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

Assessment and Accountability Obligation

The Assessment and Accountability Obligation speaks to the obligation to assess an AI system prior to and during deployment. Regarding assessment, it should be understood that a central purpose of this obligation is to determine whether an AI system should be established. If an assessment reveals substantial risks, such as those suggested by principles concerning Public Safety and Cybersecurity, then the project should not move forward.

Principle: Universal Guidelines for AI, Oct, 2018

Published by Center for AI and Digital Policy

Related Principles

Fairness Obligation

The Fairness Obligation recognizes that all automated systems make decisions that reflect bias and discrimination, but such decisions should not be normatively unfair. There is no simple answer to the question as to what is unfair or impermissible. The evaluation often depends on context. But the Fairness Obligation makes clear that an assessment of objective outcomes alone is not sufficient to evaluate an AI system. Normative consequences must be assessed, including those that preexist or may be amplified by an AI system.

Published by Center for AI and Digital Policy in Universal Guidelines for AI, Oct, 2018

Transparency

Review mechanisms will ensure citizens can question and challenge AI based outcomes Not only must the people of NSW have high levels of assurance that data is being used safely and in accordance with relevant legislation, they must also have access to an efficient and transparent review mechanism if there are questions about the use of data or AI informed outcomes. The development of AI solutions must be robust technically, legally and ethically. The community should be engaged on the objectives of AI projectsand insights into data use and methodology should be made publicly available unless there is an overriding public interest in not doing so. Projects should clearly demonstrate: a publicly available project objective and planned outcomes how the public can question and seek reviews of AI based decisions how the community can get insights into data use and methodology how the community will be informed of changes to an AI solution, including where existing technology is adapted for another purpose.

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

4. Fairness Obligation.

Institutions must ensure that AI systems do not reflect unfair bias or make impermissible discriminatory decisions. [Explanatory Memorandum] The Fairness Obligation recognizes that all automated systems make decisions that reflect bias and discrimination, but such decisions should not be normatively unfair. There is no simple answer to the question as to what is unfair or impermissible. The evaluation often depends on context. But the Fairness Obligation makes clear that an assessment of objective outcomes alone is not sufficient to evaluate an AI system. Normative consequences must be assessed, including those that preexist or may be amplified by an AI system.

Published by The Public Voice coalition, established by Electronic Privacy Information Center (EPIC) in Universal Guidelines for Artificial Intelligence, Oct 23, 2018

5. Assessment and Accountability Obligation.

An AI system should be deployed only after an adequate evaluation of its purpose and objectives, its benefits, as well as its risks. Institutions must be responsible for decisions made by an AI system. [Explanatory Memorandum] The Assessment and Accountability Obligation speaks to the obligation to assess an AI system prior to and during deployment. Regarding assessment, it should be understood that a central purpose of this obligation is to determine whether an AI system should be established. If an assessment reveals substantial risks, such as those suggested by principles concerning Public Safety and Cybersecurity, then the project should not move forward.

Published by The Public Voice coalition, established by Electronic Privacy Information Center (EPIC) in Universal Guidelines for Artificial Intelligence, Oct 23, 2018

4. Risk Assessment and Management

Regulatory and non regulatory approaches to AI should be based on a consistent application of risk assessment and risk management across various agencies and various technologies. It is not necessary to mitigate every foreseeable risk; in fact, a foundational principle of regulatory policy is that all activities involve tradeoffs. Instead, a risk based approach should be used to determine which risks are acceptable and which risks present the possibility of unacceptable harm, or harm that has expected costs greater than expected benefits. Agencies should be transparent about their evaluations of risk and re evaluate their assumptions and conclusions at appropriate intervals so as to foster accountability. Correspondingly, the magnitude and nature of the consequences should an AI tool fail, or for that matter succeed, can help inform the level and type of regulatory effort that is appropriate to identify and mitigate risks. Specifically, agencies should follow the direction in Executive Order 12866, “Regulatory Planning and Review,”to consider the degree and nature of the risks posed by various activities within their jurisdiction. Such an approach will, where appropriate, avoid hazard based and unnecessarily precautionary approaches to regulation that could unjustifiably inhibit innovation.

Published by The White House Office of Science and Technology Policy (OSTP), United States in Principles for the Stewardship of AI Applications, Nov 17, 2020