Food and Agriculture Organization (FAO)
Food and Agriculture Organization (FAO)
Description of Activities on AI
Project 1: To detect fall army worm damage using a mobile application
The FAMEWS global platform is an online resource for mapping data collected by the FAMEWS mobile app whenever fields are scouted or pheromone traps are checked for FAW. The platform provides a real-time situation overview with maps and analytics of FAW infestations at global, country and sub-country levels. The data and maps provide valuable insights on how FAW populations change over time with ecology in order to better understand its behaviour and guide best management practices.
Project 2: Port inspectors, custom agents, fish traders and other users without formal taxonomic training, iSharkFin allows the identification of shark species from a picture of the fin
iSharkFin is an expert system that uses machine learning techniques to identify shark species from shark fin shapes. The software was developed by FAO in collaboration with the University of Vigo with financial support from the Government of Japan and the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). Aimed at port inspectors, custom agents, fish traders and other users without formal taxonomic training, iSharkFin allows the identification of shark species from a picture of the fin.
Project 3: Land cover / crop classification using satellite imaginary, phenology and ground reference data
WAPOR is supervised classification methodology is applied to assign a specific class to each pixel of the image. Training data consist of seasonal and long term metrics derived from dekadal NDVI time series, phenology and spectral reflectance data combined with reference data denoting the exact location of each of the classes.
Project 4: Palm tree mapping from satellite imagery
(Internal, In use)
Project 5: Fleet estimation
(Internal, In use, Used for improving fisheries statistics)
Project 7: Detecting Fall Armyworm infestations
An innovative, talking app – Nuru – to help African farmers recognize Fall Armyworm, a new and fast-spreading crop pest in sub-Saharan Africa, so that they can take immediate steps to destroy it and curb its spread.
Project 8: FAO Data Lab
FAO’s Big Data tool on food chains under the COVID-19 pandemic. This open-access tool developed by the FAO Data Lab gathers, organizes and analyses daily information on the impact of the COVID-19 pandemic on food and agriculture, value chains, food prices, food security and undertaken measures.
Project 9: Hand-in-Hand Initiative Geospatial Platform
Using machine-learning for image classification and knowledge discovery. Using natural language processing for queries.
Challenges and Opportunities
There are claims that AI capabilities will someday exceed human capabilities, and in many areas, they already come close to this benchmark. With this in mind, we would like to comment on the seven AI principles. In general, we believe these principles are an excellent start. They are about a sine qua non condition, but are insufficient to cover all facets of this embryonic field on the fourth industrial revolution.
Therefore, FAO would like to highlight that Artificial Intelligence is an entire domain of knowledge and should not be seen only as a tool or a menace. We believe that more intensive learning and training is needed in this area to understand the technology and its implications. Today we see AI portrayed in a sensationalistic manner, and can be easily distracted by the rapid advancements and fantastic scenarios envisioned for future use, while we search for appropriate use cases in the core functions of our business. We need to understand that AI is a series of algorithms based on data (evidence or observations), which will continue to get smarter and more pervasive, eventually surpassing human capacities in many activities (faster and more precise) though never quite the same as human beings. That said, even today there are many areas of work aiming to build self-conscious machines. Therefore, to focus the core of this approach in one type of technology (AI) could be an error, and we propose to expand the scope of this approach, to understand the implications and potential benefits of technology more broadly, and how this could be oriented in terms of policymakers and principles.
The UN needs to exploit the topic widely in order to build a holistic approach local and globally. The most important role of AI is outside of the seven proposed principles and should be included as fundamental to our approach to AI. This role is the ability to use AI to predict unexpected events, threats and crises. Challenges such as hunger, climate change, and migration could be addressed before they become crises through early detection, prevention and mitigation of natural disasters, social conflicts or economic hazards.
Despite this caveat, we concur with the seven guiding principles, and will interpret them through the lens of the impact of AI on food security and ensuring that any programmes implemented by FAO do not increase the digital divide and risk creating or increasing food insecurity, especially for those at risk of being left behind.
There is no doubt that AI, and other technologies, and its applications will replace jobs4, 5, and this is a widely accepted consequence of all technology that has resulted from the industrial revolution. However, this does not need to be seen as an entirely negative consequence, assuming that we can successfully promote other types of jobs. At FAO, we believe that AI policies and programmes of member states need to be oriented to contribute to job and entrepreneurship opportunities creation for Youth in developing countries. This development should induce young people to remain in the rural areas with employment perspective and suitable livelihoods conditions.
At FAO, we would like to see a better understanding in terms of the technologies (AI and others), as an incomplete understanding can lead to biased assumptions with regards to comprehension and analysis of strategies for consideration and implementation. The text indicated that AI is complex, and therefore, we are unnecessarily limiting our analysis in terms of understanding and capability to manage this technology. In general, AI is a set of algorithms and methodology to process data and use them to improve the precision and response time to make or support decision (classification, forecast, etc.). We propose that a better understanding will lead to the ability to provide a more fair and sound assessment. Therefore, we suggest that important training is provided to those who will create the UN’s strategy for AI in terms of what the technology is, what it can do, and the implication for our business.
In terms of the four points mentioned: (a) infrastructure; (b) data; (c) human; and (d) policy/law/human rights, we concur with all of them. However, we consider that the infrastructure area is a topic by itself, because it is important to bring other capabilities, technologies and solutions to eliminate the digital divide and promote innovation, jobs and fairness. We recognize that this is a good-to-have for AI, but it is not a necessary condition, because AI solutions could also be used offline. Therefore, we propose to have a different, separate note dedicated to the need to increase connectivity and reduce the digital dive.