Cerca nel sito

03. Complexity and artificial intelligence for the challenges of Sustainable Development Goals

Complexity and artificial intelligence for the challenges of Sustainable Development Goals

This project will be carried out in collaboration with the Sony Computer Science Lab (Sony-CSL, https://csl.sony.fr/) in Paris, led by Prof. Loreto of the Physics Department of Sapienza University of Rome. The Sony CSL has been engaged for years in issues related to the goals of sustainable development, and this strategically important project is proposed for the creation of a new joint laboratory that can combine in a single space the sciences, the arts, and the business world for pushing innovation and aiming for concrete and applicable solutions on a large scale.

Our environment and our societies are clearly in danger and are undergoing major structural transformations, particularly through climate change, globalization, and digitalization. In this context, the ongoing crisis linked to COVID-19 has done nothing but make the proposals related to the Sustainable Development Goals (SGDs) even more urgent and pressing. Although the paths out of the current situation are not yet visible, it is clear that the crisis we are experiencing has the potential to profoundly change our habits and our lives. This complex system of challenges requires the development of new tools and methods to devise new solutions, plan optimal solutions, and effectively manage emergencies.

Areas of Application and Impact

The project will focus on the global innovation agenda, aiming to impact in this way various sectors linked to the Sustainable Development Goals (SDGs). In particular, due to the growing level of urbanization globally, many SDGs are destined to address issues related to urban spaces to: improve accessibility and mobility (SDG 11.2, 11.7); optimize logistics and waste management (SDG 12.4); enhance inclusion (SDG 10.7, 11.7); and promote a green transition (SDG 13.2). Another set of relevant challenges is related to disinformation. In addition to the damage to social dialogue, disinformation can have a strong impact on several SDGs. For example, disinformation is undermining awareness about vaccinations (SDG 3: Good health and well-being), is hindering decision-making on climate change (SDG 13: Climate action) and is threatening the democratic process and social cohesion (SDG 16: Strong institutions for peace and justice).

This project will set a research agenda aimed at:

  • developing data-driven, reliable modeling schemes for the problems underlying the SDGs;
    ● developing new AI tools for exploring the space of solutions to the problems underlying the SDGs;
    ● developing and implementing platforms that make it possible for all stakeholders to view the current state of the systems and to conceive and explore new scenarios, testing their effectiveness in real contexts.

The cited tools aim to accelerate the rate of innovation through assisted co-creation and co-design processes. The diagram shown here schematically illustrates the interaction between the various different activities that the project will carry out.

The structure in the figure revolves around the following three pillars, whose constant interaction represents the real novelty of the approach. Appropriate and accurate modeling schemes will be devised, exploiting statistical inference approaches, data-driven modeling, and machine learning, through a combination of human intuition and automatic inference. AI assistants will help devise relevant solutions in complex landscapes. Interactive platforms will constitute an interface for users, stakeholders, and decision makers to understand the present and to evaluate the effectiveness and relevance of specific solutions.

The project will develop data-driven modeling aimed at reconstructing the complex set of couplings between the layers and the characteristics that characterize a given problem. For example, these include, for a city, all the couplings between different areas and different layers (census, market forces, housing, public policies, services, etc.). Based on the detailed knowledge of the interaction patterns, specific indicators will be devised and tested to evaluate the present and predict the likely evolution. Being able to accurately predict the most likely evolution of complex systems is the key to creating validated scenarios to share with stakeholders and decision makers.

