The “Creativity, Innovation and Artificial Intelligence” topic, “Creativity” in short, is the newest research line of CSL Paris. It started its activities at the end of 2017 and it focuses on the investigation of the processes underlying innovation and human creativity and their interplay with the most recent advances in Artificial Intelligence, Machine Learning and Inference methods. The research, by blending in a unitary interdisciplinary effort three main activities – theoretical modelling, data-science and machine learning, gaming and participation – aims at developing a science of the “new”, focusing on how the “new” emerges in social and technological systems and how humans and machines explore the space of possibilities and find new solutions.
The main Topic includes several specific research lines that can be summarized as follows:
(i) The mathematics of the new: One of the key problems when studying innovation processes is represented by a lack of a suitable mathematical framework to describe the occurrence of events whose existence one did not even previously suspect; this is the so-called problem of ‘unanticipated knowledge’. In this framework, a beautiful notion is that of the “adjacent possible”. Originally introduced in the framework of biology, the adjacent possible metaphor already expanded its scope to include all those things (ideas, linguistic structures, concepts, molecules, genomes, technological artifacts, etc.) that are one step away from what actually exists, and hence can arise from incremental modifications and recombination of existing material. Mathematically, the notion of Adjacent Possible has been formulated by some of the core members of the Creativity topic (see tria_2014). Based on this general formulation extremely challenging problems can be faced, to investigate the topology of the space of possibilities and its dynamical evolution at the individual and collective level. The final goal is that of defining a coherent and self-consistent mathematical formulation that, beyond explaining stylized facts (statistical laws, correlation and triggering effects, etc.), is able to cast concrete predictions to be grounded on actual data.
(ii) Unfolding creativity processes: This research line is focusing on unveiling the strategies of exploration of the adjacent possible in many different systems (social, biological, technological). The goal is pursued through a data-science and machine-learning approach to datasets mirroring the emergence of novelties in very different kind of systems. This approach is paralleled by the realization of actual experiments involving people, both through online gaming and open events (see for instance: www.kreyon.net/kreyonDays) to engage people in activities that challenge them to explore their adjacent possible and come up with new ideas, recombining existing ideas, effectively triggering some evolutionary dynamics of novelties (videos of several activities are available here: https://goo.gl/Ejdy14). A special attention is devoted to the way in which machines and artificial agents are able to explore their adjacent possible and overcome the problem of the unanticipated knowledge.
(iii) Platforms for a sustainable world: The intrinsic complexity of the emerging challenges human beings collectively face requires a deep comprehension of the underlying phenomena in order to plan effective strategies and sustainable solutions: from the planning of urban infrastructures to containment strategies for pandemics, from the impact of political campaigns to measures against information pollution and misinformation. In all these cases, decision-making processes have to be supported with meaningful representations of the present situations along with accurate simulation engines to generate and evaluate future scenarios. Instrumental to all this is the possibility to gather and analyze huge amounts of relevant data and visualize them in a meaningful way also for an audience without technical or scientific expertise. Understanding the present through data is often not enough and the impact of specific decisions and solutions can be correctly assessed only when projected into the future. Hence the need of tools allowing for a realistic forecast of how a change in the current conditions will affect and modify the future scenario. In short scenario simulators and decision support tools. In this framework CSL Paris is launching a new research direction aimed at developing effective infrastructures merging the science of data with the development of highly predictive models, to come up with engaging and meaningful visualizations and friendly scenario simulation engines.
Interactions with other Topics
Creativity-Language: The topic of Creativity is strongly intertwined with the topic of Language. Language is in fact one of the most natural playground to investigate creativity and innovation processes for several reasons and CSL Paris has a strong interest in Language studies and its Language team is widely known for its seminal contributions to the developments of Construction Grammars. In addition, Language features a vast ecosystem of innovation phenomena and creative exploits. Not to mention the huge amount of language-related data already available, that are a strategic starting point for any scientific investigation, and the vast corpus of both theoretical and computer science tools for natural language processing.
Joint projects: Anticipation processes.
Misinformation threatens our societies, but little is known about how the production of news by unreliable sources relates to supply and demand dynamics. We exploit the burst of news production triggered by the COVID-19 outbreak through an Italian database partially annotated for questionable sources. We compare news supply with news demand, as captured by Google Trends data. We identify the Granger causal relationships between supply and demand for the most searched keywords, quantifying the inertial behaviour of the news supply. Focusing on COVID-19 news, we find that questionable sources are more sensitive than general news production to people’s interests, especially when news supply and demand mismatched. We introduce an index assessing the level of questionable news production solely based on the available volumes of news and searches. We contend that these results can be a powerful asset in informing campaigns against disinformation and providing news outlets and institutions with potentially relevant strategies.
