Page d'accueil // IAS // Pilot projects

Pilot projects

Audacity projects – 2019

LAIWYERS: Law and AI: WaYs to Explore Robust Solutions by M. Cole and Y. Le Traon

Artificial Intelligence (AI) is reaching out into all sectors and the law is no exception. Machine learning methods are being discussed as potential substitutes for human interaction or decision or at least for creating results on which human decisions are based. In the area of law and its application, this can start for example with police investigations, range from the use of analysis in the prosecution to the final judicial decision-making, or from a client-lawyer relationship to the decision about damages in a civil procedure. Although there is still a long way to go before human intervention in the field of law could be replaced, first use cases show already now that there is an inherent danger in using AI solutions to replace analysis of a specific situation and possible predictions derived from it when attaching legal consequences to such decisions: the data used may be biased or the way they are applied may have unintended consequences and therefore the results can be unfair, or – in terms of the law – unjust. The area of law in this context is only a possible example to illustrate the research project’s goal: firstly, testing machine learning (ML) applications, which are used to replace or as a basis of human decision-making, in order to identify whether it is technically possible to get adversarily unfair results, and, secondly, to then be able to conclude whether what is technically possible is overall desirable or not. If testing of existing AI methods and models used e.g. in the context of law as described above, show that decision outcomes are worse in light of the originally perceived goal (just/unjust) than if they are based on human intervention only, or if the outcome even contradicts legal or moral assumptions (e.g. fairness by being non-discriminatory), then a (technically) possible use of AI may still not be a favourable option.

In challenging the appropriateness of ML in the context of legal decision-making, but possibly also in other fields such as political decision-making, the ultimate goal is to identify in which cases and to what extent it is “safe” to rely increasingly on such solutions and where the limits are. The project shall not have the goal of programming a specific algorithm that can be used in the legal domain.

Nor is the idea to analyse completely a specific legal domain such as criminal law and identify the specific problems in that area. What is needed is to define concrete cases with which it can be proven that a problem with the ML exists and there is no perfect solution to fix it. In that regard, the project shall be exploratory in nature, by showing what type of testing is necessary to find out deficiencies of AI systems – in this regard it relies on computer science expertise – and what consequences this has from a legal perspective. The latter is twofold: on the one hand, the domain of law and the judicial decision-making process will be used as blueprint for an area in which – possibly with good reason – there is quite some scepticism about using ML. On the other hand, legal norms such as fundamental rights will be the scrutiny benchmark to decide whether the outcome of testing means an AI system is insufficient. With this, the project shall give the basis to then further systematically research specific domains, specific AI solutions or specific testing methodology.

In the meanwhile, another aspect of the project is to present interim findings to a general public via an open conference as there is an increasing awareness of “AI entering everyday life” whilst for many this is a fearful prospect – and thereby there is a need for science to contribute to the societal discussion in Luxembourg.

MCI-BIOME: Relationship between socioeconomic status and the gut microbiome as a risk of dementia by A. Leist, P. Wilmes and R. Krüger

Dementia is one of the greatest scientific, medical and socioeconomic challenges of our times, as it affects more than 46 million people worldwide, with incidence numbers projected to double within the next 20 years. There is no medical treatment to prevent, halt or reverse the progression of neurodegeneration underlying dementia. Educational gradients in the onset and progression of dementia on top of the contribution of innate cognitive abilities suggest a role of socioeconomic factors in the development of this syndrome. At the same time, recent evidence for dysregulation of the gut microbiome in patients with dementia indicates that the microbiome may play a role in modulating brain function linked to neurodegeneration and suggests the presence of a diet-microbiome-gut-brain axis. Interestingly, socioeconomic factors can also influence food choices, which in turn may impact the microbiome composition. However, the underlying mechanisms of how the microbiome might contribute to neurodegeneration remain unclear, and little is known on how the composition of the gut microbiome is modulated by environmental and socioeconomic factors. The proposed Audacity project, MCI-BIOME, aims to further our knowledge of the relationship between socioeconomic factors, diet, and changes in the gut microbiome in relation to Mild Cognitive Impairment (MCI). MCI is considered a prodromal stage of dementia and represents a transition state between normal cognitive ageing and dementia. We focus on MCI as, at this stage, identification of individuals at highest risk to progress towards dementia for potential preventive interventions would be most relevant. The project relies on a unique interdisciplinary team composed of clinical experts in neurodegeneration, specialists in socioeconomic research, and specialists in high-resolution microbiome characterisation (integrated multi-omics) and big data analytics. Samples and data to be collected include stool samples of 60 MCI patients of the Programme Démence Prévention (pdp) and 60 healthy controls (in terms of cognition) as well as relevant information on demographic, socioeconomic, and diet-related variables. The main objectives of the project are

