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

3 Ensure transparency, explainability and intelligibility

AI should be intelligible or understandable to developers, users and regulators. Two broad approaches to ensuring intelligibility are improving the transparency and explainability of AI technology. Transparency requires that sufficient information (described below) be published or documented before the design and deployment of an AI technology. Such information should facilitate meaningful public consultation and debate on how the AI technology is designed and how it should be used. Such information should continue to be published and documented regularly and in a timely manner after an AI technology is approved for use. Transparency will improve system quality and protect patient and public health safety. For instance, system evaluators require transparency in order to identify errors, and government regulators rely on transparency to conduct proper, effective oversight. It must be possible to audit an AI technology, including if something goes wrong. Transparency should include accurate information about the assumptions and limitations of the technology, operating protocols, the properties of the data (including methods of data collection, processing and labelling) and development of the algorithmic model. AI technologies should be explainable to the extent possible and according to the capacity of those to whom the explanation is directed. Data protection laws already create specific obligations of explainability for automated decision making. Those who might request or require an explanation should be well informed, and the educational information must be tailored to each population, including, for example, marginalized populations. Many AI technologies are complex, and the complexity might frustrate both the explainer and the person receiving the explanation. There is a possible trade off between full explainability of an algorithm (at the cost of accuracy) and improved accuracy (at the cost of explainability). All algorithms should be tested rigorously in the settings in which the technology will be used in order to ensure that it meets standards of safety and efficacy. The examination and validation should include the assumptions, operational protocols, data properties and output decisions of the AI technology. Tests and evaluations should be regular, transparent and of sufficient breadth to cover differences in the performance of the algorithm according to race, ethnicity, gender, age and other relevant human characteristics. There should be robust, independent oversight of such tests and evaluation to ensure that they are conducted safely and effectively. Health care institutions, health systems and public health agencies should regularly publish information about how decisions have been made for adoption of an AI technology and how the technology will be evaluated periodically, its uses, its known limitations and the role of decision making, which can facilitate external auditing and oversight.

Principle: Key ethical principles for use of artificial intelligence for health, Jun 28, 2021

Published by World Health Organization (WHO)

Related Principles

1. Transparency and Explainability

Transparency refers to providing disclosure on when an AI system is being used and the involvement of an AI system in decision making, what kind of data it uses, and its purpose. By disclosing to individuals that AI is used in the system, individuals will become aware and can make an informed choice of whether to use the AIenabled system. Explainability is the ability to communicate the reasoning behind an AI system’s decision in a way that is understandable to a range of people, as it is not always clear how an AI system has arrived at a conclusion. This allows individuals to know the factors contributing to the AI system’s recommendation. In order to build public trust in AI, it is important to ensure that users are aware of the use of AI technology and understand how information from their interaction is used and how the AI system makes its decisions using the information provided. In line with the principle of transparency, deployers have a responsibility to clearly disclose the implementation of an AI system to stakeholders and foster general awareness of the AI system being used. With the increasing use of AI in many businesses and industries, the public is becoming more aware and interested in knowing when they are interacting with AI systems. Knowing when and how AI systems interact with users is also important in helping users discern the potential harm of interacting with an AI system that is not behaving as intended. In the past, AI algorithms have been found to discriminate against female job applicants and have failed to accurately recognise the faces of dark skinned women. It is important for users to be aware of the expected behaviour of the AI systems so they can make more informed decisions about the potential harm of interacting with AI systems. An example of transparency in an AI enabled ecommerce platform is informing users that their purchase history is used by the platform’s recommendation algorithm to identify similar products and display them on the users’ feeds. In line with the principle of explainability, developers and deployers designing, developing, and deploying AI systems should also strive to foster general understanding among users of how such systems work with simple and easy to understand explanations on how the AI system makes decisions. Understanding how AI systems work will help humans know when to trust its decisions. Explanations can have varying degrees of complexity, ranging from a simple text explanation of which factors more significantly affected the decisionmaking process to displaying a heatmap over the relevant text or on the area of an image that led to the system’s decision. For example, when an AI system is used to predict the likelihood of cardiac arrest in patients, explainability can be implemented by informing medical professionals of the most significant factors (e.g., age, blood pressure, etc.) that influenced the AI system’s decision so that they can subsequently make informed decisions on their own. Where “black box” models are deployed, rendering it difficult, if not impossible to provide explanations as to the workings of the AI system, outcome based explanations, with a focus on explaining the impact of decisionmaking or results flowing from the AI system may be relied on. Alternatively, deployers may also consider focusing on aspects relating to the quality of the AI system or preparing information that could build user confidence in the outcomes of an AI system’s processing behaviour. Some of these measures are: • Documenting the repeatability of results produced by the AI system. Some practices to demonstrate repeatability include conducting repeatability assessments to ensure deployments in live environments are repeatable and performing counterfactual fairness testing to ensure that the AI system’s decisions are the same in both the real world and in the counterfactual world. Repeatability refers to the ability of the system to consistently obtain the same results, given the same scenario. Repeatability often applies within the same environment, with the same data and the same computational conditions. • Ensuring traceability by building an audit trail to document the AI system development and decisionmaking process, implementing a black box recorder that captures all input data streams, or storing data appropriately to avoid degradation and alteration. • Facilitating auditability by keeping a comprehensive record of data provenance, procurement, preprocessing, lineage, storage, and security. Such information can also be centralised digitally in a process log to increase capacity to cater the presentation of results to different tiers of stakeholders with different interests and levels of expertise. Deployers should, however, note that auditability does not necessarily entail making certain confidential information about business models or intellectual property related to the AI system publicly available. A risk based approach can be taken towards identifying the subset of AI enabled features in the AI system for which implemented auditability is necessary to align with regulatory requirements or industry practices. • Using AI Model Cards, which are short documents accompanying trained machine learning models that disclose the context in which models are intended to be used, details of the performance evaluation procedures, and other relevant information. In cases where AI systems are procured directly from developers, deployers will have to work together with these developers to achieve transparency. More on this will be covered in later sections of the Guide.