The adverse effects of unsustainable behaviors on human society are leading to an increasingly urgent and critical need to change policies and practices worldwide. This requires that citizens become informed and engaged in participatory governance and measures leading to sustainable futures. Citizens’ understanding of the inherent complexity of sustainable systems is a necessary (though generally not sufficient) ingredient for them to understand controversial public policies and maintain the core principles of democratic societies. In this work, we present a novel, open-ended experiment where individuals had the opportunity to solve model urban sustainability problems in a purposeful game. Participants were challenged to interact with familiar LEGO blocks representing elements in a complex generative urban economic indicators model. Players seeks to find a specific urban configuration satisfying particular sustainability requirements. We show that, despite the intrinsic complexity and non-linearity of the problems, participants’ ability to make counter-intuitive actions helps them find suitable solutions. Moreover, we show that through successive iterations of the experiment, participants can overcome the difficulties linked to non-linearity and increase the probability of finding the correct solution to the problem. We contend that this kind of what-if platforms could have a crucial role in future approaches to sustainable developments goals.
COVID-19, Electricity consumption, Electricity generation, Governmental restrictions, Renewable share, Electricity grid”,
abstract = “When COVID-19 pandemic spread in Europe, governments imposed unprecedented confinement measures with mostly unknown repercussions on contemporary societies. In some cases, a considerable drop in energy consumption was observed, anticipating a scenario of sizable low-cost energy generation, from renewable sources, expected only for years later. In this paper, the impact of governmental restrictions on electrical load, generation and transmission was investigated in 16 European countries. Using the indices provided by the Oxford COVID-19 Government Response Tracker, precise restriction types were found to correlate with the load drop. Then the European grid was analysed to assess how the load drop was balanced by the change in generation and transmission patterns. The same restriction period from 2020 was compared to previous years, accounting for yearly variability with ad hoc statistical technique. As a result, generation was found to be heavily impacted in most countries with significant load drop. Overall, generation from nuclear, and fossil coal and gas sources was reduced, in favour of renewables and, in some countries, fossil gas. Moreover, intermittent renewables generation increased in most countries without indicating an exceptional amount of curtailments. Finally, the European grid helped balance those changes with an increase in both energy exports and imports, with some net exporting countries becoming net importers, notably Germany, and vice versa. Together, these findings show the far reaching implications of the COVID-19 crisis, and contribute to the understanding and planning of higher renewables share scenarios, which will become more prevalent in the battle against climate change.
The perception of facial attractiveness is a complex phenomenon which depends on how the observer perceives not only individual facial features, but also their mutual influence and interplay. In the machine learning community, this problem is typically tackled as a problem of regression of the subject-averaged rating assigned to natural faces. However, it has been conjectured that this approach does not capture the complexity of the phenomenon. It has recently been shown that different human subjects can navigate the face-space and “sculpt” their preferred modification of a reference facial portrait. Here we present an unsupervised inference study of the set of sculpted facial vectors in such experiments. We first infer minimal, interpretable and accurate probabilistic models (through Maximum Entropy and artificial neural networks) of the preferred facial variations, that encode the inter-subject variance. The application of such generative models to the supervised classification of the gender of the subject that sculpted the face reveals that it may be predicted with astonishingly high accuracy. We observe that the classification accuracy improves by increasing the order of the non-linear effective interaction. This suggests that the cognitive mechanisms related to facial discrimination in the brain do not involve the positions of single facial landmarks only, but mainly the mutual influence of couples, and even triplets and quadruplets of landmarks. Furthermore, the high prediction accuracy of the subjects’ gender suggests that much relevant information regarding the subjects may influence (and be elicited from) their facial preference criteria, in agreement with the multiple motive theory of attractiveness proposed in previous works.
The rapid urbanization makes the understanding of the evolution of urban environments of utmost importance to steer societies towards better futures. Many studies have focused on the emerging properties of cities, leading to the discovery of scaling laws mirroring the dependence of socio-economic indicators on city sizes. However, few efforts have been devoted to the modelling of the dynamical evolution of cities, as reflected through the mutual influence of socio-economic variables. Here, we fill this gap by presenting a maximum entropy generative model for cities written in terms of a few macro-economic variables, whose parameters (the effective Hamiltonian, in a statistical-physical analogy) are inferred from real data through a maximum-likelihood approach. This approach allows for establishing a few results. First, nonlinear dependencies among indicators are needed for an accurate statistical description of the complexity of empirical correlations. Second, the inferred coupling parameters turn out to be quite robust along different years. Third, the quasi time-invariance of the effective Hamiltonian allows guessing the future state of a city based on a previous state. Through the adoption of a longitudinal dataset of macro-economic variables for French towns, we assess a significant forecasting accuracy.
In the last decades, the acceleration of urban growth has led to an unprecedented level of urban interactions and interdependence. This situation calls for a significant effort among the scientific community to come up with engaging and meaningful visualizations and accessible scenario simulation engines. The present paper gives a contribution in this direction by providing general methods to evaluate accessibility in cities based on public transportation data. Through the notion of isochrones, the accessibility quantities proposed measure the performance of transport systems at connecting places and people in urban systems. Then we introduce scores ranking cities according to their overall accessibility. We highlight significant inequalities in the distribution of these measures across the population, which are found to be strikingly similar across various urban environments. Our results are released through the interactive platform: www.citychrone.org, aimed at providing the community at large with a useful tool for awareness and decision-making.
The Letter ‘Interacting Discovery Processes on Complex Networks’ is out...
The second release of the “Online misinformation monitoring system in...