Luke Thorburn

My core research interests lie at the intersection: AI ∩ epistemology ∩ conflict. I try to understand how technology can be used to reduce the risks of entrenched political polarisation and improve collective decision-making, as well as to better understand the extent to which these goals are important. My background is in probability and statistics.

lukethorburn.com CV PDF @[email protected] @luke-thorburn Google Scholar

Education

  1. — Present

    PhD in Computer Science

    King's College London

    Group

    UKRI CDT in Safe & Trusted AI

    Topic

    The Use of Spatial Preference Models in Bridging Systems

    Supervisors

    Maria Polukarov, Carmine Ventre, Aviv Ovadya

    Researching how spatial preference models, such as the left-right political spectrum, can be used in recommender systems to perform bridging-based ranking.

    • Derived lower bounds on the expected error when using Euclidean preference models (and more generally, vector embeddings) to represent ordinal preferences.

    • Co-authored, together with Aviv Ovadya, a working paper on Bridging Systems.

    • Co-authored, together with Jonathan Stray and Pri Benghani, the Understanding Recommenders research blog at CHAI, UC Berkeley.

    • Served on Staff-Student Liaison Committees for the STAI CDT and Department of Informatics (2021-23).

  2. —

    Master of Science (with Distinction)

    The University of Melbourne

    Specialisation

    Statistics & Stochastic Processes

    Minor

    Pure Mathematics

    Supervisor

    Kostya Borovkov

    Thesis

    On integral representations of fractional Brownian motion

    Grade

    First Class Honours

    Metric and Hilbert Spaces

    Measure Theory

    Functional Analysis

    Mathematical Statistics

    Analysis of High-Dimensional Data

    Computational Statistics and Data Mining

    Advanced Mathematical Statistics

    Random Processes

    Advanced Probability

    Stochastic Calculus with Applications

    Mathematics of Risk

    AMSI Summer School (see below)

  3. —

    AMSI Summer School

    Australian Mathematical Sciences Institute

    Topological Data Analysis

    Probabilistic Methods & Random Graphs

  4. —

    Bachelor of Science

    The University of Melbourne

    Major

    Mathematics & Statistics

    Specialisation

    Statistics & Stochastic Processes

    Grade

    First Class Honours

    Accelerated Mathematics 1

    Accelerated Mathematics 2

    Introductory Microeconomics

    Intermediate Microeconomics

    Physics 1

    Physics 2: Advanced

    Weather and Climate Systems

    Foundations of Computing

    Foundations of Algorithms

    Database Systems

    Probability

    Probability for Inference

    Stochastic Modelling

    Statistics

    Linear Statistical Models

    Modern Applied Statistics

    Vector Calculus

    Differential Equations

    Geometry

    Complex Analysis

  5. —

    Diploma of Languages (Japanese)

    The University of Melbourne

    Japanese 5

    Japanese 6

    Japanese 7

    Japanese 8

    Japanese Through the Media

    Variation in Japanese Language

    Social Problems in Japan

    Reading Japanese Literature

  6. —

    International Baccalaureate Diploma

    St Leonard's College

    Grade

    43/45 (equivalent to notional ATAR of 99.60)

    Mathematics (HL)

    Physics (HL)

    Economics (HL)

    Literature (SL)

    Chemistry (SL)

    Japanese (SL)

    Theory of Knowledge

Teaching

  3. —

    St Leonard's College

    Student Study Assistant

    Academic tutoring and library support for students after school, on weekends and during school holidays.

  4. —

    Self-employed

    Private Academic Tutor

    Mainly in Mathematics, Physics and Economics. Students ranged from high school to university postgraduate level. Over 300 hours of tutoring experience.

Employment

  1. — Present

    Safra Center for Ethics, Harvard University

    Workstream Lead (GETTING-Plurality Research Network)

    Member of the GETTING-Plurality Research Network, working on bridging systems and broader questions around the ethics and governance of decentralised social technologies.

  2. — Present

    Resilient Youth Australia

    Data Scientist

    Data analysis, full stack web development and applied statistics research for an organisation that surveys 100,000 Australian school students annually on issues of resilience and mental health.