The project aims to implement a new generation of context-aware, artificially intelligent assistants to support real-time interactivity in modeling, problem-solving, and researching new strategies and solutions. AI and machine-learning tools will help humans find their way into the complex space of possible solutions to a given problem. AI assistants will be in constant dialogue with human users and serve multiple purposes: To enable human users to better understand the complexity of problems, to help infer suitable modeling schemes, and to seek relevant solutions (sweet spots) in high-dimensional space.

the project’s goal is to implement interactive platforms, adapted to the specific context of the relevant SDGs, to evaluate and visualize the present and develop what-if scenarios. The charaterisation of the present will be carried out through the visualization of appropriate significant observables. The platforms will enable users to take actions to change scenarios, encouraging out-of-the-box thinking. In this way, users will have the opportunity to explore and evaluate the validity of new solutions. The platforms will also make possible applications to a panoply of case studies in real contexts. An example, linked to mobility in urban spaces, is represented by the Citychrone platform (http://whatif.cslparis.com/citychrone.html).

Specific Objectives

The project will focus on the global innovation agenda, aiming to impact in this way various sectors linked to the Sustainable Development Goals (SDGs). In particular, due to the growing level of urbanization globally, many SDGs are destined to address issues related to urban spaces to: improve accessibility and mobility (SDG 11.2, 11.7); optimize logistics and waste management (SDG 12.4); enhance inclusion (SDG 10.7, 11.7); and promote a green transition (SDG 13.2). Another set of relevant challenges is related to disinformation. In addition to the damage to social dialogue, disinformation can have a strong impact on several SDGs. For example, disinformation is undermining awareness about vaccinations (SDG 3: Good health and well-being), is hindering decision-making on climate change (SDG 13: Climate action) and is threatening the democratic process and social cohesion (SDG 16: Strong institutions for peace and justice).

Urbanization is an irreversible trend in global demographic dynamics. The World Economic Forum (WEF) predicts that, by 2050, 68% of the world’s population will live in cities. While cities can be more energy efficient, this poses a number of challenges due to the high concentration of people and the resulting demand for resources, congestion, social divisions, and other issues. Cities today are undergoing significant changes that require informed and strategic thinking to achieve the SDGs. Urban phenomena (e.g., social exclusion and gentrification, mobility and accessibility, management of public events, recovery from natural disasters, redesign and planning of city boundaries and functional areas) act all on very different spatial and temporal scales (see figure).

The response to the challenges of urban sustainability can only come from a coordinated and multidisciplinary approach that operates on very different spatial and temporal scales—from the short time scale of the present to long-term strategic thinking and from the microscales of intervention on the ground (transport systems, logistics, etc.) to the large scale of the more complex characteristics (inclusion, gentrification, vocation of specific areas). The project will contribute to these challenges through the development of cutting-edge methods, merging the sciences of complexity, artificial intelligence, machine learning, and data science into a single approach. Furthermore, it will aim to create metrics and visualizations, modeling tools, and AI assistants whose impact will consist of their adoption by institutional agencies and policy makers around the world, to plan local interventions and remodel the cities of the future.

A specific example of the approach just described is represented by ecosystems related to mobility. Far from being only linked to travel between two physical places, this constitutes a tool to open up new opportunities—education, work, leisure—and to enrich our human experience and the potential of our communities. However, very often freedom of movement has never been accompanied by the right to mobility. In large urban areas (> 100,000 inhabitants), mobility solutions are far from optimal (urban centers are too dense  and disorganized, while the suburbs are isolated) and have a negative impact on our lives. The recent COVID19 epidemic has spawned enormous challenges to the reorganization of transport services, which are linked to the massive potential change in individual habits.

The need for physical distance between individuals is a variable that has never before entered the manuals of transport, architecture, urban planning, or work organization. Today, the meaning of this constraint is extending the meaning of the term “safety” of transport phenomena. The term security now also acquires the meaning of “the possibility of minimizing the risk of physical contact with potentially infected individuals”. Public health safety is a driver of change for mobility ecosystems. Rethinking mobility in the summer of 2020 also means conceiving a “post-covidic” era in which the theme of movement will intertwine with public health safety.

The interplay between innovative transport and potential new behaviors will revolutionize the way individuals move around the city and reshape the socioeconomic structure of the city itself. Regulating and supervising this transition is a significant challenge that requires a substantial and timely interdisciplinary effort.

