Alba (Ruiran) Su
DPhil Researcher & AI Policy Fellow
University of Oxford · Jardine Scholar
Incoming IAPS AI Policy Fellow · Summer 2026
(conf. + journal)
acceptance rate
chart-claim pairs
research grants
About Me
I am a DPhil student in Engineering Science at the University of Oxford, supervised by Prof. Janet B. Pierrehumbert, and an incoming IAPS AI Policy Fellow (Summer 2026, top 0.4% of 8,200+ applicants). As a Jardine Scholar (1.5% acceptance rate), my work bridges technical AI research and international AI governance.
Research Focus
My research addresses catastrophic risks from advanced AI, with a focus on evaluation interoperability as a mechanism for preventing race-to-the-bottom dynamics in dangerous-capability deployment. I develop verifiable safety frameworks bridging US (NIST AI 600-1), UK (AISI Frontier AI Evals), and Chinese (TC260 / GB/T series) standards — using first-hand analysis of primary Chinese standards texts to mitigate strategic miscalibration between great powers.
On the technical side, I build causal multimodal AI frameworks integrating satellite imagery, scientific charts, and text for environmental accountability, and develop the CLIMATEVIZ benchmark for statistical reasoning in high-stakes scientific domains.
Research Interests
- 🛡️ AI Safety & Frontier AI Evaluation (NIST, AISI, TC260)
- 🌐 International AI Governance & Evaluation Interoperability
- 🔗 Causal Inference & Explainability (DAGs, ICP, do-calculus)
- 🌍 Climate Misinformation Detection & Scientific Fact-Checking
- 🖼️ Multimodal AI & Vision-Language Models
- 🤖 Graph Neural Networks for Social Network Analysis
- 🔬 Argument Mining & NLP for Scientific Discovery
Recent News
Publications
Conference & Journal Papers
Book Chapters
Research Projects
Evaluation Interoperability for Frontier AI Red Lines New
As an IAPS AI Policy Fellow, I am developing a framework for evaluation interoperability between the US (NIST AI 600-1), UK (AISI Frontier AI Evals), and Chinese (TC260) dangerous-capability standards. The goal is to identify shared definitions of safety thresholds that can prevent unilateral deployment of unsafe frontier AI systems — and design zero-trust attestation architectures for where shared definitions are not yet feasible.
Methods: Structured expert interviews, standards mapping, policy-facing disagreement logs, zero-trust auditing architectures.
CLIMATEVIZ: Benchmark for Scientific Chart Reasoning
CLIMATEVIZ is a large-scale benchmark (49,862 chart-claim pairs; 2,896 expert-curated charts) sourced from NOAA, UK Met Office, and Copernicus, designed to evaluate statistical reasoning and fact-checking on scientific charts. It bridges visual and textual data to benchmark vision-language models for high-stakes scientific domains.
Key features: Multi-source authoritative climate data, expert-curated annotations, multimodal LLM evaluation, RAG applications.
Causal Graph Discovery for Climate Claims
Creating graph-based causal discovery frameworks using invariant causal prediction (ICP) to detect spurious correlations and validate scientific claims across distributional shifts. This methodology is directly transferable to AI evaluation claim verification across jurisdictions.
Methods: NOTEARS, PC, FCI, GES algorithms; do-calculus; mediation analysis; counterfactual reasoning.
Climate Discourse Analysis via Graph Neural Networks
GNN-based approaches to understand climate discourse and misinformation on social media. Graph attention networks model 1,397 scientific entities, achieving 79% accuracy in detecting climate misinformation and demonstrating causal reasoning for information diffusion.
Teaching
University of Oxford (2022 – Present)
Graduate Teaching Assistant, Department of Engineering Science
- Computational & Corpus Linguistics: Supervised tutorials on statistical analysis, graphical models, and causal reasoning in linguistic data.
- Cybersecurity: Led tutorials on threat analysis and vulnerability assessment using causal frameworks.
- NLP for Mental Health: Taught statistical methods for observational data analysis, emphasising confounding control.
- AI for Education: Supervised projects on intervention design and causal evaluation of AI-based tools.
Professional Experience
IAPS – AI Policy Fellow, International Strategy Track Incoming
Summer 2026 | London / Washington, D.C.
- Developing an evaluation interoperability framework for frontier AI "red lines" between the US, UK, and China.
- Mapping TC260-related security requirements to NIST AI 600-1 and AISI Frontier AI Evals.
- Conducting structured expert interviews with evaluation practitioners and China standards analysts.
Climatic Change (Springer) – Editorial Reviewer
2026 | Special Collection: NLP and AI as Climate Solutions
- Invited by editorial team to review submissions on opportunities and challenges of NLP and AI as climate solutions.
ACL 2025 – ClimateNLP Workshop Co-Organizer
2025 | Vienna, Austria
- Co-organized alongside Prof. Christopher Manning (Stanford) and leading NLP scientists.
- Coordinated peer review; facilitated interdisciplinary discussions on responsible AI for high-stakes applications.
China Telecom Research Institute – AI Technology Researcher
2021 – 2025 | Beijing, China
- Developed causal models for complex systems using structural equation modeling and knowledge graph reasoning.
- Applied graphical causal inference and RAG for automated decision support; presented at Linux Foundation conferences.
- Published multiple IEEE conference papers on intent-based networking and 6G systems.
Tsinghua University – Assistant Researcher, Smart City Program
2020 – 2021 | Beijing, China
- Investigated causal mechanisms of regional inequality using observational data.
- Applied causal mediation analysis to identify policy-relevant pathways.
Awards & Honors
Top 0.4% of 8,200+ applicants. Focus: US-UK-China evaluation interoperability & dangerous-capability red lines.
Premier full scholarship for exceptional academic merit and leadership potential. Acceptance rate: 1.5%.
Two-time recipient for causal verification frameworks and statistical reasoning in high-stakes AI.
Awarded for research on interpretability and safety metrics of large-scale foundation models.
Recognised for cross-disciplinary leadership in responsible AI integration.
Competitive grant supporting Sino-British academic exchange and international research collaboration.
Highest honor for students in the top 1% of the Computer Science department.
Beyond Research
🎵 Music & Composition
I am an experienced multi-instrumentalist with proficiency in piano, keyboards, guitar, ukulele, Irish flute, harmonica, handpan, and dizi. I compose original music blending folk traditions with modern genres including neo-classical, indie pop, tropical house, and EDM, and actively publish on platforms like NetEase.
🎼 Listen to My Music
🎧 Winterbird🎨 Creative Arts & Writing
I am a member of the China Writers Association, writing fiction, poetry, and environmental essays. I'm also a member of Procreate Artists, exploring digital art alongside traditional painting with acrylic and watercolor mediums — often with nature-inspired themes.
