In parallel with the definition of the SDGs, a global movement to which I belong has developed that highlights the potential of data and artificial intelligence (AI) to measure and accelerate the achievement of the SDGs. In this brief article I provide some insights into this matter, illustrating the rich intersection between AI and the SDGs through a selection of the latter:
AI techniques have been used to automatically analyze satellite, mobile or digital transactions to infer poverty or socioeconomic levels, especially in developing countries where census information is limited. Additionally, AI-based decision support systems could inform public decisions regarding poverty eradication programs both to measure the success of such programs and to guide resource allocation based on current and estimated levels. poverty rates in different regions.
Meteorological, satellite, demographic and socio-economic data have been analyzed using AI techniques for early detection of hunger in developing countries. Food demand in areas affected by natural disasters and crop yields can be predicted from climate and agricultural data, sometimes combined with satellite data. Invasive species and pests can be automatically recognized in images with AI techniques, as well as identification and recommendation of crops based on soil characteristics.
In this regard, there are two emerging and relevant concepts: precision agriculture and smart agriculture, which focus on harnessing data captured by a variety of sensors and cutting-edge technology to optimize crop performance while preserving resources. According to FAO, smart agriculture refers to the use of digital technology – including Internet of Things sensors, autonomous drones, livestock feeding robots and AI techniques to analyze data captured by sensors – to improve production systems. agricultural.
The intersection between data, AI and health is rich and full of opportunity. Overall, AI methods redefine medicine from at least three perspectives:
1. Accelerating the discovery and design of effective treatments and vaccines, which allow the prediction of expected results and side effects, in addition to automating the discovery of new pharmacological compounds and protein folding.
2. Assisting in clinical decision-making related to, for example, the diagnosis of cancer, covid or tuberculosis in radiological tests, potentially providing expert feedback and diagnoses to patients where human experts may not be available; improve pregnancy, postpartum and child care and therefore prevent deaths; and predict the effectiveness of treatments or the likelihood of needing intensive care.
3. Supporting policymaking related to public health, including mental health and infectious diseases, such as malaria, influenza, Ebola and Covid, through the analysis of multidimensional data captured by mobile network infrastructure, social media platforms and pervasive sensors.
Furthermore, the increased availability of wearable devices at affordable prices (e.g., activity bracelets, smart watches…) allows large-scale data collection on daily activities, sleep habits and physiological signals, which , analyzed through AI techniques, could be extremely valuable in the early diagnosis of diseases and in personalized, preventive and predictive medicine.
Indeed, such medicine will not be achieved without the use of AI techniques applied to genomic, behavioral, contextual (e.g., pollution, climate), and medical data.
AI has the potential to contribute to education in several ways. First, allowing a personalization of the learning experience, moving from a general educational model, from one to many, to an individual model, from one to one. Intelligent tutoring systems (ITS) through software agents, chatbots or social robots can personalize both the content and the strategies used to teach students, to maximize their learning. In addition, intelligent educational interfaces allow early detection of students with physical or cognitive functional diversity and provide the necessary tools to help them learn more effectively. Secondly, AI methods are used to enable more efficient academic management (e.g. automatically creating schedules for teachers, supporting teachers in grading, providing 24/7 support via chatbots, etc.) and to evaluate the quality of education.
However, the potential risks of using AI in education would need to be studied further. They include the violation of privacy, the subliminal manipulation of student behaviors through personalized algorithms, different types of discrimination, the lack of veracity of text generation systems such as ChatGPT, the impact of generative AI on learning and assessment methods, and the possible negative effects on the physical and mental health of students along with their behavioral development.
AI has a critical role to play in this SDG. Smart energy grids rely on AI to predict demand and optimize their maintenance and operation, and to increase their efficiency through automatic failure and cyber-attack detection and load prediction. Semi-autonomous or autonomous robots are used to inspect and maintain renewable energy plants, so they could be placed in remote or dangerous locations, but with optimal prospects for energy production.
The application of AI in nuclear engineering has been limited to date. However, AI algorithms can be used to predict the behavior of nuclear reactors, perform predictive maintenance of nuclear infrastructure or improve fire risk models.
