For the public

25 November 2021
Violeta Ivanova-Rohling
Postdoctoral Fellow, Zukunftskolleg, University of Konstanz (Germany)

“Machine-learning Approaches for Debugging a Quantum Computer”

In the past decades, the mounting evidence that quantum algorithms can solve specific tasks with efficiency beyond the capability of a state-of-the-art classical computer has attracted tremendous interest in the field. A turning point was Shor’s algorithm for prime factorization, a polynomial quantum algorithm solving a problem that is hard for classical computers. A fully functioning all-purpose quantum device would have an enormous impact on our lives, with applications in science, drug discovery, disaster preparedness, space exploration and environmental sustainability, among many others. As a consequence, an increasing number of countries and companies are investing billions of dollars in a race to produce and commercialize the quantum computer. Various physical systems for quantum computation have already been developed, and hybrid quantum algorithms, which aim at solving optimization problems more efficiently, can run on existing noisy intermediate-sized quantum devices. However, a full-size general-purpose quantum computer is still out of reach. One of the difficulties in developing such a device is that as the size and complexity of the quantum computer grow, more sophisticated techniques for calibration and evaluation of their performance are required in order to develop fault-tolerant devices. Quantum state tomography (QST) is a prominent technique for the verification of a quantum computer, which allows for the reconstruction of a given quantum state from measurement data. By providing comprehensive information for a given quantum state, QST is known as the “gold standard” for the verification of a quantum device. However, its computational costs make it infeasible for a system larger than a few qubits. Moreover, it can be time-consuming even for small systems, i.e. building blocks of a quantum computer of only one or two qubits. Efficient QST would be an important step to making a general-purpose quantum device possible. One aspect of the efficiency of the QST procedure depends on the choice of the measurement scheme, which determines the number of measurements needed to perform the QST. Finding a measurement scheme that minimizes the number of required measurements can be formulated as an optimization problem. Ivanova-Rohling’s work focuses on applying and developing various optimization and machine-learning methods, with the goal of finding measurement schemes that minimize the number of measurements needed. By using prior knowledge of the landscape of potential solutions, such as particular symmetries and invariances, it would be possible to improve the exploration of search space and find optimal measurement schemes.

16 December 2021
Lino Camprubí
Ramón y Cajal Researcher, University of Sevilla (Spain)
Resident, Institut d’Etudes Avancées d’Aix-Marseille Université (France)

“Knowing the Earth in the Digital Era”

In the popular imagination, earth and environmental scientists are adventurous explorers ready to fight the elements in the field, on the mountain peak, in the dark cave or the blue ocean. This naturalist type still exists in real life. But much of the knowledge generated about the Earth in the last decades has come from remote sensors gathering and processing data that scientists then study on their computers. What is lost and what is gained? In this conference, the question was explored historically and philosophically, with special attention to the role of digital and satellite data in ocean sciences.

27 January 2022
Ruben Pauwels
Biomedical Scientist, Associate Professor, the Aarhus Institute of Advanced Studies (Denmark)

“Evolution of the Radiology Profession due to Artificial Intelligence” 

In this talk, a particular example of ‘Knowledge in a Digital Era’ was discussed. Digitization of medical imaging started well over 30 years ago and is still ongoing. Within this time span, certain innovations did not require a significant adaptation in radiological know-how (e.g. photostimulable phosphor plate), whereas other technologies required significant adaptations of training curricula and/or additional postgraduate certification (e.g. CT, MRI). Currently, artificial intelligence (AI) is revolutionizing the radiological landscape. Following the development of highly performant algorithms for deep learning, along with the ever-increasing computational power available to us, the use of AI for diagnostics as well as image processing has been explored. While the development of clinical AI tools is very much work in progress, the hype around AI has already raised several questions regarding its impact on the radiological profession. Although it is generally believed that radiologists will not become obsolete any time soon, and that the ultimate responsibility for a patient’s diagnosis and treatment will remain in human hands, several questions can be raised regarding the effect of AI on the required knowledge and competencies of future radiologists. Most of the current perspectives regarding the use of AI in radiology are somewhat simplified and static, and focus mainly on the ways in which AI can enhance the diagnostic workflow (‘augmented radiology’). The reality, however, may be much more complex and dynamic, and might require a significant alteration to the training requirements of radiologists or, perhaps, the inception of an entirely new profession.

