My research explores the two-way interaction between AI and economic systems,
studying how AI can enhance economic decision-making and
how economic principles can guide the design of intelligent, adaptive AI systems.
Learning-Based Mechanisms for AI Economies
As digital assets—such as data, privacy, APIs, and AI-generated content—gain economic significance,
markets must learn to price, trade, and utilzie them smartly and efficiently.
I design profit-aware, data-efficient, incentive-aligned mechanisms, such as
(1) AI-driven digital marketplaces
that learn buyer values over time and attain optimal revenue faster, and
(2) Incentive-aligned privacy markets
where users, instead of being forced to surrender their privacy,
can decide to whom and to what extent to share their data in exchange for economic rewards.
AI Systems with Economic and Statistical Reasoning
I study how AI systems can integrate causal and economic reasoning
to make interpretable, uncertainty-aware decisions, including
(1) Prediction-powered uncertainty quantification
that leverages foundation models to construct prediction intervals using inexpensive model-generated data,
reducing reliance on costly real-world observations, and
(2) Retrieval-augmented LLMs
that not only retrieve relevant information but also reveal how responses are generated,
identify which resources to rely on, and rank results by their reliability.
By embedding statistical inference within adaptive learning,
my work enables AI systems that quantify uncertainty and act strategically
under limited or evolving information.
Economic Alignment in Generative and Agentic AI
I explore how large language models (LLMs) and agentic AI systems
can interact through economic incentives to align behavior and value creation.
Building on insights from auction and market design,
I investigate mechanisms that link bids, rewards, and generated outcomes—ensuring
that AI-generated content or actions reflect proportional economic value.
This economic interpretation strengthens fairness, trust, and transparency
in AI-driven environments such as recommendation and advertising systems.
Ultimately, I aim to build economically intelligent, inference-driven AI systems
that not only learn from data but also reason, adapt, and coordinate through incentives.
By integrating causal inference for uncertainty quantification and
mechanism design principles for incentive alignment,
my research bridges machine learning and economics to create scalable, value-aligned AI ecosystems.