EconCS lab opening

As yesterday’s blog mentioned the opening premiere, here’s the fresh recap. The first section features ceremonial talks, and the second is a more lively discussion of mechanism design, economics, and artificial intelligence.

Section 1: Ceremonial Talks

The first section of the event was filled with what we can affectionately describe as “ceremonial talks.” These talks are often marked by grand visions and mutual compliments, leaving the audience yearning for more substance. However, there was a beacon of hope in the form of my advisor who is also the director of the lab, managed to infuse genuine content into his speech.

Section 2: Insights from Mechanism Design, Economics, and AI

This is where the meat of the event was. Three scholars presented inspirational talks:

• Talk 1: Jianwei Huang - Data Privacy and Correlation

  Jianwei Huang, representing CHUK-SZ, delivered a fascinating talk on mechanism design related to data privacy in correlated social network. He delved into questions about aligning individuals’ willingness to contribute data and the incentives required from data collectors.

  Huang presented a mathematical model that included two crucial layers: the data layer (modelled as a weighted undirected graph) and the social layer (weighted directed graph), with asymmetric characteristics. His insights included the role of conditional variance in individual privacy loss and the trade-off between adding noise and self-interest.

• Talk 2: Xiaole Wu - Endogenous Merger Decisions Among Competitors: Impact of Limiter Capacity and Loyal Segment

  Professor Xiaole Wu from FDU provided insights on merger decisions restricted with productivity capacity and loyal consumer. It’s quite a classical economic-game-theory work. About merger, real-world examples ranged from microchips, to e-commerce, like to the union of Didi and Uber China. Wu used a notion of “strong Nash” equilibrium (immune to collusions) to analyze merger decisions, emphasizing the importance of capacity constraints. Her analysis illustrated the implications for competition and social welfare, revealing which merger methods could reduce competition.

  But they’re limited to simplified market model involving 3 firms asymmetrically or multiple firm symmetrically. Huh, economists.

• Talk 3: Jirong Wen, Renmin University - Concepts: Intelligent Agent vs. LLM

  Jirong Wen, representing Renmin University, explored the world of intelligent agents driven by Large Language Models (LLMs). In his talk, he introduced the notion of a “reasoning chain” and the more intricate concept of “multi-track reasoning chains.” These chains form the foundation for understanding the evolution of intelligent agents and their decision-making processes.

  I was in the restroom in the third talk because personally, still not a big fan of LLM. But my friend really liked it so I was like ok sure (and pretending to be appreciating the third talk as well). But the event in general was fruitful and happy. Liked it.