Science, Technology and Innovation Forum

The Seventh Annual Multi-Stakeholder Forum on Science, Technology and Innovation for the Sustainable Development Goals (STI Forum) is taking place from 5 to 6 May in New York in a hybrid format. In addition to the main programme, there have been a number of pre-events and side events throughout the forum that have tackled issues ranging from disaster risk reduction to energy-saving technologies such as Tiny Machine Learning and green hydrogen.
using solar energy in rural Africa

Strategies designed to reduce the impact of disasters

From 2000 to 2019, disasters claimed 1.23 million lives around the world, affected 4 billion people and cost US$2.97 trillion in economic losses. This year is not shaping up to be any better’, observed Shamila Nair-Bedouelle, Assistant Director-General for Natural Sciences at UNESCO on 6 May.

She was speaking at a side event organized by the World Federation of Engineering Organizations (WFEO) with the support of UNESCO, the International Science Council and American Society of Civil Engineering, to launch WFEO’s publication on Engineering Resilience in Disaster Risk Management for Sustainable Development. The book provides guidance on planning and policies to reduce the impact of natural disasters in high-risk areas, the design and development of resilient infrastructure systems, data systems on natural disasters and advice on best practices for institutional frameworks.

The growing human and financial cost of these tragedies is pushing risk reduction up the agenda across the United Nations system. Although the Sendai Framework for Disaster Risk Reduction to 2030 is our roadmap in this domain, other global agendas, including the Sustainable Development Goals, the Paris Agreement on climate action and the New Urban Agenda all have targets that will only be attained if we manage to limit our exposure to disaster risk.

UNESCO recalled that its own report on Engineering for Sustainable Development (2021) analysed many of the issues raised by the session. In turn, the International Science Council presented the achievements of its Integrated Research on Disaster Risk programme.

Open science can build resilience in times of crisis – and afterwards

The need to promote sustainability science and reduce disaster risk was also on Dr Nair-Bedouelle’s mind as she took the floor in the session on Global Research Cooperation, Funding and Partnerships. ‘The greatest science-related risks that threaten our world now and in the years to come’, she said, ‘are failure of climate change mitigation and adaptation; extreme weather events; natural hazards, human-made environmental hazards; biodiversity loss and ecosystem collapse; and water crises’.

Dr Nair-Bedouelle observed that global research cooperation guaranteed access to scientific knowledge and its benefits that could, in turn, build resilience not only in times of crisis, but, above all, in post-crisis environments.

To promote global research cooperation, UNESCO adopted two normative instruments last year which are both world firsts: the Recommendation on Open Science and the Recommendation on the Ethics of Artificial Intelligence.

Now, UNESCO is urging governments to apply the principles contained in these recommendations to ensure maximum impact. ‘We believe that the implementation of the Recommendation on Open Science will contribute to reducing the digital, technological and knowledge divides between and within countries’, affirmed Dr Nair-Bedouelle.

As she told participants in the side event on Open Science Promoted Quality Graduate Education in the Global South organized on 4 May by the China Association for Science and Technology’s Consultative Committee on Open Science and Global Partnership, ‘quality education and open science are inseparable and their relationship is reciprocal. Open science aims to provide access to science and scientific advancements to everyone, everywhere and quality education enables people to develop and contribute to open science’.

Tiny Machine Learning, a low-cost, low-power solution

The Recommendation on the Ethics of Artificial Intelligence will support countries’ efforts to build policies and regulatory frameworks that ensure that these emerging technologies benefit humanity as a whole.

Tiny Machine Learning (TinyML) – a subfield of machine learning – was one of the emerging technologies highlighted by a session on 6 May organized by the Interagency Team on Emerging Technologies, which groups sister agencies that include the United Nations Department of Economic and Social Affairs, the International Labour Organization, UNESCO and the United Nations Industrial Development Organization.

Dr Marco Zennaro from UNESCO’s Abdus Salam International Centre for Theoretical Physics presented a policy brief to the session on TinyML that had been prepared in tandem with Brian Plancher and Vijay Janapa Reddi from Harvard University in the USA.

The TinyML process starts by collecting data from devices equipped with the Internet of Things. It then trains the collected dataset in the cloud to extract knowledge patterns. These are subsequently packaged into a TinyML model that takes into account the target microprocessor’s limited resources, such as in terms of memory and processing power. The resulting model is then deployed on embedded devices where it is used to evaluate new sensor data in real-time without communicating with the cloud.

By focusing on developing models that can be executed on small, real-time, low-power and low-cost embedded devices, TinyML could save a tremendous amount of energy at little cost. It would also require low connectivity and ensure privacy – as no raw information would be sent over the internet.

Given these advantages, ‘we believe that TinyML can enable research in artificial intelligence in the developing world and, thereby, foster local solutions with a small carbon footprint’, explained Dr Zennaro.

He went on to cite TinyML applications that address the Sustainable Development Goals. These included a prototype water-borne cholera detector kit that had been developed in Rwanda, camera traps to help track wildlife, a model for detecting optimum tea fermentation and the Ribbit Network, which provides accurate readings of carbon dioxide gas through a crowdsourced network of open-source, low-cost, smart sensors. ‘This kind of high-quality data will help scientists to understand better and predict the impact of climate change’, observed Dr Zennaro.

TinyML: Applied AI for Development

Growth in sustainability science strongest in lower middle-income countries

A second policy brief submitted to the same session by UNESCO summarized a study in the UNESCO Science Report (2021) on global trends in scientific publishing on topics of keen relevance to sustainable development. The authors found that scientific output on topics such as climate-ready crops and local disaster risk reduction strategies remained marginal in the publication record.

Growth in output in sustainability science was fastest between 2011 and 2019 in lower middle-income countries, particularly as concerns problem-solving research. For instance, lower-income countries contributed to one-fifth of global output on ecological approaches to industrial waste management and to research on photovoltaics and biofuels and biomass in 2019, despite accounting for just over 4% of global research expenditure… ‘This suggests a real potential for the development of [sustainable] industries’ note the authors ‘but, currently, only one in five publications in supportive fields such as engineering and cross-cutting strategic technologies involves international collaboration’.


Avenues for using Science for Smarter Development
Brief for IATT report to STI forum, 2022