Donnerstag, 29. Februar 2024
Was ist GenAI?
Colossus 1970
Die Warnung vor Colossus
The legend of “Munich 1938”
Die Legende von „München 1938“
Mittwoch, 28. Februar 2024
What is GenAI?
Dienstag, 27. Februar 2024
Emergence has nothing to do with consciousness
Emergenz hat nichts mit Bewusstsein zu tun
Ein zentraler Punkt in der Debatte um Emergenz und Bewusstsein ist die Erkenntnis, dass Bewusstsein – charakterisiert durch die Sein Wahrnehmung – eine Qualität darstellt, die sich grundlegend von den Phänomenen unterscheidet, die typischerweise in KI-Systemen als "emergent" beschrieben werden. Während emergente Eigenschaften in KI-Systemen beeindruckend und unerwartet sein können, wie die Fähigkeit zur Lösung komplexer Probleme oder die Entwicklung neuartiger Strategien durch Schwarmintelligenz, liegen diese Phänomene in einer gänzlich anderen Kategorie als Bewusstsein.
Die Verknüpfung von Emergenz in KI-Systemen mit dem Auftreten von Bewusstsein zu vergleichen, muss aus einer kritischen wissenschaftlichen Perspektive als spekulativ und theoretisch unbegründet betrachtet werden. Die faszinierenden Entwicklungen in der KI-Forschung und die Beobachtung emergenter Phänomene erweitern zweifellos unser Verständnis von Komplexität und Informationsverarbeitung. Sie führen jedoch nicht zu einem tieferen Verständnis des Bewusstseins oder bieten eine Grundlage für die Annahme, dass Bewusstsein in KI-Systemen emergieren kann.
Donnerstag, 22. Februar 2024
Sora: From Text to Film
The introduction of Sora by OpenAI marks a significant milestone in the development of artificial intelligence (AI). Sora, a text-to-video model, represents the latest advancement in AI's ability to generate complex media content. It highlights the rapid progress in the field of generative AI and raises questions about the future role of AI in media production, creative endeavors, and information dissemination.
Sora builds on the achievements of its predecessors, such as DALL-E 3, a text-to-image model, and expands them to include the generation of moving images. The technology uses a denoising latent diffusion model, supported by a transformer, to create videos from text descriptions in latent space and then transfer them to standard space. This ability to create detailed and visually appealing videos relies on extensive training data, including publicly available and licensed videos.
The videos produced by Sora, ranging from creative scenarios to realistic depictions, demonstrate the immense creative potential that AI-powered systems can bring to the media and creative industries. By automating video creation, content could be produced faster, more cost-effectively, and in greater variety. Sora also offers opportunities for education, training, and entertainment by visually representing complex concepts or simulating historical events and future scenarios.
At the same time, Sora's capabilities raise serious questions regarding responsibility and ethics in AI development. The risk of generating disinformation, manipulating imagery, or creating inauthentic content requires strict control mechanisms and ethical guidelines. While OpenAI has implemented safety practices that restrict the creation of content with sexual, violent, hateful images, or depictions of celebrities, the effectiveness and enforceability of these measures remain critical questions.
Business Intelligence in the Age of AI
Digitalization and the increasing use of data have revolutionized business analytics and business intelligence in recent years. Where data had to be manually aggregated from databases and reports had to be created in the past, modern business intelligence tools like Power BI or Tableau are now used. These tools enable the merging of data from various sources and present it in meaningful dashboards and reports.
Another trend currently shaping business intelligence is the use of artificial intelligence. AI systems can evaluate and analyze data on a scale and in real-time previously unimaginable. They help companies derive valid insights and forecasts from vast amounts of data. Machine learning aids in identifying patterns and correlations that remain hidden to the human eye.
