Zum Hauptinhalt springen

Programm 27. April

Öffentliches Programm – Livestream: kicamp2021.de

Mit Prof. Dr. Ina Schieferdecker (BMBF) und Daniel Krupka (GI)

Speaker*innen: Dr. Michael Meister, Prof. Dr. Ziawasch Abedjan, Prof. Dr. Hiroaki Kitano, Prof. Dr. Jana Koehler

Beschreibung:

AI technologies are no longer just research objects. Robotics, knowledge-based systems and pattern recognition, among others, are increasingly revolutionising global research and enabling new ways of gaining knowledge across disciplines. However, AI systems are also increasingly becoming the driving force behind economic and social developments outside of the academic world. Researching and shaping these systems requires transdisciplinary approaches.
The panel "Visions for the Future of AI Research" addresses the future of international AI research with a special focus on Germany and Japan. The thematic framework ranges from the political framework conditions of AI research, such as the AI strategy of the German government and the European Union, to new transnational research approaches and the question of how European AI research can become even more transdisciplinary, inclusive and diverse in the future.

Speaker: Prof. Dr. Peter Dayan

Beschreibung:  

Reinforcement learning has become a wide and deep conduit that links ideas and results in artificial intelligence, computer science, statistics, control theory and economics to a near century's worth of psychological data on animal and human decision-making, and a fantastic wealth of findings concerning the neural basis of choice. There is a ready and free flow of ideas among these disciplines, providing a powerful foundation for exploring some of the complexities of both normal and abnormal behaviors of humans, other animals and machines. I will provide an overview, illustrating the themes with examples showing how far we have come – and how far we still have to go.

Speaker*innen: Horst Hörtner, Lily McCraith, Dr. Drew Hemment
Moderation: Paula Böhme

Beschreibung:

Art functions both as a seismograph of social developments associated with artificial intelligence and as a catalyst for the successful and socially responsible transformation of this new technology into products. It can open up new perspectives on the subject and take the discussion out of specialist circles and into a broader social stratum. The session "AI + Art", organised by the Goethe-Institut, not only aims to shed light on the potential of art in terms of critically observing and positively influencing technological change. It also aims to question the extent to which art itself can benefit from this change. After all, the skills of artists can be extremely expanded through the use of algorithmic tools. With the help of AI, entirely new aesthetic forms of expression can be realised. Three speakers, all of them border crossers between computer science and art, will discuss the extent to which works of art represent innovative solutions for a society in upheaval or portray the influence of technology on people. At the interface of art, technological innovation, economy and society, all three artists are working on the interconnection of the methods of creativity and AI in order to critically accompany developments and at the same time shape visions of the future.

Speaker*innen: Prof. Dr.-Ing. Ina Schieferdecker, Kenza Ait Si Abbou Lyadini, Dr. Mario Gleirscher, KI-Newcomer*in 2021
Moderation: Frithjof Nagel

Beschreibung:

AI research not only includes different groups of actors, such as universities, research institutes, foundations and companies, but also a wide variety of disciplines.  One common feature, however, is that a large part of the research is carried out by junior scientists. They often work and research under very different conditions depending on the institution or discipline, if one compares, for example, businesses with universities or computer science with the humanities. With this in mind, this panel discussion will address what junior researchers need in order to conduct cutting-edge AI research in Germany.

Nach der öffentlichen Nominierung, einem öffentlichen Publikumsvoting und der Bewertung durch eine hochkarätige Jury stehen die zehn KI-Newcomer*innen 2021 endlich fest. Heute ist es so weit: Die jungen Forscher*innen der Disziplinen Informatik, Natur- und Lebenswissenschaften, Geistes- und Sozialwissenschaften, Technik- und Ingenieurwissenschaften und Kunst werden öffentlich bekannt gegeben und geehrt

Zur Auszeichnung: Der KI-Newcomer*innen-Award zeichnet junge Talente aus, die mit ihrer Forschung die Entwicklung der KI in Deutschland besonders vorantreiben. Ziel ist es, die verschiedenen Facetten der KI-Forschung einer breiten Öffentlichkeit zugänglich zu machen, den Dialog zwischen den wissenschaftlichen Disziplinen zu fördern und ein Schlaglicht auf die junge KI-Community zu werfen.

