AICS: Artificial Intelligence and Complex Systems

The special session aims to explore the intricate relationship between Artificial Intelligence (AI) and complex systems through Complexity Theory but also philosophy, delving into the understanding of modern AI systems such as deep learning architectures and large language models (e.g., GPT/ChatGPT, Gemini, Llama, Claude, etc.) as complex dynamic systems. With a focus on stochastic processes, explainable AI, cognitive approaches, multimodal learning, AI and security, and AI and ethics/bias, the session will serve as a multidisciplinary platform that extends beyond engineering to include cognitive science, (computational) linguistics, philosophy, and other relevant fields. Specifically, we welcome researchers and engineers in AI and complex systems theory, cognitive scientists, linguists, and philosophers as well as industry professionals seeking to apply complexity theory and, in general, multidisciplinary approaches, in AI solutions to submit their works to this special session to investigate on the following four objectives:

To investigate how Complexity Theory can offer invaluable tools for analyzing AI systems, particularly in the context of dynamic behavior, emergent properties, and stochastic processes.
To explore how AI can be employed to study and understand complex systems, including information granulation (Granular Computing) and multi-agent environments.
To establish new methodologies for measuring the intelligence and linguistic understanding of AI systems through the lens of Complexity Theory.
To discuss the future directions and next steps in the intersection of AI and Complexity
Theory in reaching artificial general intelligence (AGI).
To investigate the implications of complexity in AI for language models, meaning, and other linguistic aspects, as well as philosophical considerations, such as ones related to word models, the problem of meaning and representation, ethics of information.

AICS is part of the International Joint Conference on Neural Networks (IJCNN)

Submission deadline: 30 January 2025 (exstended)

Topics of interest include, but are not limited to:

Emergent Behavior and Stochastic Processes in AI Systems: Understanding how complex systems theory can explain the adaptive, emergent characteristics and stochastic outputs of AI.
Multifractal and Stochastic Analysis of AI: Utilizing complexity theory for the multifractal and stochastic analysis of AI systems.
Dynamic Systems Theory in AI: How dynamic systems theory can be applied to AI for better system analysis and prediction.
Explainable AI: Investigating methods for making complex AI systems understandable and interpretable.
Cognitive Approach in AI: Exploring how cognitive science can inform and improve AI systems.
Multimodal Learning: The role of multimodal learning in enhancing the capabilities of AI systems.
AI in Multi-Agent Complex Systems: Exploring the role of AI in the study and understanding of multi-agent complex systems.
AI and Security: Discussing the implications of complexity theory for AI in cybersecurity.
AI and Bias: Addressing the challenges and solutions for bias in AI systems.
Linguistic and Philosophical Aspects: Investigating the implications of complexity in AI for language models, meaning, and other linguistic aspects, as well as philosophical considerations (agency, intentionality, meaning, etc.).

Keywords: Artificial Intelligence, Complex Systems, Natural Language Generation, Large Language Models, Distributional Semantics, Compositionality, Quantitative Linguistics, Dynamical Systems, Explainable AI

2025 ed. - Presented works

Session Chair(s): Alessio Martino, Enrico De Santis

Chain of Thoughts-Based Prompting for Incident Detection Using Large Language Model

Lorraine Chambers (Conway Packaging Services (CPS) Ltd., Birmingham City University), Affiong Udokisai (Birmingham City University), Haitham Hassan Mahmoud (Birmingham City University), Khalid Ismail (Birmingham City University), Nazim Mohammed (Conway Packaging Services (CPS) Ltd.), Waldo Cervantes (Birmingham City University), Mohamed Gaber (Birmingham City University), Rehan Bhana (Birmingham City University)

High-level classification based on meta-graph of visibility graphs for Autism detection

Ricardo Lima Filho (Universidade Federal de Uberlândia), Robinson Sabino-Silva (Universidade Federal de Uberlândia), Murillo Carneiro (Universidade Federal de Uberlândia)

LSTM in Recursive Feedback Loops: A Study on Textual Evolution and Complexity

Enrico De Santis (University of Rome "La Sapienza"), Alessio Martino (LUISS University), Francesca Ronci (University of Rome "La Sapienza"), Antonello Rizzi (University of Rome "La Sapienza")

2025: A GPT Odyssey. Deconstructing Intelligence by Gradual Dissolution of a Transformer

Enrico De Santis (University of Rome "La Sapienza"), Alessio Martino (LUISS University), Edoardo Bruno (Sis.Ter. Pomos S.r.l.), Antonello Rizzi (University of Rome "La Sapienza")

LogReader: General-Purpose Log Analysis via Open-Source Large Language Models

Jiaxing Qi (Beihang University), Zhongzhi Luan (Beihang University), Shaohan Huang (Beihang University), Yao Lu (Beihang University), Hailing Yang (Beihang University), Depei Qian (Beihang University)