The following areas particularly benefit from AI in business intelligence:
- Prediction: Using predictive analytics, AI systems can forecast future trends and developments such as demand, growth, or risks. This helps leaders make informed decisions
- Personalization: Based on historical data, AI models learn to understand customer behavior. On this basis, products, services, and campaigns can be individually tailored to customers
- Process optimization: By capturing and evaluating processes, resource use, and error analyses, AI can identify weaknesses and show potential for improvement
- Automated analyses: Many simple or routine evaluations and reports can now be fully automated by AI systems. This leaves analysts more time for creative and strategic tasks
However, the use of AI also brings new challenges for business intelligence. Insights derived from AI must always be questioned and validated. Moreover, the transparency and traceability of the results are often limited. It is important for companies to select the right AI tools and train employees accordingly to fully leverage the benefits of AI-based analyses.
AlphaGeometry – The Digital Archimedes
AlphaGeometry could go down in the annals of artificial intelligence as the next revolutionary AI system, inspired by the achievements of Google DeepMind. This system promises to surpass the current state of the art in geometric problems by achieving a precision and efficiency that human experts have not considered possible until now. The ability to prove new geometric theorems and triumph in mathematical competitions is just a fraction of what sets AlphaGeometry apart.
From AlphaGo to AlphaFold, DeepMind has already demonstrated how its systems can be groundbreaking in their respective fields. AlphaGeometry builds on this legacy, striving to radically change our understanding of geometry. With the help of deep neural networks and symbolic logic, AlphaGeometry can recognize geometric patterns and use these insights to provide proofs that were previously unreachable. This could be invaluable not only for science but also for practical applications such as cryptography, materials science, and robotics.
The vision that AlphaGeometry embodies is that of a system that can accelerate scientific research and technological development, as well as fundamentally change the way we approach and solve problems. Deciphering geometric problems that once seemed insurmountable could pave the way for groundbreaking new discoveries and help tackle some of humanity's greatest challenges. From combating climate change through more efficient energy sources to curing diseases through a better understanding of biological structures – the impacts of AlphaGeometry could be profound.
In an optimistic vision of the future where AlphaGeometry becomes a reality, we could witness a new era of science and technology where AI is seen not just as a tool but as a partner in the quest for knowledge and progress. The possibility that AlphaGeometry demonstrates the ability to reason logically, potentially even competing with human scientists, opens a discussion on the role of AI in our society.
Looking to the future, AlphaGeometry could pave the way for developments in robotics and towards general artificial intelligence (AGI). The question arises whether and how such systems could expand and change our understanding of consciousness and cognitive abilities. Research in this area could benefit enormously from such advances, leading to a deeper understanding of AGI that goes beyond the limits of current technology.
Zum 80. Geburtstag von Wolf Singer
Am 9. März 2023 feierte der renommierte Hirnforscher Wolf Singer seinen 80. Geburtstag. Singer hat in seinem Leben viel geleistet und viele wichtige Erkenntnisse zur Erforschung des Gehirns beigetragen. Seine Arbeit hat nicht nur in der wissenschaftlichen Gemeinschaft Anerkennung gefunden, sondern auch in der breiten Öffentlichkeit große Aufmerksamkeit erlangt.
Wolf Singer wurde am 7. April 1943 in München geboren und studierte Physik und Medizin. Nach seiner Promotion in Medizin an der Ludwig-Maximilians-Universität München zog er 1973 in die USA, um als Forscher am Max-Planck-Institut für biophysikalische Chemie in Göttingen zu arbeiten. Dort entwickelte er gemeinsam mit David Hubel und Torsten Wiesel die Hypothese, dass bestimmte Nervenzellen im visuellen Kortex des Gehirns auf bestimmte Reize reagieren. Diese Entdeckung war bahnbrechend und bildet bis heute die Grundlage für das Verständnis der Funktionsweise des Gehirns.
In den Jahren 1980 bis 2005 leitete Singer das Max-Planck-Institut für Hirnforschung in Frankfurt am Main, wo er eine Vielzahl von Forschungsprojekten initiierte und leitete. Er prägte die moderne Hirnforschung durch seine Erkenntnisse zur neuronalen Plastizität und zur Funktion von neuronalen Netzwerken im Gehirn.
