This is JiangXiaojuan’s March 2026 speech at Southwest University of Political Science & Law, Chongqing. She is a professor at Chinese Academy of Social Sciences (also where she got her economic PhD), and former deputy secretary-general of the Chinese state council.

AI for Good: What Is Good, How to Achieve It, Who Should Act?

Core argument International consensus on AI ethics principles (safety, fairness, transparency, etc.) is strong but vague on implementation. Jiang proposes a social-science framework defining three dimensions of “good”—rationality, utility, and social consensus—each with concrete metrics, responsible actors, and enforcement mechanisms.

Three dimension of ‘good’ that AI brings:

  • 合理, I find it best translate as rationality — efficient allocation, welfare gains, and fair distribution. These part of AI’s produce are measurable via standard economic indicators: total factor productivity, input-output ratios, income growth, innovation investment. By these metrics, AI clearly qualifies as “good.”

    Though on equity AI is currently not good: wealth concentrates among a few innovators while job displacement hits middle-and-low income workers hardest.

    Her prescription is firms must pursue inclusive innovation; government must balance labor-displacing AI deployment with new job creation and strengthen long-term social safety nets.

  • 用益, means consumer surplus beyond GDP: AI delivers massive un-priced benefits: self-service booking, free digital media, AI-assisted creative tools. These replace GDP-counted services, so GDP understates AI’s true welfare contribution.

    This is actually an equalizing effect: Low-income and low-literacy users gain proportionally more from free AI tools (e.g., image/video content for those with weak reading ability; free access to expensive cultural products). Measurable via contingent valuation / willingness-to-pay surveys.

    On the backside: addictive design (gaming, filter bubbles) causes long-term psychological harm. Producers bear product-liability-style responsibility; government and society must co-regulate.

  • 合意: social consensus on legitimacy for irreversible scientific frontiers

    This is the most novel contribution. Jiang distinguishes traditional science (“discovering natural laws”) from AI-era science that actively creates new realities—altering human physiology, cognition, reproduction, and even consciousness. Because such innovations are potentially irreversible and affect all of humanity, they require explicit public legitimation through transparent collective deliberation, not unilateral decisions by scientists or firms.

    Scientists must disclose all possible consequences (not just benefits); society must be given space for informed debate to reach a “greatest common denominator” consensus. The technical logic of what can be done must never override what the public will accept.

Governance Mechanisms for Good

  1. Market self-discipline: AI requires massive adoption to succeed, so social disapproval of “not-good” behavior is a powerful and fast-acting corrective. Reputational incentives are stronger in the AI era than ever (e.g., OpenAI’s swift pivot on user-data training after backlash).
  2. Distributed governance: AI deployment is inherently scene-based and cluster-oriented. Each application ecosystem (smart city, health platform, e-commerce marketplace) forms its own community with tailored rules. These communities simultaneously function as self-regulating governance units.
  3. Public authority (essential but bounded): Government’s role is to set a “negative list” of prohibited conduct (privacy violations, disinformation, terrorism), mandate transparency of user agreements, issue guidance and best-practice cases, and signal norms through informal tools (criticism, regulatory interviews) while formal law catches up.

Reference:

Jiang Xiaojuan, “AI for Good: What, How, Who,” 2026 Academic Conference on Digital Economy Development & Governance, SWUPL Chongqing, March 7, 2026. Published by New Economist / Tsinghua Service Economy & Digital Governance Institute https://mp.weixin.qq.com/s/JffV9i3zvSAbk1vA5dmJWA