Planning tools can now exploit a large amount of longitudinal GPS data, making possible high-resolution and real-time monitoring of individual habits, together with socioeconomic indicators and the states of the infrastructures. In parallel, the modeling schemes have reached maturity to support the conception of new scenarios. In this framework, the present project aims to play a central role in implementing this agenda, developing algorithmic and analytical tools capable of merging information from heterogeneous data sources to enable the orchestration of validated scenarios for the now unavoidable transition to new mobility systems for both developing and developed countries.

●WHO and WHY: Rethinking mobility needs and corresponding priorities. This implies understanding who has to move (WHO) and for what reasons (WHY). All these studies will be the basis for the conception of new mobility models that combine safety, inclusiveness, and sustainability.
● WHERE and WHEN: A crucial element in planning the transition to the mobility of the future will be a careful monitoring of the demand for mobility, i.e., understanding where people have to go and when and under what constraints, such as costs, duration, security, etc. In this context, the project aims to provide: (i) a reference framework in which data can be collected and organized in such a way as to be able to answer questions related to the fine-grained resolution levels affecting both the spatial and temporal domain; (ii) a set of tools to expand the set of observables for mobility to include, for example, safety distance factors.
● HOW: Once the mobility question has been assessed, it must be investigated whether and how this question can be satisfied or not. In this context, the extraordinary explosion of means or modes of transport must be taken into consideration. Shared mobility is now a reality that encompasses a variety of modes of transport, including car-sharing, bike-sharing, peer-to-peer ride-sharing, on-demand services, microtransit, and other modes, not to mention electric vehicles (EVs), semi-autonomous or fully autonomous vehicles (AV) and all the hybrid variants of these technologies.
● Conceive new solutions for safe, inclusive and sustainable mobility through a modular “what-if” platform (see for example the what-if machine platform developed by Sony CSL-Paris: http://whatif.cslparis.com/) to conceive new possible solutions for the transition of mobility and to validate them through a rigorous data-driven modeling of the complex phenomena underlying mobility. To this end, it will also be important to evaluate the models of individual and collective adoption of the new solutions, i.e. the integration of the new solutions into the fabric of the needs and habits of users.
● Conceive orchestrated scenarios for a transition to a new mobility. This objective comprises the synthesis of all the previous objectives and consists of the orchestration of new global mobility scenarios. The scenarios will be presented through the online interactive platform and discussed in depth with scholars, planners, stakeholders, and decision makers.

During the COVID-19 crisis, we have seen yet another confirmation of how crucial information and information technologies (Internet, social media, etc.) are in the life of a country. Particularly in countries with high levels of democracy, where the political class is strongly linked to the opinions of the population, the information enjoyed by the population is the source from which many mass behaviors arise, as well as the selection of the political class itself. However, if information technologies have radically changed the dynamics according to which these mechanisms historically occurred, they have also opened this process to new influences and new phenomenologies.

It is quite evident that in the last century the conflicts between states that previously took place mainly on the military level have been transferred mainly to the economic level. A recent and striking example of this is the trade war between the United States and China, which is actually part of a wider conflict for world economic hegemony. However, if the competition on the economic level is clear, another fundamental conflict level still in the shadows is that of information. Information technologies, which, in democratic countries, have reached very high levels of penetration, represent the new battlefields of world conflicts. However, the clashes are no longer between armies, but between narratives. There are many documented cases of attempts to influence democratic life during crucial and delicate moments, as in the case of Brexit or the US elections of 2016 and 2020.

However, it would be superficial to reduce the impact of social media on political discourse to attempts at external influences (which in fact have always existed since propaganda exists, but in other forms). The speed and interconnection capacity of new media, new technological tools for content creation, and new platforms have created a vast plethora of new phenomena, as well as changed the dynamics of historically known phenomena, even without the intervention of external influences. For example, the unprecedented interaction of confirmation bias with the gargantuan availability of content and resources made possible by the various platforms is one of the phenomena behind the rise of the so-called echo chamber. Another phenomenon of interest is the use of bots to “dop” the visibility of a profile or the dissemination of some news for commercial purposes, or the explosion of hate speech and trolling phenomena linked to the crisis of trust on the part of social media users.