Singer wurde für seine Arbeit vielfach ausgezeichnet, darunter mit dem renommierten Brain Prize im Jahr 2013. Er war Mitglied der Nationalen Akademie der Wissenschaften Leopoldina und der Berlin-Brandenburgischen Akademie der Wissenschaften sowie Ehrenmitglied der Royal Society of Edinburgh.
Besonders bekannt ist auch seine Zusammenarbeit mit dem französischen Neurobiologen Francisco Varela und dem buddhistischen Mönch und Autor Mathieu Ricard. Gemeinsam erforschten sie die neuronalen Grundlagen von Meditation und Mitgefühl und prägten damit eine neue Richtung in der Neurowissenschaft, die sich mit dem Zusammenhang zwischen Geist und Gehirn beschäftigt.
Wolf Singer hat durch seine Arbeit die moderne Hirnforschung geprägt und zahlreiche Erkenntnisse zur Funktionsweise des Gehirns geliefert. Seine Forschung hat nicht nur in der wissenschaftlichen Gemeinschaft Anerkennung gefunden, sondern auch in der breiten Öffentlichkeit für Aufsehen gesorgt. Gratulation zum 80. Geburtstag und alles Gute für die Zukunft!
Wie entstehen demokratieferne Einstellungen in einer Kommune?
Raiko Hannemann
Im Rahmen der Forschungslinie „Zusammenhalt stärken in Zeiten von Krisen und Umbrüchen“ des Bundesministeriums für Bildung und Forschung erforschte das Projekt mit diesem Titel die Entstehung demokratieferner Auffassungen in einer Kommune am Beispiel des Berliner Bezirks Marzahn-Hellersdorf, einer ostdeutschen Kommune. Aufschlussreich waren biografische Interviews im Anschluss an eine repräsentative Befragung. Es wird gezeigt, dass Menschen sensibel auf soziale Verwerfungen reagieren und sich insbesondere dann von Demokratie abwenden , wenn diese ihr Versprechen auf soziale Gleichheit nicht einhält. Dennoch ist eine hohe Bereitschaft zu Engagement erkennbar. Dieses Potenzial könnte ausgeschöpft werden, wenn die Sprachlosigkeit zwischen Politik und Bevölkerung überwunden wird. Viele Interviewpartnerinnen äußerten sich froh darüber, dass sich endlich jemand für ihre Meinungen interessiert. Auf der anderen Seite sind die Möglichkeiten, sich in etablierten politischen Formen zu engagieren, sehr hochschwellig.
Als schwer erreichbar sind also nicht die sozial Ausgeschlossenen, sondern die etablierte Politik. Ansätze zu einer sozialräumlichen Demokratieentwicklung und zu einer Überwindung der Lücke zwischen etablierter Politik und den Menschen werden unter anderem in einem Ausbau von Räumen für nachbarschaftliche Begegnung gesehen sowie in einer verstärkten Demokratiebildung in Schulen, einer größeren Vernetzung von Stadtteilarbeit auch im digitalen Raum und in Formen aufsuchender politischer Bildung. Das Projekt stand in einem fachlichen Austausch mit den Universitäten Bremen, Göttingen und Jena, welche ebenfalls zu der Frage der Stärkung lokaler Demokratie geforscht haben. Zentrale Erkenntnisse werden im Hinblick auf ihren gesellschaftlichen Transfer gemeinsam diskutiert.
Link zum 1. Teil des Interviews mit Raiko Hannemann
Link zum 2. Teil des Interviews mit Raiko Hannemann
What is AGI?
In contrast to the currently existing AI systems that are designed for specific tasks and referred to as "weak AI," AGI (Artificial General Intelligence) encompasses the endeavor to develop machine intelligence that has the capability to learn, understand, and apply any intellectual task that a human being can perform.
The idea behind AGI is not new; it is rooted in the origins of Artificial Intelligence as an academic field in the 1950s and 1960s when scientists first developed the vision of a machine that could simulate human intelligence in its entirety. This vision has evolved over the decades and has become more precise, while the gap between current AI, which is limited to specialized tasks, and the ambitious goal of AGI has become increasingly apparent.