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

Plan and Design:

1 The planning and design of the AI system and its associated algorithm must be configured and modelled in a manner such that there is respect for the protection of the privacy of individuals, personal data is not misused and exploited, and the decision criteria of the automated technology is not based on personally identifying characteristics or information. 2 The use of personal information should be limited only to that which is necessary for the proper functioning of the system. The design of AI systems resulting in the profiling of individuals or communities may only occur if approved by Chief Compliance and Ethics Officer, Compliance Officer or in compliance with a code of ethics and conduct developed by a national regulatory authority for the specific sector or industry. 3 The security and protection blueprint of the AI system, including the data to be processed and the algorithm to be used, should be aligned to best practices to be able to withstand cyberattacks and data breach attempts. 4 Privacy and security legal frameworks and standards should be followed and customized for the particular use case or organization. 5 An important aspect of privacy and security is data architecture; consequently, data classification and profiling should be planned to define the levels of protection and usage of personal data. 6 Security mechanisms for de identification should be planned for the sensitive or personal data in the system. Furthermore, read write update actions should be authorized for the relevant groups.

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

· Plan and Design:

1 When designing a transparent and trusted AI system, it is vital to ensure that stakeholders affected by AI systems are fully aware and informed of how outcomes are processed. They should further be given access to and an explanation of the rationale for decisions made by the AI technology in an understandable and contextual manner. Decisions should be traceable. AI system owners must define the level of transparency for different stakeholders on the technology based on data privacy, sensitivity, and authorization of the stakeholders. 2 The AI system should be designed to include an information section in the platform to give an overview of the AI model decisions as part of the overall transparency application of the technology. Information sharing as a sub principle should be adhered to with end users and stakeholders of the AI system upon request or open to the public, depending on the nature of the AI system and target market. The model should establish a process mechanism to log and address issues and complaints that arise to be able to resolve them in a transparent and explainable manner. Prepare Input Data: 1 The data sets and the processes that yield the AI system’s decision should be documented to the best possible standard to allow for traceability and an increase in transparency. 2 The data sets should be assessed in the context of their accuracy, suitability, validity, and source. This has a direct effect on the training and implementation of these systems since the criteria for the data’s organization, and structuring must be transparent and explainable in their acquisition and collection adhering to data privacy regulations and intellectual property standards and controls.

