Cloud & AI
What is the “black box” called AI?
Andreea Giurgiu - AI Master Student @ Vrije Universiteit Amsterdam
Room 32
12th November, 13:00-13:30
Artificial Intelligence often feels like a mysterious black box — powerful, unpredictable, and sometimes a little too smart for comfort. We ask: Is my phone listening to me? Does AI really “think”? But the truth is less magical and far more fascinating.
After 4 years of studying AI, I can finally take the curtain of AI to reveal what’s actually happening inside these systems. Drawing from four years of research and hands-on work — from probabilistic game-playing bots to behavioral modeling and conversational assistants — I explores what AI really is: a system of probabilities trained on patterns, not an all-knowing mind.
Then I connect the session with the science behind AI to the ethics that must guide it. Through the lens of European regulation: the GDPR and the upcoming AI Act
Then discusses why fairness, transparency, and data integrity aren’t barriers to innovation, but the foundation for trustworthy technology. Also examines bias as a product of historical data and emphasizes how both developers and users share responsibility in understanding and using AI wisely.
The talk closes with a practical and hopeful perspective on the future of AI — arguing that humans won’t be replaced by algorithms, but by those who know how to use them effectively and ethically.
Andreea Giurgiu
Vrije Universiteit Amsterdam
Andreea is an AI researcher and community builder originally from Cluj, Romania, now based in Amsterdam. After moving to the Netherlands four years ago, she completed her bachelor’s and is currently pursuing a master’s in Artificial Intelligence at Vrije Universiteit Amsterdam.
Her passion for AI began after winning a global AI hackathon, which inspired her to organize five international hackathons and to found TeensInAI Romania — a national initiative that introduces teenagers to AI through hackathons, mentoring, and workshops.
Andreea’s research focuses on hybrid intelligence: building cognitive agents that collaborate with humans and modeling behavior inspired by animal groups. She has also explored LLM-based labeling for unstructured data and conversational systems using the Google Dialogflow API.
Alongside her studies, Andreea has served as a teaching assistant and course coordinator, worked in corporate operations for nearly two years, and developed an ontology, as a research assistant, now used in a master’s course on hybrid intelligence research methods.