Die Verleihung des Awards erfolgt durch Dr. Michael Meister und Prof. Dr.-Ing. Ingo Timm.

 

Speakerin: Dr. Timnit Gebru

Beschreibung:

Feminist and race and gender scholars have long critiqued "the view from nowhere" that assumes science is "objective" and studied from no particular standpoint. In this talk, I discuss how this view has resulted in a hierarchy of knowledge in machine learning and related fields, devaluing some types of work and knowledge (e.g. those related to data production, annotation and collection practices) and mostly amplifying specific types of contributions. This hierarchy also results in valuing contributions from some disciplines (e.g. physics) more than others (e.g. race and gender studies). With examples from my own life, education and current work, I discuss how this knowledge hierarchy limits the field and potential ways forward.

Abschließende Worte von Prof. Dr. Ina Schieferdecker (BMBF) und Prof. Dr. Hannes Federrath (GI-Präsident)

Exklusivprogramm für vorab angemeldete KI-Camp Teilnehmende

Gesellschaft

Speaker*innen: Barbara Rolf, Moritz Riesewieck, Hans Block, Moderation: Lorenz Widmaier

Beschreibung:

Ist weltliche Unsterblichkeit dank KI nahende Realität oder digitaler Irrglaube? Können uns errechnete Doppelgänger ersetzen oder ist der Mensch unberechenbar? Verheißt Künstliche Intelligenz auf der Datenbasis Verstorbener deren Auferstehung oder nur lebendige Erinnerung für Hinterbliebene? Wollen wir das ewige Leben auf Erden oder haben Sterblichkeit und Abschied einen zu schlechten Ruf?

"Dank der atemberaubenden Fortschritte maschinellen Lernens scheint die Überwindung des Todes zum Greifen nah” – so Moritz Riesewieck und Hans Block in ihrem Buch "Die Digitale Seele".

Die beiden Autoren diskutieren gemeinsam mit Barbara Rolf, Bestatterin und Theologin, und Lorenz Widmaier, der zu digitalen Nachlässen forscht, über das Leben nach dem Tod.

Speaker*innen: Nikolas Becker, Leonie Beining

Beschreibung:
Insbesondere beim Einsatz von KI-Systemen im Arbeitsumfeld zeigen sich die Risiken der Technologie: Mögliche Diskriminierung in Personalentscheidungen, Sicherheitsrisiken in der Produktion oder schlicht fehlende Robustheit der Systeme. Welche Rolle können Testing- und Auditing-Verfahren spielen, um diese Risiken zu minimieren? Dieser Frage widmet sich das Forschungsprojekt "ExamAI – KI Testing und Auditing". In der Session werden wir uns einerseits die rechtlichen und ethischen Anforderungen an KI genauer ansehen und andererseits erste Forschungsergebnisse zu den Möglichkeiten aber auch Grenzen von KI-Testverfahren vorstellen.

Wissenschaft

Speakerin: Prof. Dr. Gentiane Venture

Beschreibung:

In this session you will learn how data science can pair with motion science to create new robots that are engaging, complex and unpredictable.
While many talk about robots taking over humans, we still have time before this happens, however artificial intelligence in its weak form: data science, has a tremendous impact in robotics and its development, in particular in anything that relates to interacting or living in close contact with humans. Robots can be programmed to react to extremely complex sensory inputs from humans such as actions and moods, paving the path for personalization.