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

· Transparency and explainability

37. The transparency and explainability of AI systems are often essential preconditions to ensure the respect, protection and promotion of human rights, fundamental freedoms and ethical principles. Transparency is necessary for relevant national and international liability regimes to work effectively. A lack of transparency could also undermine the possibility of effectively challenging decisions based on outcomes produced by AI systems and may thereby infringe the right to a fair trial and effective remedy, and limits the areas in which these systems can be legally used. 38. While efforts need to be made to increase transparency and explainability of AI systems, including those with extra territorial impact, throughout their life cycle to support democratic governance, the level of transparency and explainability should always be appropriate to the context and impact, as there may be a need to balance between transparency and explainability and other principles such as privacy, safety and security. People should be fully informed when a decision is informed by or is made on the basis of AI algorithms, including when it affects their safety or human rights, and in those circumstances should have the opportunity to request explanatory information from the relevant AI actor or public sector institutions. In addition, individuals should be able to access the reasons for a decision affecting their rights and freedoms, and have the option of making submissions to a designated staff member of the private sector company or public sector institution able to review and correct the decision. AI actors should inform users when a product or service is provided directly or with the assistance of AI systems in a proper and timely manner. 39. From a socio technical lens, greater transparency contributes to more peaceful, just, democratic and inclusive societies. It allows for public scrutiny that can decrease corruption and discrimination, and can also help detect and prevent negative impacts on human rights. Transparency aims at providing appropriate information to the respective addressees to enable their understanding and foster trust. Specific to the AI system, transparency can enable people to understand how each stage of an AI system is put in place, appropriate to the context and sensitivity of the AI system. It may also include insight into factors that affect a specific prediction or decision, and whether or not appropriate assurances (such as safety or fairness measures) are in place. In cases of serious threats of adverse human rights impacts, transparency may also require the sharing of code or datasets. 40. Explainability refers to making intelligible and providing insight into the outcome of AI systems. The explainability of AI systems also refers to the understandability of the input, output and the functioning of each algorithmic building block and how it contributes to the outcome of the systems. Thus, explainability is closely related to transparency, as outcomes and ub processes leading to outcomes should aim to be understandable and traceable, appropriate to the context. AI actors should commit to ensuring that the algorithms developed are explainable. In the case of AI applications that impact the end user in a way that is not temporary, easily reversible or otherwise low risk, it should be ensured that the meaningful explanation is provided with any decision that resulted in the action taken in order for the outcome to be considered transparent. 41. Transparency and explainability relate closely to adequate responsibility and accountability measures, as well as to the trustworthiness of AI systems.

Published by The United Nations Educational, Scientific and Cultural Organization (UNESCO) in The Recommendation on the Ethics of Artificial Intelligence, Nov 24, 2021

4 Foster responsibility and accountability

Humans require clear, transparent specification of the tasks that systems can perform and the conditions under which they can achieve the desired level of performance; this helps to ensure that health care providers can use an AI technology responsibly. Although AI technologies perform specific tasks, it is the responsibility of human stakeholders to ensure that they can perform those tasks and that they are used under appropriate conditions. Responsibility can be assured by application of “human warranty”, which implies evaluation by patients and clinicians in the development and deployment of AI technologies. In human warranty, regulatory principles are applied upstream and downstream of the algorithm by establishing points of human supervision. The critical points of supervision are identified by discussions among professionals, patients and designers. The goal is to ensure that the algorithm remains on a machine learning development path that is medically effective, can be interrogated and is ethically responsible; it involves active partnership with patients and the public, such as meaningful public consultation and debate (101). Ultimately, such work should be validated by regulatory agencies or other supervisory authorities. When something does go wrong in application of an AI technology, there should be accountability. Appropriate mechanisms should be adopted to ensure questioning by and redress for individuals and groups adversely affected by algorithmically informed decisions. This should include access to prompt, effective remedies and redress from governments and companies that deploy AI technologies for health care. Redress should include compensation, rehabilitation, restitution, sanctions where necessary and a guarantee of non repetition. The use of AI technologies in medicine requires attribution of responsibility within complex systems in which responsibility is distributed among numerous agents. When medical decisions by AI technologies harm individuals, responsibility and accountability processes should clearly identify the relative roles of manufacturers and clinical users in the harm. This is an evolving challenge and remains unsettled in the laws of most countries. Institutions have not only legal liability but also a duty to assume responsibility for decisions made by the algorithms they use, even if it is not feasible to explain in detail how the algorithms produce their results. To avoid diffusion of responsibility, in which “everybody’s problem becomes nobody’s responsibility”, a faultless responsibility model (“collective responsibility”), in which all the agents involved in the development and deployment of an AI technology are held responsible, can encourage all actors to act with integrity and minimize harm. In such a model, the actual intentions of each agent (or actor) or their ability to control an outcome are not considered.

Published by World Health Organization (WHO) in Key ethical principles for use of artificial intelligence for health, Jun 28, 2021