An earlier speech by JiangXiaojuan at THU. She knows economic and knows what’s actually going on. It’s genuinely good reasoning and insightful ideas:

On Gov-Market Boundary in the Digital Era

Core Argument Digital and AI technologies have empowered governments to collect granular information and enforce accountability much better. However, the same technologies have equally empowered markets and civil society as free market to function better. Therefore, the question of whether the digital era justifies expanding government’s role has no predetermined answer, and the traditional binary of “government vs. market” should give way to a more nuance multi-stakeholder cooperative governance model.

Note: I translate and summarize based on my pov. All contribution are the speaker’s and all mistakes are mine.

The logic is:

  1. AI dramatically lowers the cost of government information and strengthens accountability

    Example: Beijing 12345 hotline handles ~80,000 citizen requests/day with AI-powered intake, semantic classification into 2,000+ categories, and intelligent routing—achieving 99% response, 96% resolution, 97% satisfaction. This real-time “good/bad review” feedback loops create powerful incentives for officials, with bottom-ranked units publicly held accountable. And granular data (e.g., “excavator index” tracking 85% of active machines nationwide) gives government precise, unfiltered views of economic activity—replacing distorted bottom-up statistical reporting.

  2. Government controls critical “landing opportunities” for AI deployment

    Networked, digital, and intelligent technologies require broad application to mature. In China, the public sector is the gate keeper of many of these deployment scenes: smart cities, emergency response, healthcare, low-altitude airspace. Example: BYD’s early EV market relied on local governments providing public-transit deployment scenarios before commercial viability.

    In other words, government’s power to allocate “scenarios” is a de facto resource-allocation lever that persists even after AI industry’s infancy.

  3. But gov services still require fiscal expenditure. Just because the state can do something does not mean it should—market solutions may be cheaper, more responsive, and more diverse. Sometimes the market do well:

    Example: government-built unmanned street libraries stocked few titles, rarely updated, and were eventually dismantled—while WeChat Read’s platform serves hundreds of millions with constantly refreshed catalogues, free or paid.

    So there are risks of expanded government role: scenario allocation can distort fair competition; technical complexity of big-data/AI projects can mask rent-seeking and rigged procurement.

  4. Which implies: embrace multi-stakeholder cooperative governance

    Example: After Red Note users flagged that Tencent’s “Yuanbao” AI could use uploaded novels for training per its ToS, public pressure forced three ToS revisions in five days—before regulators even reacted.

    A governance architecture combining a thin layer of binding “hard rules” (eg information transparency) from public authority with a rich diversity of community-level “soft rules” (platform ToS, open-source norms, industry codes).

Reference

Jiang Xiaojuan, “Government-Market Boundary in the Digital Era,” 2025 Academic Conference on Digital Economy Development & Governance, Tsinghua University, July 3, 2025. https://mp.weixin.qq.com/s/dnEnsbkfsNtFi9lJ9Mp5UQ