大数据

• •    

可信AI治理框架探索与实践

夏正勋,唐剑飞,罗圣美,张燕   

  1. 星环信息科技(上海)股份有限公司,上海 200233
  • 作者简介:夏正勋(1979- ),男,星环信息科技(上海)股份有限公司高级研究员,主要研究方向为大数据、数据库、人工智能、流媒体处理技术等。 唐剑飞(1986- ),男,星环信息科技(上海)股份有限公司大数据技术标准研究员,主要研究方向为大数据、数据库、图计算等。 罗圣美(1971- ),男,博士,星环信息科技(上海)股份有限公司大数据研究院院长,主要研究方向为大数据、并行计算、云存储、人工智能等。张燕(1985- ),女,星环信息科技(上海)股份有限公司大数据技术研究员,主要研究方向为大数据、人工智能等。

Exploration of trusted AI governance framework

XIA Zhengxun, TANG Jianfei, LUO shengmei, ZHANG Yan   

  1. Transwarp Information Technology (Shanghai) Co., Ltd., shanghai 200233, China

摘要: 人工智能(Artificial Intelligence,AI)进一步提升了信息系统的自动化程度,但在其规模应用过程中也暴露了一些新问题,比如数据安全、隐私保护、公平伦理等。为了解决这些问题,推动AI由可用系统向可信系统转变,本文提出了T-DACM可信AI治理框架,从数据、算法、计算、管理四个层级入手,提升AI的可信性,设计了不同组件针对性解决数据安全、模型安全、隐私保护、模型黑盒、公平无偏、追溯定责等具体问题。T-DACM框架通过各层之间的分工协作,将各模块的可信方法结合起来,提供多样化的可信AI解决方案,模块化设计也方便未来灵活引入新的可信措施,其为业界提供了一个完善可行的可信AI治理框架示范,对于后续基于可信AI治理框架的产品研发提供一定的参考。

关键词: 可信AI, 治理框架, AI公平伦理, AI可解释, AI监管

Abstract: Artificial Intelligence (AI) has further improved the automation of information systems, however, some issues have been exposed during its large-scale application, such as data security, privacy protection, and Fair ethics. In order to solve these issues and promote the transition of AI from available systems to trusted systems, this paper proposes the T-DACM trusted AI governance framework to improve the credibility of AI from the four levels of data, algorithm, calculation and management. Different components are designed to solve specific issues such as data security, model security, privacy protection, model black box, fairness, accountability and traceability.The T-DACM framework combines the trusted methods of each module through the division of labor and cooperation among various layers to provide diversified trusted AI solutions.The modular design also facilitates the flexible introduction of new trusted measures in the future. It provides a complete and feasible demonstration of trusted AI governance framework for trade unionandprovides some reference for subsequent product development based on the trusted AI governance framework.

Key words: trustedAI, governanceframework, AI ethics and fairness, AI interpretability, AI supervision

No Suggested Reading articles found!