Researcher in the Spotlight: Xiaotong Sun

Name: Xiaotong Sun
Position in the faculty: Doctoral Researcher, Marie Skłodowska-Curie Fellow in the HAIF project
Degrees: Juris Master (JM), Master of Laws (LL.M.)
Fields of interest: AI regulation and governance, human oversight, EU law, law and technology

Describe your career path. What led you to your current position?

A cross-jurisdictional education shaped the way I learned to think about law. My undergraduate education introduced me to both Chinese and US law, while my graduate studies continued this path through Chinese and EU law, exposing me to differences not only in legal rules but also in how problems are framed, which interests are made visible and what law is expected to accomplish. Moving among these traditions taught me not to regard any single perspective as self-evident and gave me a comparative instinct to look for the assumptions, institutions and social contexts underlying a particular legal answer.

That instinct also shaped the direction I chose at the beginning of my career. While focusing on environmental protection in international investment law during my master degree studies, I became increasingly aware that the rapid rise of AI was beginning to rearrange the questions that law would need to answer, and I sensed that it would become one of the defining areas of legal and regulatory development in the years ahead. When choosing my first full-time position, I therefore decided to follow this emerging field rather than enter conventional legal practice, joining a think tank in Beijing as a researcher in AI governance.

Working in AI policy and industry research shaped both my desire for greater analytical depth and my ability to approach AI as an interdisciplinary problem. The need to respond quickly to emerging issues made me want more time and methodological grounding to investigate why problems arise and why proposed solutions may or may not work in practice. Because my work ranged from examining AI technologies themselves to exploring ethical questions and the mechanisms through which risks emerge, it also taught me to follow a problem across disciplinary boundaries without losing sight of its legal and institutional stakes.

Human oversight brought these concerns together. AI makes autonomous workflow attractive, but as judgement and control move from people to AI systems, failures may become harder to detect, interrupt or remedy. I therefore chose to pursue doctoral research on how law can preserve not merely human presence, but a genuine human capacity to understand and act. This question brought me to the University of Turku in 2025, and to my current position within the HAIF project.

What projects are you currently working on?

I carry out my doctoral research within HAIF, Human-Centric Artificial Intelligence for a Sustainable Future, an EU co-funded doctoral programme at the University of Turku. At the centre of my research is one question: What makes human oversight effective?

I approach this question by diagnosing the effectiveness gap under the EU AI Act. This gap refers to the possible distance between the capacity for human oversight established by Articles 14 and 26 and the effective exercise of that capacity in practice. Assigning a human overseer is only the starting point. A person cannot oversee what they cannot perceive or understand, and cannot intervene without sufficient time, authority, technical means and organisational support. Drawing on legal analysis and insights from human factors, human-computer interaction and organisational studies, I examine the conditions that shape effective oversight. My aim is to explain why human presence may remain nominal and what would be needed to turn it into effective when AI systems create risks.

Have your interests evolved since you finished your studies?

My interests have evolved less by changing direction than by becoming more critical of the assumptions with which I began. When I first started studying human oversight, I tended to see it as a relatively straightforward safeguard. Greater human control would make AI safer, while AI operating without sufficient oversight was inherently dangerous. The deeper I looked, however, the more complex the relationship became. Placing a human "in the loop" does not necessarily give that person the practical capacity to oversee the system. Without adequate information, time, competence, authority and support, the safeguard may exist in an organisational chart but fail at the very moment it is needed.

The EU AI Act became a particularly revealing example of this shift in my thinking. At first sight, its human oversight provisions appear strict, because oversight is a mandatory requirement for high-risk AI systems. On closer reading, expressions such as "can be" and "aim to" suggest an arrangement oriented partly towards creating a capacity for human oversight rather than guaranteeing a particular result. Read alongside other articles and the standards that may further specify them, the framework turns out to be neither simply strict nor simply lenient, and its practical force depends on how these elements interact. My attention has therefore shifted from asking whether a human is present to asking what makes that presence informed, capable and effective.

If you were not a researcher, what would your profession be?

It is surprisingly difficult for me to imagine a life outside research, because research has been my work ever since I graduated. I enjoy working with words and the slow discipline of refining thought. An idea is placed on the page, questioned and gradually made more precise. Research has made me more willing to revise my first impressions and more open to the possibility that a problem may look very different from another perspective.

My neck and back occasionally remind me that intellectual compatibility and ergonomic compatibility are not quite the same thing. If I were not a researcher, I might work in public administration or public policy. Even then, however, I suspect that I would turn every policy question into a small research project, so perhaps I would not escape research after all.

What inspires you?

I am most inspired by conversations that allow a question to cross disciplinary boundaries. When I encounter a technical aspect of AI that I do not fully understand, I value being able to discuss it with colleagues in computer science. When a question concerns ethics, organisations or human behaviour, speaking with researchers in those fields can reveal dimensions that legal analysis alone might not capture.

Conferences, seminars and discussions with my supervisors inspire me for the same reason. A good conversation does more than answer the question I brought with me. It may expose an assumption I had not noticed, show why a method does not quite work or point towards a path I had not considered. For me, inspiration often emerges through this process of asking, listening, being questioned and trying again.

If you'd like, feel free to share how you've adapted to life in Turku and Finland, and the Calonia community.

Turku has been an easy city for me to grow fond of. I love nature and quiet surroundings, and the city seems to leave space for both thought and everyday life. The Finnish people I have encountered have also been remarkably kind in the small, practical moments that make up daily life. Whenever I have been uncertain about how something works, people have usually been patient and willing to help. Adapting socially has involved a subtler kind of learning. Finnish life allows a degree of personal space that I genuinely value, although it can also make it easy to remain within my own quiet orbit. A year in, I have come to read that space as an invitation rather than a distance, and to see that the move is usually mine to make. Calonia and the HAIF community have been welcoming and easy to talk to, and I would like to make that move more often.