  3. — Present

    Future Problem Solving Program Australia

    International Evaluator & Future Scene Writer

    Evaluating written competition booklets and setting competition problems at state, national & international levels.

  4. —

    Ofcom (via research consultancy Pattrn)

    Technical Analyst

    Worked on a comprehensive review of methods for evaluating the extent to which recommender systems are implicated in the spread of illegal content.

  5. —

    Hunt Lab for Intelligence Research, University of Melbourne

    Research Associate (Machine Learning & Statistics)

    Machine learning and other research as part of a group working to improve the quality of reasoning in intelligence organisations.

    • Prototyped an argument processor—analogous to a word processor but for structured argumentation—which integrates with a large language model to partially automate common argumentative reasoning tasks.

    • Built Navigator, a prototype application that uses argument mining to augment analysts’ abilities to extract insight from large document corpora.

    • Co-facilitated the collaboration of over 100 Australian academics that led to the Roadmap to Recovery report, advising the National Cabinet on strategies for COVID-19 pandemic recovery.

        The Australian · The Age · Pursuit

    • Helped run the 2020 Hunt Challenge, a study that demonstrated that crowdsourced teams can produce analysis of comparable quality to that of professional intelligence analysts.

    • Helped conduct a major Delphi method expert elicitation on the nature of analytic rigour in intelligence work, along with its contributing factors and opportunities for improvement.

    • Conducted research in accordance with open science principles, pre-registering studies (1, 2) and sharing data.

  6. —

    Population Interventions Unit, University of Melbourne

    Data Visualisation

    Worked with epidemiologist Tony Blakely to communicate modelling of SARS-CoV-2 outbreaks, vaccination, and the consequences of government lockdowns to the public via an interactive web tool.

      ABC News · Sydney Morning Herald · Pursuit

  7. —

    CSIRO Manufacturing (Biomedical)

    Research Assistant

    Built an interactive tool to visualise and explore phase space in protein crystallisation. Originally 6-month appointment through the Undergraduate Research Opportunities Program (UROP), with several stints of follow-up work.

      Biomedical Research Victoria blog

  8. —

    The University of Melbourne

    Research Assistant

    Worked with A/Prof. Tony Wirth and Dr James Saunderson in the Department of Computing and Information Systems on algorithms for the finding of densest subgraphs.

  9. —

    Mercer

    Stochastic Modelling

    Fixed-term summer project, stochastically modelling mortalities for a new retirement income product.

Service

  1. — Present

    Goodenough College

    Treasurer, Members' Council

    Elected member representative and Chair of the Members’ Council Funding Committee for a longstanding, independent, 700-head postgraduate residential college in central London.

  2. — Present

    newDemocracy Foundation

    Research & Computational Support

    Building an accessible online tool for performing stratified random selection.

  3. —

    Future Problem Solving Program Australia

    Moderniser of Evaluator Resources

    Provided technical advice on the adoption/creation of an electronic evaluation platform. Previously, I rewrote the Evaluator Handbook.

  4. —

    Young Australians Cancer Initiative

    Director

    Inaugural co-director of an enrichment program for Australian high school students focused on cancer science.

  5. —

    iGEM UniMelb

    Videographer

    Produced three short videos to help explain and promote the work of the team from The University of Melbourne entering the International Genetically Engineered Machine competition.

  6. —

    St Leonard's College

    Peer Support Leader

  7. —

    St Leonard's College

    SCOPE Leader

    Mentored an overly reserved grade three student.

Awards & Honours

  1. 2021

    Rae and Edith Bennett Travelling Scholarship

    The University of Melbourne

  2. 2019

    Mathematics and Statistics School Scholarship

    The University of Melbourne

  3. 2018

    Maurice H. Belz Scholarship

    The University of Melbourne

  4. 2016

    Undergraduate Research Opportunities Program (UROP) Scholar

    Biomedical Research Victoria

    Leaders in Communities Award (LiCA)

    The University of Melbourne

  5. 2015

    Keith Frampton Roll of Honour (Member)

