Large language models appear closer than ever to passing the Turing Test, yet true language understanding may still be far away. In this talk, Monojit Choudhury proposes “Turing Test 2.0”: evaluating AI not by its ability to imitate an average human, but by its capacity to model specific individuals shaped by context, culture, and experience. This shift offers a new perspective on intelligence, generalization, and human-centered AI with potential applications in education, therapy, and social science.

The workshop aims to bring together researchers from various fields and different cultural perspectives to examine how increasingly capable AI systems may influence the way people understand and experience the "self." As interactions with AI systems (including chatbots) become frequent and personal, AI-generated responses may shape self-image, decisions, and identity. At the same time, the line between humans and artificial agents is becoming less clear. These developments raise practical and conceptual questions about agency, authenticity, and what "real self" could mean in everyday life.

The AI revolution has accelerated in recent years, propelled by the widespread use of large language models (LLMs). Today, AI systems are not only transforming technical environments but also shaping our thoughts, emotions, and everyday linguistic practices. Our conference will explore the role of affective computing, emotionally laden human-machine interaction, conversational AI models, reinforcement algorithms, and recommender systems in the wake of the LLM revolution. In this light, we will discuss what we can learn about language—both in its explicit, logical, grammatical structure and in its emotional, expressive dimension—when AI accesses these depths of human expression.

The possibility of absolutely general quantification is often taken to be obvious, since even its critics appear to presuppose it. This paper critically examines and rejects standard arguments for that assumption. It then develops an alternative approach that acknowledges the inherent restrictions of every form of quantification. Second-order quantification is used as a case study to illustrate these restrictions and to contrast the resulting view with Hellman’s position.

Drawing on philosophical epistemology, the lecture introduces a conceptual framework for knowledge in animals and machines. Guest speaker Prof. Laurie Paul (Yale) uses it to develop a new notion of LLM knowledge and to propose a taxonomy spanning biological and artificial systems—from C. elegans to LLMs. This approach clarifies how knowledge is realized across these different domains while accounting for their distinct biological and physical foundations.

AI in search engines, on the phone, or in the defense industry – headlines about new developments seem to outdo each other every day. While the EU has adopted its AI Act, new technological possibilities are already emerging.
How can we use AI ethically and integrate our values into its design? The lecture explores ethical concerns and risks for individuals and democratic societies, while highlighting ways we can promote responsible technology development and use.

Following the success and strong interest in last year’s events, the online seminar series “Cross-Cultural Approaches to Desirable AI” continues this year, starting on October 22. The series aims to encourage intercultural and interdisciplinary discussion on the ethics of artificial intelligence (AI). Its primhttps://www.cst.uni-bonn.de/en/events/events-1/edit?_authenticator=b126c4a555b4087b7797604c18ad1632ef58074e#ary focus is to support and develop research on AI and social justice, particularly from intersectional feminist and anti-racist perspectives. “Cross-Cultural Approaches to Desirable AI” is a collaboration between the Universities of Cambridge, Bonn, and Tokyo, and the University of Europe for Applied Sciences.