Keynotes

Abstract

Human-AI coevolution, a process in which humans and AI algorithms continuously influence and adapt to each other, increasingly characterises our society. AI-based recommender systems, including generative-AI-based tools, play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human behaviour on online platforms, such as social media, online marketplaces, geographic mapping and chatbots. The interaction between humans and AI-based recommenders generates a feedback loop: users’ choices shape the data used to train algorithms, and in turn, these algorithms influence future user choices, initiating a potentially never-ending cycle. This feedback loop may lead to unexpected and often unintended social consequences; nonetheless, it is understudied in AI and complexity science. Our research aims at forging a synergy between AI and complexity science to address several challenges:

1. understanding human-AI coevolution studying the feedback loop’s effect on the structure and dynamics of AI-aided complex systems;
2. modelling the fundamental mechanisms of human-AI coevolution to capture specific and universally emerging patterns;
3. governing human-AI coevolution developing complexity-informed machine learning models to manage the systemic feedback loop’s effects;
4. designing a framework of responsible and transparent AI-based recommenders to dynamically balance individual and collective goals.

These research activities are grounded in empirical and simulation-based experiments on four key human-AI ecosystems: social media, geographic mapping, online retail, and chatbots based on large language models. We aim at advancing an open science of human-AI coevolution able to understand the impact of AI-based recommenders on society and develop potential tools to manage it.

Abstract

Process mining is essential to move beyond isolated data analysis and uncover the real operational pathways in healthcare systems. Unlike data mining, process mining leverages AI techniques to extract and model end-to-end processes, offering insights into how care actually works and should work. This approach empowers collaboration with decision makers to identify and optimize value-generating pathways. By engaging stakeholders in co-creating sustainable process designs, we can achieve better health outcomes and contribute to the sustainability of our healthcare systems, making it easier to adopt innovations identified in these collaborative processes. The keynote will showcase real-world examples of value generation and share key lessons learned that can be scaled to other fields.

Abstract

In industrial environments, the effective application of Artificial Intelligence requires more than data and algorithms, it demands the integration of people. This keynote explores how Human-in-the-Loop AI systems can enhance predictive quality in manufacturing, based on real-world implementations in factory settings. Drawing from practical experience, it examines how engineers, shift supervisors, operators, and AI systems can collaborate in hybrid networks to improve decision-making, adaptability, and trust in AI. The session reflects on the organizational, cultural, and technological challenges of embedding AI in production processes, and how collaborative design is key to unlocking its full potential.