31 March 2022
James D. Hollan
Professor of Cognitive Science and Computer Science, University of California, San Diego
2021/2022 Fellow, Paris Institute for Advanced Study

“Addressing the Challenge of Human-Technology Partnership in the Digital Era: A Human-Centred Information Space Approach”

“The computer desktop was an amazing design for its time, but does not reflect the complexity, flexibility, and sociality of human activity...Eventually we will have to reorganize the desktop to reflect the complex mix of activities users engage in and move beyond the rigidity of separate applications and files-and-folders.” – Bonnie Nardi, Acting with Technology: Activity Theory and Interaction Design, 2009.

For far too long we have conceived of thinking as something that happens exclusively in our heads. Thinking happens in the world as well as in the head. We think with things, with our bodies, with marks on surfaces and with other people. Increasingly, we think with computers, often the one we carry with us everywhere. Computers are now ubiquitous and embedded in virtually every new device and system, ranging from the omnipresent cell phone to the complex web of socio-technical systems that pervade and shape modern life. They connect our activities to ever-expanding information resources with previously unimaginable computational power. Yet with all the increases in capacity, speed and connectivity, computer-mediated information activities remain fragmented and frustrating.
Designing the future of work at the human-technology frontier is one of the ten long-term science and engineering challenges identified by the National Science Foundation in the US. In this presentation, Hollan argued that a core aspect of this challenge arises from an unquestioned view of information systems as collections of separate passive tools rather than active partners. The scale of information available and the sophisticated cognition demanded by contemporary information work has outpaced innovation in user interfaces. In modern computing systems, information is encapsulated in silos, leaving users to shuttle files between applications, cobbling together workflows, requiring troublesome context switching and increasing attentional demands. In short, we lack a human-centred information work space, a cognitively supportive visual space for intellectual work.
A human-centred information space is both an idea and a computational environment. It is the idea of a spatial cognitive workspace – a desktop for intellectual activity – reified as a computational environment that actively supports the coordination of information activities. It should develop awareness of the history and structure of a user’s action: how they accomplish activities through discrete tasks across devices, programmes and working sessions. Through use, each representation in the linked computational environment accumulates structure and context: not only who accessed it and when, but relationships to concurrent and other semantically related information and activities. This context and history of activity should drive the behaviour of information representations. To the user, their information should seem alive, have awareness, know where it came from, how it got there, what it means – and behave accordingly. These dynamic representations will in turn guide the user’s future action, providing a supportive personalized information context. It is important to emphasize that the human-centred information space will not replace the user’s ecosystem of documents and applications, but be a separate space linked to them, acting as a home, a control centre, a multi-modal but fundamentally ‘spatial workshop’, where information across applications will converge, augmented with visual features and active behaviours to support the user in not only completing their tasks, but accomplishing long-term overarching activities.

*This presentation benefitted from a FIAS Fellowship at the Paris Institute for Advanced Study (France). It has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 945408, and from the French State programme “Investissements d’avenir”, managed by the Agence Nationale de la Recherche (ANR-11-LABX-0027-01 Labex RFIEA+).

28 April 2022
Nadine Sutmöller
Postdoctoral Researcher, Center for Interdisciplinary Research, Bielefeld University
Coordinator, Research Group Economic and Legal Challenges in the Advent of Smart Products

“Filter Clashes and Democracy: The Dissemination of Information via Social Media and its Impact on Freedom”