Speakerin: Prof. Dr. Ute Schmid

Beschreibung:

Machine learning is considered as an important technology with high potential for many application domains in industry as well as society. Impressive results of deep neural networks, for instance for image classification, promise that complex decision models can be derived from raw data without the need of feature engineering (end-to-end learning). However, there is an increasing awareness of the short-comings of data-intensive black box machine learning approaches: For many application domains it is either impossible or very expensive to provide the amount and quality of data necessary for deep learning. Furthermore, legal or ethical or simply practical considerations often make it necessary that decisions of learned models are transparent and comprehensible to human decision makers. Consequently, AI researchers and practitioners alike proclaim the need for the so-called 3rd Wave of AI to overcome the problems and restrictions of an AI which is focusing on purely data-driven approaches. In the talk, it is shown that machine learning research offers many alternative, often less data-intensive, approaches. Current topics and approaches for explainable and interactive machine learning will be introduced and illustrated with some example applications.

 

Nachhaltigkeit

Speaker: Dr. Roy Schwartz

Beschreibung:

The computations required for deep learning research have been doubling every few months, resulting in an estimated 300,000x increase between 2012 and 2018. These computations have a surprisingly large carbon footprint. Moreover, the financial cost of the computations can make it difficult for researchers, particularly those from emerging economies, to engage in deep learning research. In this talk I will discuss the conditions that led to this unprecedented growth in cost, touching on the deep learning revolution in AI, and the contextual representation revolution in natural language processing. I will then suggest ways to mitigate the negative side effects of this growth.

Speaker*innen: Dr. Nishtha Srivastava, Prof. Dr. Horst Stöcker, Dr. Alexander Kies

Beschreibung:

The possibilities to solve big data based complex problems with deep learning and machine learning (DL/ML) are undeniable. Owing to improvement in computation & storage devices and better data sharing platforms, these novel methods have gained prominence and opened up several new avenues in data intensive fields such as seismology, renewable energy, weather data modelling and so on. In this session, we will talk about various projects which are currently being executed at Frankfurt Institute for Advanced Studies within the field of sustainability. We will explain how we are using DL/ML-based algorithms to solve real life problems affecting humanity. 

Mobilität und Smart Spaces

Speaker: Prof. Dr. Iyad Rahwan

Beschreibung:

Machine intelligence plays a growing role in our lives. Today, machines recommend things to us, such as news, music and household products. They trade in our stock markets, and optimise our transportation and logistics. They are also beginning to drive us around, play with our children, diagnose our health and run our government. How do we ensure that these machines will be trustworthy? This talk explores various psychological, social, cultural and political factors that shape our trust in machines. It will also propose an interdisciplinary agenda for understanding and improving our human–machine ecology.

Speaker*innen: Dr. Suzanna Randall

Beschreibung:

Schon als Kind hat mich das Weltall mit seinen unendlichen Weiten fasziniert. Ich wollte alles darüber erfahren und gerne auch selbst dorthin fliegen!
So entschied ich mich als junge Erwachsene, Astrophysik zu studieren. Inwischen arbeite ich als Wissenschaftlerin bei der Europäischen
Südsternwarte (ESO) und bin dort für das ALMA-Teleskop tätig. Eigentlich mein Traumjob – doch dann sah ich eine Annonce, die mein Leben
verändern sollte: "Astronautin gesucht!". Vier Jahre später stecke ich mitten im Training des privaten Raumfahrt-Startups "Die Astronautin" mit dem Ziel,
als erste Deutsche zur internationalen Raumstation ISS zu fliegen. Ich berichte vom harten Auswahlverfahren, vom Spaß und den Herausforderungen
des Trainings, von unserer Mission und gehe auf KI im Rahmen unserer Arbeit ein.

Gesundheits- und Lebenswissenschaften

Speaker: Dr. Hector Zenil

Beschreibung:

AI in medicine and healthcare is getting great attention but often bold, and often unjustified, claims magnified by media coverage are made. In this talk, I will cover some of the very important challenges and limitations of AI in medicine and healthcare and how to take a responsible approach using approaches from neurosymbolic computation as opposed to simply statistical machine learning. I will introduce Algocyte and the work that we have done and continue doing at Oxford Immune Algorithmics to make the introduction of AI in medicine responsible for value-based and precision healthcare. We believe that the best AI is the kind that does not need to be presented as AI but is translated into a tangible benefit for all stakeholders, including payers, clinicians, and patients.

