October 31, 2024 10:00 AM | TUM Think Tank | Richard-Wagner Str. 1, 80333 Munich

Who is conducting research on LLMs and ChatGPT, where are they from, and how does their work connect across different fields and countries? In this talk, Moses Boudourides from the Northwestern University shares his latest insides into this meta study, and discusses the networks of collaboration and influence formed among researchers working in this rapidly growing area of artificial intelligence.

Comprising over 50,860 unique publications, the study analyzes patterns of diversity in post-2020 publications on "LLMs" and "ChatGPT," specifically concerning gender representation, institutional and country collaboration, interdisciplinary research, amounts of supporting grants, citation networks among publications, co-authorship networks, and shared fields of research networks. Join us for a deep dive into the research field on LLMs.

Discover the Future of AI with Professor Martin Hilbert

Join us for an enlightening session with Professor Martin Hilbert from the University of California, Davis, as he delves into the transformative power and potential pitfalls of Large Language Models (LLMs). In this era marked by rapid technological advances, LLMs have surpassed the Turing test and are now reshaping industries from education to politics with their ability to derive meaning from mere language statistics.This talk will explore the significant productivity boosts AI has brought to fields like coding and discuss the broader implications of automating knowledge creation. What does the rise of AI mean for human uniqueness in knowing and understanding? Join us as we explore these questions and more, offering a comprehensive look at both the opportunities and challenges presented by the digital age.

About Martin Hilbert

Professor Martin Hilbert is renowned for his groundbreaking research on the societal implications of digitalization. With a dual doctorate in Economic and Social Sciences and Communication, his contributions range from pioneering studies on global information volumes to strategic digital planning with the United Nations. His insights have informed digital strategies in over twenty countries and continue to influence both academia and industry discussions worldwide.

Join us for a vibrant discussion delving into promising avenues and potential challenges on the horizon of AI technologies.

Matti Nelimarkka presents his work on methods for social inquiry combining computational social science and design-based research.

Matti Nelimarkka introduces his textbook entitled "Computational Thinking and Social Science" (SAGE, 2023) on June 19, from 12:00pm - 1:00pm, elaborating on the pedagogical challenges faced when introducing computational concepts to scholars from non-computational backgrounds. He expands on the chapters focused on validity, reliability, and the integration of social theory in computational social sciences. He also shares insights into his ongoing work on developing a Marxist Large Language Model (LLM).

Moreover, he showcases preliminary research aimed at examining how political ideologies manifest in designing digital systems. By posing thought-provoking questions like "What would a tax collection app designed by left-wing parties look like?" and "What would a conservative dating app entail?", he highlightes how digital platforms could serve as conduits for advancing political values. He illustrates this points by examining which kind of interfaces for political discussion are preferred by different parties. This exploration underlines the intersection of politics, design, and technology.

Matti Nelimarkka is currently doing part of his sabbatical at TUM SOT until June 23. People who wish to meet with him one on one or in small groups are cordially invited to email Matti at matti.nelimarkka@helsinki.fi.

When talking about AI and data regulation, we are looking at quite different approaches on the global political stage. For our newly established TUM Think Tank AI Speaker Series, our first guest, Silicon Valley based attorney Lothar Determan provided a comprehensive overview of the contrasting approaches to AI and data regulation in the European Union and the United States. He highlighted the stark contrasts between the US's reactive, issue-specific AI and data regulations and the EU's proactive, precautionary approach.

The event provided a deep dive into the regulatory philosophies of the US and the EU, offering attendees a clear understanding of the challenges and opportunities presented by each approach. Understanding these differences is crucial for businesses operating internationally, as it informs strategies for compliance, innovation, and data management. Read on to find out our main insights from the talk:

Regulatory Approaches

The US regulatory framework is characterized by a reactive, harm-focused approach. Regulations are often created in response to specific issues, resulting in a multitude of narrowly focused laws. This method allows innovation to flourish but can create a challenging landscape for companies to navigate due to the complexity and specificity of the regulations.
The EU on the other hand employs a precautionary principle, preemptively restricting activities until clear conditions for safe operation are established. This approach is particularly stringent when it comes to personal data processing, aiming to ensure high standards of data protection and security.

AI Regulation

AI regulation in the US is managed through a series of specific laws targeting particular issues. Key examples include:

• • California Business & Professions Code 17941: Requires the disclosure of AI bots.
  • New York City’s regulation on Automated Employment Decision Tools: Governs the use of AI in hiring processes.
  • Colorado AI Act (effective May 2024): Focuses on preventing algorithmic discrimination.
  • Presidential Executive Order: Provides guidance for federal agencies, though it leaves many details open-ended, making it distinct from the EU's comprehensive regulatory approach.

The EU AI Act primarily regulates high-risk AI systems, mandating extensive data for training these systems. However, there is still ambiguity regarding how the AI Act aligns with the General Data Protection Regulation (GDPR), particularly in terms of data requirements and privacy concerns.

Discussion Highlights:

• The US's approach allows for rapid innovation but requires companies to manage a complex array of specific regulations.
• The EU's precautionary stance ensures robust data protection but can be restrictive for businesses seeking to innovate.
• There is a significant difference in regulatory philosophy: the US intervenes when problems arise, while the EU sets strict rules upfront to prevent issues.

This discussion was invaluable for professionals navigating the complexities of compliance and innovation in these regions or students and legal scholars seeking to understand the dynamics of regulatory bodies, law, and the private sector. Many thanks to Lothar Determann for his insightful presentation and to all attendees for their participation.

Lothar Determann is an attorney in Silicon Valey focused on data privacy law compliance, information technology, artificial intelligence, copyrights, product regulations and international commercial law.

Discover the nexus where science, technology, and social justice converge to sculpt the future! Join our event on June 3rd featuring Alondra Nelson who gives insights from her research on race, health, and the interplay with science and technology.

Renowned scholar and author Alondra Nelson holds the esteemed Harold F. Linder Professorship at the Institute for Advanced Study. Formerly, Dr. Nelson held pivotal roles as deputy assistant to President Joe Biden and interim director of the White House Office of Science and Technology Policy (OSTP). Recognized by Nature and TIME100 for her significant contributions to science and technology policy, she is the author of the award-winning book "The Social Life of DNA".

to the program

In a dynamically ever-evolving technological landscape, the proliferation of artificial intelligence (AI) has become a defining feature of the digital transformation. As AI continues to permeate various aspects of society, questions regarding its ethical implications, societal impact, and regulatory frameworks have come to the forefront. This talk delves into the commonalities and differences of AI and data regulation between the European Union (EU) and the United States (US). While both regions grapple with similar concerns surrounding privacy, bias, and accountability, their approaches to regulation diverge significantly. Through a comparative analysis, we explore foundational principles, legislative initiatives, and enforcement mechanisms shaping AI and data governance on both sides of the Atlantic.

Professor Lothar Determann practices and teaches international data privacy, technology, commercial and intellectual property law. At the global law firm Baker McKenzie in San Francisco and Palo Alto, he has been counseling companies since 1998 on data privacy law compliance and taking products and business models international. He has authored more than 150 articles and contributions and many books, including Determann’s Field Guide to Data Privacy Law (5th edition 2022) and Determann’s Field Guide to Artificial Intelligence Law (2024).