Bingchen Wang

I am an independent researcher working at the intersection of artificial intelligence and economics. My work focuses on incentive design for collaborative and decentralized machine learning; preference alignment through preference reconstruction; and bias and mitigation in AI-mediated decision systems, especially in labour market contexts.

Previously, I worked as a research assistant at the National University of Singapore, focusing on mechanism design and AI for collaborative learning systems, and earlier at the University of Hong Kong on political economy projects. I hold an MPhil in Economics from the University of Oxford, where I was fortunate to work under the guidance of Prof. David Hendry and Prof. Jennifer Castle and to contribute research within Climate Econometrics, which played an important role in shaping my research ethos and intellectual curiosity. Before that, I earned a B.A. in Mathematics–Statistics from Columbia University and a B.Sc. (Hons) in Computing Mathematics from CityU Hong Kong.

Email / CV / Scholar / LinkedIn / Github / Sites

Research Interests

I study how AI and ML systems interact with—and can be designed to better serve—human objectives, institutions, and incentives. Thematically, my interests include:

Mechanism design and incentives for AI systems / digital markets;
AI alignment and preference modeling;
Bias and mitigation in large language models (LLMs);
Economics of AI and societal impact.

Design-wise, my research is shaped by a recurring principle: reduce the fixed cost of asking a serious question. Working between machine learning and economics, I am drawn to problems where methodological ambition has been too tightly coupled to resource privilege—abundant compute, data access, institutional infrastructure, or bespoke human pipelines. I build alternatives that relax these dependencies without trivializing the task:

preference alignment without sensitive demographic data nor finetuning (P2P);
incentive design that enables small players to collaborate on model training none could afford alone (Paid with Models);
synthetic agent feedback for resource-constrained product iteration (SIM-PANEL, in development);
synthetic auditing of algorithmic bias without access to proprietary data (ResumeBench, in development).

What threads these projects together is the conviction that accessibility can be treated as a first-class design parameter rather than an afterthought.

Selected Works

Prompts to Proxies: Emulating Human Preferences via a Compact LLM Ensemble
Bingchen Wang*, Zi-Yu Khoo, Jingtan Wang
Preprint out. Under review, 2026
arXiv

Prompts to Proxies (P2P) develops a new way to align large language models with the diversity of human preferences. Instead of training or prompt-tuning one model per demographic group, P2P builds a compact ensemble of proxy agents—each defined by structured prompts that span a latent preference space. Inspired by revealed preference theory, it learns how to weight these agents to reproduce real survey data, offering a cost-efficient and theoretically grounded approach to pluralistic alignment, bridging AI alignment and social-scientific inference.

Paid with Models: Optimal Contract Design for Collaborative Machine Learning
Bingchen Wang*, Zhaoxuan Wu, Fusheng Liu, Bryan Kian Hsiang Low
AAAI 2025 (Oral Presentation)
arXiv / code / poster / slides / talk / slides (B-side)

This work studies how to incentivize collaboration in machine learning when monetary rewards are unavailable. Drawing on contract theory from economics, we model contributors’ incentives and information asymmetries to derive the optimal reward scheme theoretically and numerically in which trained models serve as payment. The results highlight how well-designed contracts can pre-empt collaboration failures and create win-wins among heterogeneous participants.

Software

Research software I build and maintain.

P2P
Pre-release research software

P2P is a modular system for reconstructing population preferences from LLM agent ensembles. P2P constructs diverse proxy agents via structured prompting, then selects a compact weighted subset to match target survey distributions—no finetuning, no demographic data, under a dollar per survey*.

Status: Pre-release (demo available upon request).
*Cost estimated using Gemini-2.0-Flash API and ATP data.

SIM-PANEL
In active development

SIM-PANEL generates synthetic panel datasets where LLM agents evaluate products under controlled experimental designs—random assignment, manual mapping, or self-selection. Designed for reproducible benchmarking, pipeline development, and simulation design in agent-based research. CLI-driven, YAML-configured, CPU-friendly.

Status: Internal research prototype.

Presentations

Let’s Talk About Trust — Paid with Models: Optimal Contract Design for Collaborative Machine Learning — End-of-Project Meeting, Trusted CollabML Lab, Singapore (Invited Talk, Mar 2025)
Slides
Fine-Tuning LLMs with Noisy Data for Political Argument Generation — Good-Data Workshop, AAAI-25, Philadelphia, USA (Presented on behalf of Svetlana Churina & Kokil Jaidka, Mar 2025)
Slides
Paid with Models: Optimal Contract Design for Collaborative Machine Learning — AAAI-25, Philadelphia, USA (Oral Presentation, Mar 2025)
Slides / Recording

Selected Awards

Phi Beta Kappa Society — Columbia University (2019)
Honor Society Fellow — School of General Studies, Columbia University (2018)
Joseph and Norma Preziosi Endowed Scholarship — Columbia University (2018)
Dean’s Scholarship — College of Science and Engineering, City University of Hong Kong (2017)
HKSAR Government Scholarship — City University of Hong Kong (2015–2016)
Chan Feng Men-ling Chan Shuk-lin ELC Scholarship — City University of Hong Kong (2014)
Full Tuition Scholarship — City University of Hong Kong (2014)
Dean’s List (all enrolled semesters) — Columbia University & City University of Hong Kong (2014–2018)

Beyond Research

Beyond research, I am deeply enamoured with art and culture, often finding inspiration in the hidden corners of cities. This passion has led me to write occasional essays on art, history, and society—including an early art-historical study of a Tang sancai figurine at the Metropolitan Museum of Art. Read selected essays in my Outside the Box series.