Media and, increasingly, social media play a key role in democratic societies by disseminating information. However, the new instruments and their dynamics are fundamentally changing this game: as often explained, the new media are characterized in particular by the algorithmically controlled curation of content based on collected data. To avoid information overload, the filtering approach makes perfect sense. However, at the same time, this procedure raises questions that strike at the heart of a democratically ordered society: What impact do these activities have on individual freedom? What does this mean for shaping social coexistence? And what consequences do these developments have for the democratic system? The claim to grant citizens equal freedoms is an essential component of democracy. This enables people to work together, since everyone can assume that they do not have to fear for their place in society. Social networks and their dynamics are increasingly calling this essential requirement into question: the user-specific presentation of content creates different horizons of experience. The phenomenon of the filter bubble is referred to with concern in this context. The term describes the situation where users no longer participate in the opinions and experiences of others and thus distance themselves from one another. Another fear is that when these different horizons of experience collide, members of society no longer succeed in exchanging ideas with one another. Rather, a deep irritation and speechlessness is to be expected (filter clash). That this is not a wrong premonition was evident at the beginning of 2021, when increasingly hardened fronts could be observed in disputes in the wake of the 2020 US election, with participants facing each other aggressively. This development poses a threat to freedom in that members of society are increasingly given the feeling that they have to fight for and defend their own views and opinions. This can be impressively demonstrated not least by the storming of the Capitol on 6 January. In this situation, it was no longer possible to exchange views and develop common solutions based on fundamentally shared knowledge. This ultimately endangers democratic order, since social connectedness is increasingly lacking, and thus trust that cooperation is possible cannot be readily assumed. To consider the issues raised, Sutmöller linked different disciplines (including computer science, media studies) and, in particular, illuminated the situation from the perspective of political philosophy – especially in the context of John Rawls’s theory.

26 May 2022
Roar Høstaker
Professor of Sociology, Inland Norway University of Applied Sciences
2020 Fellow, Nantes Institute for Advanced Study

“Including or Excluding Context – from the Information Crisis to Natural Language Processing”

A common thread in the debates over the digital age, since its inception in the 1940s, is the question whether information must have a meaning or not. Or, whether a text depends on its context or not. Both meaning and context are at times kept away from being relevant topics in these debates, but they somehow come back again: from the controversies between librarians and documentalists in the 1950s to mass digitization of books and the contemporary controversy over the sources for the datasets in Natural Language Processing (NLP). The talk looked into some of these debates in order to show how and why the question of context and meaning tends to return.

30 June 2022
Jennifer Guiliano
Associate Professor, Department of History, Indiana University – Purdue University Indianapolis, USA
2022 Fellow, Institute for Advanced Studies in the Humanities, University of Edinburgh

“Decolonizing Knowledge Production through Linked Open Data”

A hallmark of the North American colonial process was the production and dissemination of knowledge about Indigenous peoples through the journals and records of colonizers. The violent and virulent practices that led to widespread disease, genocide, trauma and displacement in the Americas were bolstered by data collection and distribution that relied on the physical death and cultural destruction of Indigenous peoples. Equally as damaging were 20th century preservation efforts by non-Indigenous peoples that form the core of most cultural heritage collections. Analogue archival collections about Native American, First Nations and Indigenous peoples were constructed through “salvage” ethnography that sought to document “disappearing” peoples. Collectors, anthropologists and historians embarked on decades-long collecting efforts that led to the extraction (forcibly and otherwise) of cultural objects, knowledge and even physical bodies from Native communities. They created the data culture that most historians operate within as they work with indigenous materials. Historians are struggling to connect data and decolonize data practices so that they align with indigenous communities and their ways of knowing. This becomes further complicated by the fact that an overwhelming amount of historical data is held by colonial repositories and not Native communities who have different epistemological and cultural priorities. There are general ethical and epistemological issues to which researchers need to be attentive when exposing historical materials (especially photographs, documents and artifacts) authored by and about Indigenous peoples. First and foremost, there is the issue of identity politics: Who has the right to speak for/about whom and what role should non-members play in articulating a community’s history, authority or beliefs? Significantly, in colonial-centric collections, only legal access is required and/or commonly completed. Every community, every tribe, and even a single family might differ in their sense of what is appropriate for research or reuse and dissemination. When national borders divide those families, the question of research ethics becomes more complex. Can linked open data account for any of these issues or does it rely on colonial systems of knowledge production that cannot be teased apart from issues of rights and access? This presentation highlighted preliminary answers to these questions, while seeking to present a vision of what a collaborative, shared authority model of indigenous digital humanities and digital history would look like.