Joshua C. Yang

Hello 👋🏽 I am Josh, a postdoctoral researcher at MIT 🇺🇸 in Michiel Bakker’s group and the Laboratory for Information and Decision Systems (LIDS). I work on AI safety, studying AI agent coordination and collective intelligence: how AI agents can work together and with people to solve problems and make collective decisions. I also focus on public input in AI alignment and democratic oversight, exploring how people can help shape AI’s goals and hold increasingly capable systems accountable.

My PhD at ETH Zurich 🇨🇭 began with AI–Human collaboration, developing voting algorithms and tools to support human collective decision-making in digital democracy. This work, together with social choice theory, later informed my study of how AI agents coordinate. I was supervised by Prof. Dirk Helbing in the Computational Social Science Lab.

With a background in Computer Science, I draw on Complex Systems, Computational Social Choice, and Human-AI Interaction, including LLM agent-based simulations. I am also an affiliated researcher at the Collective Intelligence Project.

I continue to put my digital democracy research into practice by advising governments and organisations on digital participation. I also developed and launched Murmi, a real-time, audio-based tool for iterative collective sensemaking that helps groups surface shared priorities, tensions, and overlooked perspectives as conversations unfold. In Switzerland, I have worked with the city of Aarau on the StadtIdee Participatory Budgeting programme and with Kultur Komitee Winterthur on its annual citizen assembly for arts and cultural funding. I am also the process designer for vTaiwan, part of Taiwan’s 🇹🇼 civic tech g0v community, where we explore AI and digital tools for public deliberation and consensus-building.

During my Computer Science (NLP) master’s at the University of Melbourne 🇦🇺, I conducted fieldwork in the Northern Territory, focusing on Speech Recognition for Indigenous languages. My research explored a human-in-the-loop transcription pipeline that combined respeaking and word-spotting to improve recognition for endangered Indigenous languages.

I call Australia, Taiwan and Switzerland home.

Hacking Democracy / ETH Zurich, September 7, 2026

Building Digital Democracy in Practice

Explore how people make collective decisions, with research, fieldwork and working tools from Switzerland and beyond.

Explore the presentation →

news

Sep 25, 2026	Our Belief Engine paper has just been accepted to NeurIPS 2026!
Aug 26, 2026	I’m organising the First Workshop on Sovereign AI for Collaborative and Pluralistic AI Ecosystems at ACM HCOMP × Collective Intelligence 2026 near Washington, D.C., on September 27. The goal is to bring the community together around more collaborative and pluralistic AI ecosystems. We’re currently looking for paper submissions, from early ideas and prototypes to full papers and real-world case studies.
Aug 12, 2026	Our TA-SWISS project team hosted four workshops in Zurich and Lausanne across three languages, bringing together 100 people to discuss Swiss AI futures using the AI deliberation tool Murmi.
Jul 23, 2026	This autumn, I’ll be joining Michiel Bakker’s lab at MIT as a postdoc! 🍂
May 10, 2026	I’m hosting the Swiss AI Futures Citizen Assembly in Zurich and Lausanne! We’re testing out how AI can help people deliberate both online and in-person for national AI policy. The assembly will run in June, we’ll be sharing the results with the Swiss Parliament. Check it out: swissaifutures.org

selected publications

NeurIPS
Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation

Joshua C Yang, Maurice Flechtner, Damian Dailisan, and Michiel A Bakker

In Advances in Neural Information Processing Systems (NeurIPS) , 2026

Accepted for publication

Page Abs arXiv Bib

We introduce Belief Engine, an auditable belief-update layer for multi-agent LLM deliberation. The system extracts arguments into structured memory and updates stance with configurable log-odds dynamics, making belief change more inspectable and useful for studying social simulation, disagreement, and deliberation.
@inproceedings{yang2026belief, title = {Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation}, author = {Yang, Joshua C and Flechtner, Maurice and Dailisan, Damian and Bakker, Michiel A}, booktitle = {Advances in Neural Information Processing Systems (NeurIPS)}, year = {2026}, note = {Accepted for publication}, url = {https://arxiv.org/abs/2605.15343}, }
AIES
LLM Voting: Human Choices and AI Collective Decision Making

Joshua C Yang, Marcin Korecki, Damian Dailisan, Carina I Hausladen, and Dirk Helbing

