Safe and Ethical AI (SEA) Platform Network · Linking Artificial Intelligence Principles (LAIP)

· Algorithmic fairness

Ethics by design (EBD): ensure that algorithm is reasonable, and date is accurate, up to date, complete, relevant, unbiased and representative, and take technical measures to identify, solve and eliminate bias Formulate guidelines and principles on solving bias and discrimination, potential mechanisms include algorithmic transparency, quality review, impact assessment, algorithmic audit, supervision and review, ethical board, etc.

Principle: "ARCC": An Ethical Framework for Artificial Intelligence, Sep 18, 2018

Published by Tencent Research Institute

Related Principles

· Article 6: Transparent and explainable.

Continuously improve the transparency of artificial intelligence systems. Regarding system decision making processes, data structures, and the intent of system developers and technological implementers: be capable of accurate description, monitoring, and reproduction; and realize explainability, predictability, traceability, and verifiability for algorithmic logic, system decisions, and action outcomes.

Published by Artificial Intelligence Industry Alliance (AIIA), China in Joint Pledge on Artificial Intelligence Industry Self-Discipline (Draft for Comment), May 31, 2019

Chapter 3. The Norms of Research and Development

　　10. Strengthen the awareness of self discipline. Strengthen self discipline in activities related to AI research and development, actively integrate AI ethics into every phase of technology research and development, consciously carry out self censorship, strengthen self management, and do not engage in AI research and development that violates ethics and morality. 　　11. Improve data quality. In the phases of data collection, storage, use, processing, transmission, provision, disclosure, etc., strictly abide by data related laws, standards and norms. Improve the completeness, timeliness, consistency, normativeness and accuracy of data. 　　12. Enhance safety, security and transparency. In the phases of algorithm design, implementation, and application, etc., improve transparency, interpretability, understandability, reliability, and controllability, enhance the resilience, adaptability, and the ability of anti interference of AI systems, and gradually realize verifiable, auditable, supervisable, traceable, predictable and trustworthy AI. 　　13. Avoid bias and discrimination. During the process of data collection and algorithm development, strengthen ethics review, fully consider the diversity of demands, avoid potential data and algorithmic bias, and strive to achieve inclusivity, fairness and non discrimination of AI systems.

Published by National Governance Committee for the New Generation Artificial Intelligence, China in Ethical Norms for the New Generation Artificial Intelligence, Sep 25, 2021

Fairness

Use of AI will include safeguards to manage data bias or data quality risks The best use of AI will depend on data quality and relevant data. It will also rely on careful data management to ensure potential data biases are identified and appropriately managed. AI solutions that rely on sub optimal quality data may result in sub optimal project outcomes and recommendations. Algorithms that contain systemic and repeatable errors may lead to prejudiced decisions or outcomes. Projects should clearly demonstrate: a data model that is designed with a focus on diversity and inclusion use of a dataset that is representative for the problem to be solved regular monitoring of data models and outputs.

Published by Government of New South Welsh, Australia in Mandatory Ethical Principles for the use of AI, 2024

10. Responsibility, accountability and transparency

a. Build trust by ensuring that designers and operators are responsible and accountable for their systems, applications and algorithms, and to ensure that such systems, applications and algorithms operate in a transparent and fair manner. b. To make available externally visible and impartial avenues of redress for adverse individual or societal effects of an algorithmic decision system, and to designate a role to a person or office who is responsible for the timely remedy of such issues. c. Incorporate downstream measures and processes for users or stakeholders to verify how and when AI technology is being applied. d. To keep detailed records of design processes and decision making.

Published by Personal Data Protection Commission (PDPC), Singapore in A compilation of existing AI ethical principles (Annex A), Jan 21, 2020

Ensure fairness

We are fully determined to combat all types of reducible bias in data collection, derivation, and analysis. Our teams are trained to identify and challenge biases in our own decision making and in the data we use to train and test our models. All data sets are evaluated for fairness, possible inclusion of sensitive data and implicitly discriminatory collection models. We execute statistical tests to look for imbalance and skewed datasets and include methods to augment datasets to combat these statistical biases. We pressure test our decisions by performing peer review of model design, execution, and outcomes; this includes peer review of model training and performance metrics. Before a model is graduated from one development stage to the next, a review is conducted with required acceptance criteria. This review includes in sample and out of sample testing to mitigate the risk of model overfitting to the training data, and biased outcomes in production. We subscribe to the principles laid out in the Department of Defense’s AI ethical principles: that AI technologies should be responsible, equitable, traceable, reliable, and governable.

Published by Rebelliondefense in AI Ethical Principles, January 2023