All this vast and heterogeneous phenomenology has three main ingredients in common:

  • human cognitive and communicative abilities, with their peculiarities and their biases, which have always influenced social interactions, have now fallen into a new context with consequences that are only minimally understood;
  • the new information technologies, which not only offer an unprecedented speed and capacity of use and dissemination of resources, but which are also made up of algorithms and systems to manage the exploration of these resources that heavily affect the dynamics of exploration and fruition;
  • the emergence of collective phenomena from individual behaviors mediated by new technologies, which occurs in a rapidly evolving manner but which the new availability of data and experimental possibilities, makes it possible to study from an absolutely unprecedented quantitative point of view.

It is the co-occurrence of these elements that makes the collaboration between the CREF and the Sony CSL in Paris the ideal convergence to scientifically address the issues related to the new dynamics of information. The collaboration aims to address the problems set out on two different but parallel directives.

The first directive consists of a tactical approach for the frontal study of critical phenomenologies such as:

  • misinformation, and in particular the spread of fake news;
  • the creation and dissolution of the echo chambers and more generally the phenomena of polarization of opinions;
  • hate speech and trolling;
  • the use of bots on social networks;
  • the competition between conflicting narratives;
  • information imbalances, i.e., the overabundance of information on some topics as opposed to the lack of information in others.

In recent years, these research topics have been attracting growing interest from both the scientific community and the institutions, which have understood their crucial importance for democratic stability and the health of public discourse. The study of these issues will be addressed thanks to the scientific advancement of the techniques of modeling the Dynamics of Opinion, Network Theory, Machine Learning and, in general, the technical armamentarium of Data Science. The purpose of this directive is first of all to offer a deep scientific understanding of these phenomenologies that increases the transparency of the public debate. Secondly, this understanding will give rise to real-time monitoring and disclosure tools to verify and make visible, in a transversal and transparent manner, the state of health of the public debate both to the population and to policy-makers.

From a more strategic point of view, the collaboration aims to improve public discourse through the study of the conditions in which this occurs and through the proposal of new tools to avoid vicious circles and enhance virtuous behavior. The study subjects will be, for example:

  • the algorithms for selecting and filtering resources on social media (the so-called recommendation systems) and their impact on the dynamics of exploration and formation of opinions. Such systems have often been linked to the creation of echo chambers due to their drive towards the preferences expressed in the past by users. Therefore, the strategic objective will be to verify this effect and develop new systems that can help the user in exploring new content without making the experience less pleasant, thus playing on the border of the so-called comfort zone.
    • reputation systems to improve the dynamics of trust in information sources. Paradoxically, we have information aggregated and organized into reviews and ratings on almost any type of content available on the net (movies, music, etc.) but not on the information sources themselves. The strategic objective consists of the study of these reputation systems and the experimental introduction of these systems, appropriately adapted, in the context of information sources. In fact, the aggregation systems of the evaluations will necessarily have to be adapted to compensate for the effects due to the echo chambers, avoiding that each “supporter” validates their own source of trust, by attributing a privileged weight to the transversality of the evaluations. These systems will be able to play a strategic role in building healthier and more transparent information dynamics.

These two directives, parallel but communicating, consist of pure research activities, as well as experimentations and case studies. This is also why the axis between Centro Fermi and Sony CSL in Paris is fundamental, as a collector of skills, experiences, and know-how that cover both the scientific and technological aspects. In addition, the Sony CSL in Paris boasts a historic collaboration with AGCOM (Italian Communications Authority), with several active projects on topics close to those explained. AGCOM is a partner of exceptional value because it has standing with both with the community of information professionals, with the stake-holders, and with the policy-makers, whose strategic value, both for theoretical and more experimental initiatives, has a very high impact.