Finally, there are numerous examples of how data-driven AI methods are key enablers for creating efficient renewable energy systems (wind, solar, geothermal, hydropower, ocean, bioenergy and hybrid) by providing accurate predictions of expected energy source behavior. renewable energy and, therefore, allow the optimization of energy generation systems.
Smart cities depend on AI. There are numerous initiatives around the world to make these a reality, including projects that analyze data captured by Internet of Things devices to measure and optimize energy consumption, recycling levels, pollution and garbage collection in cities. .
Urban security is a critical area that contributes to the quality of life in cities and has not escaped the impact of AI to automatically detect and predict crime hotspots in cities.
AI techniques can help improve urban planning by estimating urban density from aerial imagery, informing decisions related to road and public transportation planning, detecting traffic incidents, and predicting future traffic conditions or transportation needs. mobility.
Intelligent urban transport systems are only possible thanks to AI methods, leading to safer, more inclusive and efficient public transport. In addition, modern commercial vehicles leverage AI to increase intersection safety, detect incoming traffic and pedestrians, avoid collisions – e.g. by detecting inattentive drivers – predicting driver maneuvers, predicting pedestrian behaviors or warning drivers when they cross into other lanes, and assisting drivers in adverse weather conditions.
AI enables the development of intelligent production systems that minimize energy consumption, anticipate demand, detect manufacturing failures, automate tasks, and perform systematic evaluations to detect areas for improvement. In addition, AI methods can be used to better predict and regulate transport flows, to help plan more efficient public transport routes and to deploy autonomous vehicles for passengers and cargo on land, rail or even air transport. Digital twins can also be used to optimize production systems.
For its part, consumption patterns can be predicted with AI algorithms, allowing production systems to be more efficient with minimal excess production. For example, AI techniques can automatically create land use maps to provide a more accurate picture of the state and actual use of natural resources or, for example, to estimate the impact of logging on forests, optimize logging processes and guarantee their sustainability.
Likewise, AI methods can be leveraged to predict solid waste in municipalities and therefore allow for more efficient planning. Finally, socially responsible consumption and waste disposal behavior can be automatically inferred through AI algorithms, and this information could be used to encourage and reinforce consumer behaviors that contribute to sustainability.
The potential for AI to help address the climate emergency is unquestionable. In fact, we will not be able to combat climate change without the help of AI. Its methods are used to model climate and weather, identify patterns and make accurate predictions based on the analysis of multidimensional weather and climate data sets. In addition to being used to create more accurate climate models and predictions, AI can also be applied to improve next-generation weather modeling systems by enabling, for example, automatic labeling of climate data.
AI has also proven to be a valuable ally in predicting extreme weather events and their impact, such as heavy rain, hail, wildfires, floods and earthquakes and in enabling a more efficient and faster response to natural disasters. Autonomous drones – guided by AI – can be used to control heat and prevent fires and to search for survivors in floods and earthquakes.
Beyond the direct application of AI techniques to model and predict climate, its methods can be applied to industries or sectors that have a negative environmental impact to enable the reduction of greenhouse gas emissions.
On the contrary, data-driven AI systems have a significant contribution to the CO2 footprint. which should be systematically measured and mitigated and which constitutes one of the challenges of current AI systems.
Despite this immense opportunity, today’s reality is far from this vision. To date, there have been few successful examples of real-world systems that systematically leverage large-scale data and AI methods to help humans make better decisions for social good. In this context, it is necessary to consider five types of barriers that limit or prevent the use of AI for sustainable development in a safe and ethical way: 1. political and regulatory; 2. technological-scientific; 3. governance and ethics; 4. economic, and 5. environmental/climatic.
AI offers us a unique opportunity for social good, to ensure not only the sustainability of our societies and our planet, but our own survival. This potential will only become a reality with international cooperation and ambitious investments that also address the challenges and barriers posed by AI. As Theodore Roosevelt said, “Sometimes a revolution is necessary.”
Nuria Oliver is scientific director and co-founder of the ELLIS Alicante Foundation, co-founder and vice president of ELLIS, chief scientific advisor of the Vodafone Institute and chief data scientist at DataPop Alliance.