  1. to develop a novel interdisciplinary framework at the interface between social sciences, neurology and microbial ecology to open new research horizons and to strengthen the University's excellent research profile,
  2. to develop a multivariate microbiome biomarker model for MCI based on integrated multi-omic data by comparing MCI patients to a group of healthy controls,
  3. to explore how microbiome characteristics differ between sub-groups defined by socioeconomic factors including education and diet,
  4. to build a research infrastructure that will be vital for acquiring high-level follow-up funding for future projects including the analysis of progression of MCI in relation to microbiome shifts and the validation of the microbiome biomarker model in the large Luxembourg-based cohort of the National Centre for Excellence in Research on Parkinson Disease (NCER-PD) including individuals with and without MCI and dementia.

DSEWELL: Data Science and The Economics of WELLbeing by C. d’Ambrosio and A. Tkatchenko

We propose to bring together machine learning approaches, physics-inspired descriptors, and the economics of wellbeing to address questions broadly related to predicting life satisfaction and health of individuals in a data-driven manner. In the context of improving individuals’ life, one is often faced with the question of ordering individual/societal parameters (i.e. health and wealth) in terms of their importance. Up to now, state-of-the-art approaches relied on linear regression with very low Pearson correlation coefficients (R^2=0.2-0.3). The main goal of this project is to apply modern nonlinear machine-learning techniques to data on individual and social wellbeing with the aim to:

  1. understand the structure of the data and signal/noise ratio of many existing datasets,
  2. go significantly beyond linear regression with kernel-based methods and neural networks to search for multi-property correlations,
  3. assess different descriptors of individuals and metric definitions in ‘individual spaces’. To our knowledge, such fundamental ‘first-principles’ approaches to data analysis and nonlinear regression are only now starting to be applied to data on individual and social wellbeing, hence our project is timely and of potentially substantial impact.

Luxembourg Time Machine by A. Fickers (University of Luxembourg), E. Schymanski (University of Luxembourg) and L. Pfister (LIST)

There is hard evidence that global change is having severe and persistent impacts on environmental systems - eventually compromising socio-economic development in many regions of the world. Luxembourg is far from being sheltered from these threats. There is an pressing need for a better understanding of the involved processes, interactions and feedbacks, both via a better understanding of past evolutions and the development of new prediction tools. Here, we propose the “Luxemburg Time Machine” (LuxTIME) project for exploring radically new ways for analysing and interpreting factual evidence of the past. We intend to build an interdisciplinary framework for investigating “big data” of the past, inspired by the conceptual premises of the “European Time Machine” Flagship project (see point 2), While being a high-risk interdisciplinary project, we see multiple avenues for high reward. Within Luxembourg, LuxTIME has the potential to function as a “team building project” for the Team Luxembourg (LIST, LISER, LIH, University), fostering the energies and skills across two interdisciplinary centres at the University in cooperation with LIST - integrating historical, informatics, environmental and health expertise that can easily be expanded at later stages. By building a digital dataset that will include information from three different fields and scientific perspectives, namely eco-hydrology, medicine and history, LuxTIME will use a local showcase (i.e. the industrialisation of Belval / Minette region) as a testbed for methodological and epistemological reflections on how to study the impact of environmental changes on the health of the local population in a long term perspective. In combining past evidence from eco-hydrological studies (informing on water and pollutant sources, flow paths and transit times; non-stationarity of hydrological systems, and topographic / geological transformations), medical records (describing disease patterns, mortality rates, social/psychological well-being) and bio-chemical data (based on imaging and mass-spectrometry techniques as well as digital forensics on teeth) and history (archival sources documenting economic, social, political and cultural changes), LuxTIME will eventually study the past in completely new ways. By mixing “contextual information” based on archival evidence with “scientific evidence” derived from chemical, biological, or medical investigations, the project explores new ground in interpreting “big data of the past” in a truly interdisciplinary setting. The Belval-case is meant to critically test the analytical potential of a multi-layered research design which – this is the mid-term ambition – can be expanded into a national case study; that is the building of a real “Luxembourg Time Machine” including many different kinds of data from many different institutions (such as demographic, economic, climate and socio-economic data from LISER, STATEC etc.).