The distinction between "weak" and "strong" AI is central to understanding the concept of AGI. Whereas "weak" AI systems are designed to perform certain tasks with efficiency and precision predefined by humans, "strong" AI or AGI aims to achieve a universal problem-solving capability that equals or even surpasses human intelligence. Such a system would not only have a broad applicability but could also independently learn, adapt, and perhaps even develop its own consciousness and emotions.
The definitions of AGI vary greatly and reflect the diversity of perspectives and research approaches in this area. Some definitions emphasize the importance of cognitive flexibility and the ability to solve new problems without prior specific programming. Others focus on the concept of consciousness or the machine's ability to understand and simulate human emotions and experiences.
The question of when AGI might be achieved is the subject of intense debate among scientists, technologists, and philosophers. Estimates range from optimistic assumptions that expect breakthroughs in the coming years to skeptical forecasts that question the realization within the next centuries or even the fundamental impossibility of true AGI. These disagreements are based on different assessments of the complexity of the human brain, the limits of computer technology, and the ethical and societal implications of such a development.
The VDW Statement on the Asilomar Principles for Artificial Intelligence
The Association of German Scientists (VDW) views the Asilomar Principles as a valuable basis for discussion on ethical and regulatory issues in the field of Artificial Intelligence (AI). These principles offer a starting point to address the potential opportunities and dangers of AI. However, the VDW emphasizes that the Asilomar Principles alone do not provide a sufficient normative framework, particularly because they do not adequately limit the possibility of uncontrollable progress in AI technology. The organization advocates for a stronger consideration of the precautionary principle, the necessity to evaluate the limits of technology development, and calls for comprehensive investigations into the impacts of AI on society, culture, and human rights. A broad, democratically legitimized discourse and international cooperation are seen as essential for the development of safe AI systems. Furthermore, the VDW recommends initiating scientific research on impact assessment, ethics, and legal frameworks to inform political decision-makers at an early stage. The VDW also deems it necessary to implement bans and moratoria on certain AI applications, such as autonomous weapons, as well as to maintain privacy and informational self-determination.
The Renaissance of Moore's Law
Moore's Law, named after Gordon Moore, co-founder of Intel, is one of the fundamental theses that shaped the rapid development of computer technology over decades. In 1965, Moore posited that the number of transistors on a microchip would double approximately every two years, which equates to an exponential increase in computing power while simultaneously reducing costs. This law has established itself as a guideline for the semiconductor industry and has driven the development of ever more powerful and efficient processors.
However, in recent years, Moore's Law has encountered physical and economic limits. The miniaturization of transistors approached the limits of quantum mechanics, which meant that performance improvements could not continue at the accustomed pace. In addition, the costs of developing and manufacturing advanced microchips grew exponentially, leading to a slowdown in the previous doubling rate. These developments led to the assumption that Moore's Law is reaching its limits and can no longer serve as a reliable forecast for the future development of chip technology.
A promising approach to maintain Moore's Law lies in the research of new materials such as graphene. Graphene, a single-atom layer of carbon atoms arranged in a two-dimensional lattice, is characterized by exceptional electrical, thermal, and mechanical properties. These make it an ideal candidate for the development of next-generation microchips. Graphene-based transistors could be significantly faster and more energy-efficient than their silicon-based counterparts. This breakthrough has the potential to continue the miniaturization of transistors and the increase in computing power, which could initiate a renaissance of Moore's Law.
The further development and implementation of graphene in microchip technology are still in their infancy but hold the potential to revolutionarily change computer technology and, in particular, artificial intelligence (AI). Higher computing performances with lower energy consumption would enable complex AI algorithms to be operated more efficiently, which, in turn, could accelerate the development of more advanced AI systems. This development is crucial as the demands for computing power and energy efficiency in AI grow exponentially. Overcoming the current limits with innovative materials such as graphene could not only carry Moore's Law forward but also have a decisive influence on the future development of AI and its applications.