    Future Problem Solving Program Australia

    Highest Mark, Weather & Climate Systems

    The University of Melbourne

  6. 2014

    Second Place, Introductory Microeconomics Film Competition

    The University of Melbourne

  7. 2013

    Year 12 IB Dux

    St Leonard's College

    Full College, Music and House Colours

    St Leonard's College

    Old Collegian's Award for Positive Involvement

    St Leonard's College

  8. 2012

    Monash Prize for Excellence & Year 11 IB Dux

    St Leonard's College

    E. Paul Torrance Youth Award

    Future Problem Solving Program Australia

Grants & Funding

  1. —

    Computational Social Choice and Machine Learning for Ethical Decision Making

    Centre for Doctoral Training in Safe & Trusted AI

    Sponsor

    UKRI Engineering & Physical Sciences Research Council

    Value

    ≈ GBP 26,000 tuition + GBP 19,000 stipend / year

  2. —

    Momentum Building for Bridging

    Project Funding from Plurality Research Network Conference

    Sponsor

    Plurality Institute

    Value

    USD 2590.26 (amount determined via quadratic voting)

  3. —

    Transfer Learning for Argument Operations

    Artificial Intelligence for Decision Making Initiative (Phase 2)

    Sponsor

    Office of National Intelligence, Commonwealth of Australia

    Value

    AUD 50,000

  4. —

    Augmenting Corpus Analysis through Argument Mining and Networked Notes

    Artificial Intelligence for Decision Making Initiative (Phase 1)

    Sponsor

    Office of National Intelligence, Commonwealth of Australia

    Value

    AUD 20,000

Publications

  16. Analytic Rigour in Intelligence

    A Barnett, T Primoratz, R de Rozario, M Saletta, L Thorburn and T van Gelder (alphabetical)

    Hunt Lab for Intelligence Research, 2021.

Talks

  1. Open Problems in Bridging Systems

    Goodenough College

    Plurality Research Network Conference, UC Berkeley    

  2. What Does it Mean to Give Someone What They Want?

    AGI Safety Reading Group, Imperial College London

  3. Epistemic Injustice and AI

    AI Ethics Reading Group, KCL

  4. Bridging in Attention-Allocation Systems

    Attention: An Interdisciplinary Workshop, LSE  

  5. Optimizing Language Models for Argumentative Reasoning

    International Workshop on Argumentation and Machine Learning  

  6. Algorithmic Risks to Collective Decision-Making

    Goodenough-Oxford Joint Colloquium on AI and Regulation  

  7. Metaphors for Public Discourse

    Newspeak House  

  8. Argument Processing

    Sensemaking Systems Symposium, UTS

    Laboratory for Analytic Sciences, NCSU  

Symposia

  1. 2023

    Designing Tech for Social Cohesion

    Council on Technology and Social Cohesion

    attendeeposter presenter

    Plurality Research Network Conference

    University of California, Berkeley

    speaker

  2. 2022

    Safe and Trustworthy AI Workshop (STAIws)

    UKRI CDT in Safe and Trusted AI, Imperial College London

    participantposter presenter

    Attention: An Interdisciplinary Workshop

    Department of Anthropology, London School of Economics

    speaker

    International Workshop on Argumentation and Machine Learning (ArgML)

    speaker

    ACM Conference on Economics and Computation (EC)

    reviewer

    The Race to Regulate AI: Global Comparative Perspectives

    Bonavero Institute of Human Rights, University of Oxford

    attendee

    Oxford-Goodenough Joint Colloquium on AI and Regulation

    Jesus College, University of Oxford and Goodenough College

    participant

  3. 2021

    International Conference on Algorithmic Decision Theory (ADT)

    attendee

    Conference for Truth and Trust Online (TTO)

    attendee

  4. 2020

    International Conference on Computational Models of Argument (COMMA)

    attendee

  5. 2019

    Effective Altrusim Global (EAG)

    attendee

    INFORMS Applied Probability Society Conference (INFORMS-APS)

    attendee

  6. 2017

    Summer Institute on Topics in the History of Economics

    Center for the History of Political Economy, Duke University

    participant

  7. 2016

    Conference on Neural Information Processing Systems (NeurIPS)

    attendee

Memberships

  1. — Present

    Trust & Safety Professional Association

  2. — Present

    Information Professionals Association

  3. — Present

    Newspeak House

  4. — Present

    Statistical Society of Australia

  5. — Present

    Australian Mathematical Society