Iran Roman
What is it about Artificial Intelligence that you enjoy so much?
What captivates me most about AI is its paradoxical nature. On one hand, it animates abstract mathematics, transforming lifeless equations into dynamic intelligence. It allows us to build synthetic minds that parse reality and act meaningfully. Yet, its limitations are equally revealing: when AI generates flawless code but struggles to explain its reasoning, or when it composes music that is technically sound but emotionally vacant, it becomes an uncanny mirror. These failures expose AI’s constraints while highlighting what makes human intelligence unique. If machines excel in structured domains we devised, why do they unravel where creativity, intuition, and leaps of insight emerge? Perhaps true intelligence is not just in what we can teach machines, but in what they may never fully grasp.
Why do you think this module is critical for students to learn? What will they gain from it?
This module provides a systematic foundation in AI’s core principles, methodologies, and historical evolution. Students engage with major paradigms—symbolic reasoning, machine learning, neural networks—and implement foundational algorithms. By studying the progression of these techniques and their role in contemporary AI, students develop both technical proficiency and critical awareness. They learn to discern how today’s advancements reinterpret classical ideas with modern data and computational power. The goal is to bridge theory, historical context, and practice, fostering a deeper, more informed approach to AI.
How would you describe your teaching style?
My teaching approach is structured around three key principles:
• Content Mastery: I demonstrate mathematical derivations, algorithmic implementation, and problem-solving workflows in lectures.
• Active Modeling: If students are expected to perform a task (coding, proofs, analytical reasoning), I show my own approach in real-time.
• Student-Centric Adaptation: I actively seek feedback to identify knowledge gaps, ensuring students have opportunities to meet outside of lectures and discuss their learning experiences.
This method prioritises clarity, replicability, and the seamless integration of theory and practice, avoiding passive, slide-based instruction.
What components of the module would be transferable to the industry students enter?
This module cultivates industry-relevant skills by grounding students in core AI techniques—symbolic systems, machine learning, neural networks. Students design, optimise, and debug real-world AI agents while learning to critically evaluate advancements within their historical context. This ability to contextualise innovation is essential in research and development, ensuring graduates can navigate an industry often driven by rapid, sometimes superficial, trends.
What advice do you have for students considering a degree in Data Science, AI, or related fields?
Approach the field with curiosity, but temper the hype. Prioritise rigorous fundamentals—mathematics, algorithms, ethical design—over chasing trends. The AI landscape rewards those who become principled engineers and scientists, capable of distinguishing durable innovations from transient buzzwords. True impact lies in mastering core theory to tackle problems that machines alone cannot solve.
What has been the highlight of your time at EECS?
Though I recently joined Queen Mary in 2024, the most rewarding aspect has been engaging with students—both in my module and through their MSc research projects. Their curiosity, technical skill, and originality continue to redefine my expectations for this field, making every interaction an opportunity to expand the boundaries of what AI can achieve.