Can we leverage Artificial Intelligence (AI) methods for Social Good? Can we take advantage of this discipline to help us limit the impact of natural disasters, to combat the spread of pandemics, to improve crime control, to increase agricultural sustainability and productivity, to help with medical diagnoses or to promote social inclusion?
Yes, we certainly can and we should, as our very survival may depend on it.
Artificial Intelligence (AI) is the discipline within engineering or computer science devoted to building intelligent non-biological computational systems, taking human intelligence as a reference. Although it has existed as a discipline since the 1950s, we have witnessed in the last 15 years a great prominence of AI in our society, thanks to the confluence of 3 factors: the existence of huge amounts of data (Big Data), the availability of low-cost high computing capabilities and the development of complex machine learning models –inspired by the neural networks developed in the 1950s– called deep learning models that, trained on lots of data and taking advantage of large scale computing capabilities, are at the heart of the digital services that we use in our daily lives, and of smart cars, cities, homes and mobile phones.
The intersection between data, Artificial Intelligence and Social Good is rich and full of opportunities. From the point of view of the impact channels and the use of Artificial Intelligence to contribute to a better design and implementation of policies and social actions, such an intersection may be structured into four large areas:
- AI for better measurement and monitoring, with the aim of bridging data gaps and improving situational awareness on specific development indicators, allowing a more detailed and accurate picture to be drawn of the underlying social reality. This area is indirect because the impact of AI will depend on the extent to which actions are taken based on the conclusions derived from the analysis of such data using AI methods. Examples of the use of AI in this sense include the analysis of satellite, meteorological, demographic, mobile phone network, social network or banking transaction data to infer the poverty of a region; to automatically detect or even predict famine areas; to model the spread of infectious diseases and pandemics; to predict the academic results of students; to identify and predict crime hot spots in cities; or to detect fraud automatically.
- AI for greater precision and intelligence, through AI-based products and services that are explicitly designed to have an impact in one or more areas for social development. This includes AI applications for decision support using algorithms and considerations on whether or not decisions around fundamental rights and human development should be automated. Examples in this category include the use of AI algorithms in public health to automatically analyze radiological images to diagnose diseases, such as cancer or COVID-19; to predict the likelihood of inmate redicivism based on their historical behavioral data; to decide the granting of credits based on historical patterns of use of technology and/or financial services; or to determine the granting of aid or scholarships based on demographic, academic and socio-economic data.
AI to design, execute and evaluate policies, developing approaches that harness AI to design and implement improved policies and programs over existing ones. There are not many examples of this AI impact channel yet. It is worth highlighting our project on the use of AI techniques during the COVID-19 pandemic in the Valencian Community to help design optimal public policies, that is, with the best balance between the economic and social cost of said policies and the resulting number of cases. of COVID-19. Other examples include the use of AI algorithms to better identify potential welfare recipients by predicting both false positives and false negatives.
- AI in other applications and services, which, although not directly linked to Social Good, have an enormous impact – positive or negative – on people, from content or product recommendation services to social networks or smart speakers . The ubiquity of AI in our lives inevitably links it to the progress –or not- of our society. In this channel of action, developing mechanisms that guarantee compliance with ethical principles in accordance with European values is of utmost importance. Such principles include preserving privacy, non-discimination, guarantees of veracity, respecting human dignity and autonomy, transparency, diversity, reliability and a clear accountability, among others.
Thanks to Artificial Intelligence we can better understand the world, predict trends, identify vulnerablel regions or social groups –together with the variables that most contribute to such vulnerability. We can evaluate the impact of social projects and define new actions and social impact policies that are effective and evidence-driven. The opportunities that AI research offers us to have positive social impact are almost limiteless. It is precisely this social aspect of AI what motivates and has always motivated our work, and it is the focus of our Foundation.