Speakerin: Jun.-Prof. Dr. Lena Kästner

Beschreibung:

Correlation does not imply causation. Nowadays, that is almost a truism. Nevertheless, correlation-based computational models are increasingly used to study, predict, and explain complex systems’ behavior. This is not only true for climate science but also for traditionally laboratory-, field- or clinic-based disciplines such as genetics, biology, and psychiatry.

Based on large amounts of data, powerful computers can build models of complex systems such as human beings or neural populations in the brain. When presented visually, these models often look like networks: variables are depicted as nodes connected by weighted edges. These networks invite a causal reading: If I change something in variable X over here, that will cause a change in variable Y directly connected to it.

Indeed, this kind of reasoning is familiar from discussions about structural causal models in philosophy of science that consist of nodes representing components or causally relevant factors and directed edges representing causal relations between them. But despite these similarities, it is far from clear how exactly network models and causal modeling relate. After all, correlation is not causation. But does it point us in the right direction?

After a short talk, we will enter into a discussion on the matter.

Kunst und Medien

Speakerin: Emily L. Spratt, Ph.D.

Beschreibung:

While generative AI has allowed artists to engage the digital medium in ever-transformative ways and digital ledger technologies have enhanced our record-keeping abilities, these tools have also been used to intentionally distort information or to promote ethically questionable practices around art and media. Often incurring significant socio-political and financial consequences, the uses of AI and blockchain technologies are often cast as either good (GANs and NFTs) or bad (deepfakes) players in the increasingly complex world of digital innovations. In this lecture/discussion, the long-standing question of authenticity in art and media will be addressed and reassessed in light of the recent applications of AI and blockchain in the arts.

 

Speaker: Lawrence Lek

Beschreibung:

Who will the artists of the future be? When AIs attain superhuman levels of creativity, will we embrace or exile these non-human creators? In my 2017 science fiction film Geomancer, an international group of pro-human activists formulate the 'Anti-AI Art Law' because of their fears about the rise of creative AIs. As automated technology continues to challenge traditional notions of human authorship, creativity in the new millennium will become an increasingly political frontier. Using examples from my open-world video games and films, I will explore how I use video game engines to explore the relationship between human and machine creativity.

 

Produktion

Speaker*innen: Ann-Kathrin Koster, Dr. Thorsten Thiel, Kaan Sahin, Moderation: Dr. Jeannette Neustadt

Beschreibung:

In den letzten Jahren mussten wir lernen, dass unsere demokratische Ordnung keinesfalls selbstverständlich ist. Die politische Meinungsbildung vollzieht sich zunehmend im Internet, antidemokratische Tendenzen in der Gesellschaft nehmen zu, gefälschte Nachrichten fördern rassistische und sexistische Einstellungen. Die immer rascher voranschreitenden Entwicklungen im Bereich der Künstlichen Intelligenz werden für Filter Bubbles, Fake News, Manipulationen, Diskriminierungen und eine immer effizientere Überwachung von Menschen und mithin für die Zersetzung fundamentaler demokratischer Ideen verantwortlich gemacht. Darüber hinaus verlieren Demokratien die Möglichkeit, große Tech-Konzerne und Plattformen zu kontrollieren, während diese die Gesellschaft zunehmend steuern können. Fest steht: Die Künstliche Intelligenz fordert die Demokratie heraus. Wie genau, erfahrt ihr in dieser Session.

Speaker*innen: Grayson Earle, Sabina Hyoju Ahn, Danae Tapia, Shirin Fahimi, Peter Polack

Beschreibung:

This session presents the work of five artists who use computational media to challenge normative ideas of perception, communication, justice, and rationality. Through research threads, that concern policing, algorithm regulation, practices of divination and biological systems, the artists will discuss how creative engagements with algorithmic and digital systems can inform our understanding of social relationships at the same time as they open up new avenues for knowledge discovery broadly conceived. Following an introduction and screening of five projects – "Sonomatter", "Diagram", "why don’t the cops fight each other?", "Digital Witchcraft: Automated Oracular Poetry" and "Umm-al-Raml / Sand Narratives" – the session will open up for a Q&A and discussion.