In 7th AAAI Conference on AI, Ethics, and Society , 2024

Page Abs Bib

This paper explores how Large Language Models (LLMs) can participate in and support human collective decision making. Through experiments comparing human and LLM voting behaviors, the work investigates alignment, bias, and interpretability in AI-assisted democratic processes.
@inproceedings{yang2024llm, title = {LLM Voting: Human Choices and AI Collective Decision Making}, author = {Yang, Joshua C and Korecki, Marcin and Dailisan, Damian and Hausladen, Carina I and Helbing, Dirk}, booktitle = {7th AAAI Conference on AI, Ethics, and Society}, year = {2024}, }
CHI EA
Komitee Equal Shares: A Hybrid Algorithm for Priceable Participatory Funding

Joshua C. Yang, Noemi Scheurer, and Dirk Helbing

In Proceedings of the Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems , 2026

Page Abs Bib

Public funding processes demand fairness, learning, and outcomes that participants can understand. We introduce Komitee Equal Shares, a priceable virtual-budget allocation framework that integrates two signals: in voter mode, participants cast point votes; in evaluator mode, small groups assess proposals against collectively defined impact fields. The framework extends the Method of Equal Shares by translating both signals into virtual spending power and producing explicit “voting receipts.” We deployed the framework in the 2025 Kultur Komitee in Winterthur, Switzerland (N = 38), allocating CHF 380,000. Our results show that this hybrid approach generates a more diverse portfolio than simple aggregation and that participants value the fair process over individual control. The framework offers a model where citizens act as both voters and evaluators within a single, proportional, and explainable allocation.
@inproceedings{10.1145/3772363.3798958, author = {Yang, Joshua C. and Scheurer, Noemi and Helbing, Dirk}, title = {Komitee Equal Shares: A Hybrid Algorithm for Priceable Participatory Funding}, year = {2026}, isbn = {9798400722813}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/3772363.3798958}, doi = {10.1145/3772363.3798958}, booktitle = {Proceedings of the Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems}, articleno = {422}, numpages = {7}, keywords = {Participatory Budgeting, Algorithmic Accountability, Method of Equal Shares, Deliberation, Digital Democracy}, series = {CHI EA '26} }

ACM

Designing Digital Voting Systems for Citizens: Achieving Fairness and Legitimacy in Participatory Budgeting

Joshua C Yang, Carina I Hausladen, Dominik Peters, Evangelos Pournaras, Regula Hänggli Fricker, and Dirk Helbing

Digital Government: Research and Practice, 2024

Page Abs Bib

@article{yang2024designing,
  title = {Designing Digital Voting Systems for Citizens: Achieving Fairness and Legitimacy in Participatory Budgeting},
  author = {Yang, Joshua C and Hausladen, Carina I and Peters, Dominik and Pournaras, Evangelos and H{\"a}nggli Fricker, Regula and Helbing, Dirk},
  journal = {Digital Government: Research and Practice},
  volume = {5},
  number = {3},
  pages = {1--30},
  year = {2024},
  publisher = {ACM},
  url = {https://dl.acm.org/doi/10.1145/3665332},
}

ACM
Bridging Voting and Deliberation with Algorithms: Field Insights from vTaiwan and Kultur Komitee

Joshua C Yang, and Fynn Bachmann

In ACM Conference on Fairness, Accountability, and Transparency (FAccT) , 2025

Page Abs Bib

Based on field experiments in Taiwan and Switzerland, this paper presents hybrid algorithmic designs that integrate deliberation and voting for digital democracy.
@inproceedings{yang2025bridging, title = {Bridging Voting and Deliberation with Algorithms: Field Insights from vTaiwan and Kultur Komitee}, author = {Yang, Joshua C and Bachmann, Fynn}, booktitle = {ACM Conference on Fairness, Accountability, and Transparency (FAccT)}, year = {2025}, url = {https://arxiv.org/abs/2502.05017}, }

projects

Murmi

A real-time, audio-based tool for iterative collective sensemaking that helps groups turn live conversations into shared, reviewable decisions.

Equal Shares Voting Receipts

A public voting-receipt interface for Kultur Komitee Winterthur 2026, showing how Equal Shares turns votes, costs, and equal budgets into explainable project funding.

Swiss AI Futures

I am running two citizen assemblies on AI in Switzerland, using AI to connect broad online public input with in-person deliberation for policy.

teaching

Spring 2026

Co-lecturer

Complexity and Global Systems Science (860-0022-00L)

ETH Zurich, Zurich, Switzerland

Co-lecturer for a seminar on complex dynamical systems, networked risks, systemic instability, and global-scale socio-technical problems.

2024-2025

Co-lecturer

Computational Social Science (851-0585-41L)

ETH Zurich, Zurich, Switzerland

Co-lecturer for an interdisciplinary seminar connecting computational modelling, simulation, empirical and data-driven social research, and global systems science.