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

5. Fairness

a. Ensure that algorithmic decisions do not create discriminatory or unjust impacts across different demographic lines (e.g. race, sex, etc.). b. To develop and include monitoring and accounting mechanisms to avoid unintentional discrimination when implementing decision making systems. c. To consult a diversity of voices and demographics when developing systems, applications and algorithms.

Principle: A compilation of existing AI ethical principles (Annex A), Jan 21, 2020

Published by Personal Data Protection Commission (PDPC), Singapore

Related Principles

2. Fairness and Equity

Deployers should have safeguards in place to ensure that algorithmic decisions do not further exacerbate or amplify existing discriminatory or unjust impacts across different demographics and the design, development, and deployment of AI systems should not result in unfair biasness or discrimination. An example of such safeguards would include human interventions and checks on the algorithms and its outputs. Deployers of AI systems should conduct regular testing of such systems to confirm if there is bias and where bias is confirmed, make the necessary adjustments to rectify imbalances to ensure equity. With the rapid developments in the AI space, AI systems are increasingly used to aid decision making. For example, AI systems are currently used to screen resumes in job application processes, predict the credit worthiness of consumers and provide agronomic advice to farmers. If not properly managed, an AI system’s outputs used to make decisions with significant impact on individuals could perpetuate existing discriminatory or unjust impacts to specific demographics. To mitigate discrimination, it is important that the design, development, and deployment of AI systems align with fairness and equity principles. In addition, the datasets used to train the AI systems should be diverse and representative. Appropriate measures should be taken to mitigate potential biases during data collection and pre processing, training, and inference. For example, thetraining and test dataset for an AI system used in the education sector should be adequately representative of the student population by including students of different genders and ethnicities.

Published by ASEAN in ASEAN Guide on AI Governance and Ethics, 2021

Fairness

Ensure that algorithmic decisions do not create discriminatory or unjust impacts when comparing across different demographics (e.g. race, sex, etc).

Published by Fairness, Accountability, and Transparency in Machine Learning (FAT/ML) in Principles for Accountable Algorithms, Jul 22, 2016 (unconfirmed)

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

Principle 1 – Fairness

The fairness principle requires taking necessary actions to eliminate bias, discriminationor stigmatization of individuals, communities, or groups in the design, data, development, deployment and use of AI systems. Bias may occur due to data, representation or algorithms and could lead to discrimination against the historically disadvantaged groups. When designing, selecting, and developing AI systems, it is essential to ensure just, fair, non biased, non discriminatory and objective standards that are inclusive, diverse, and representative of all or targeted segments of society. The functionality of an AI system should not be limited to a specific group based on gender, race, religion, disability, age, or sexual orientation. In addition, the potential risks, overall benefits, and purpose of utilizing sensitive personal data should be well motivated and defined or articulated by the AI System Owner. To ensure consistent AI systems that are based on fairness and inclusiveness, AI systems should be trained on data that are cleansed from bias and is representative of affected minority groups. Al algorithms should be built and developed in a manner that makes their composition free from bias and correlation fallacy.

Published by SDAIA in AI Ethics Principles, Sept 14, 2022

· Plan and Design:

The fairness principle requires taking necessary actions to eliminate bias, discrimination or stigmatization of individuals, communities, or groups in the design, data, development, deployment and use of AI systems. Bias may occur due to data, representation or algorithms and could lead to discrimination against the historically disadvantaged groups. When designing, selecting, and developing AI systems, it is essential to ensure just, fair,non biased, non discriminatory and objective standards that are inclusive, diverse, andrepresentative of all or targeted segments of society. The functionality of an AI system shouldnot be limited to a specific group based on gender, race, religion, disability, age, or sexualorientation. In addition, the potential risks, overall benefits, and purpose of utilizing sensitivepersonal data should be well motivated and defined or articulated by the AI System Owner. To ensure consistent AI systems that are based on fairness and inclusiveness, AI systems shouldbe trained on data that are cleansed from bias and is representative of affected minority groups.Al algorithms should be built and developed in a manner that makes their composition free frombias and correlation fallacy.