The National Guidelines On AI Governance & Ethics PDF

THE NATIONAL GUIDELINES ON AI GOVERNANCE & ETHICS “AI FOR MALAYSIA, AI FOR ALL” To Enhance The Development and Deployment of AI Technology Published by: Ministry of Science, Technology and Innovation Floors 1-7, Blo...

THE NATIONAL GUIDELINES ON AI GOVERNANCE & ETHICS “AI FOR MALAYSIA, AI FOR ALL” To Enhance The Development and Deployment of AI Technology Published by: Ministry of Science, Technology and Innovation Floors 1-7, Block C4 & C5, Complex C, Federal Government Administrative Centre 62662 Putrajaya, MALAYSIA 603 8885 8084 | [email protected] | mosti.gov.my ©Publisher’s Copyright All rights reserved. No part of this publication may be reproduced, copied, stored in any retrieval system or transmitted in any form or by any means-electronic, mechanical, photocopying, recording or otherwise, without prior permission. Table of Contents Section 2: Understanding AI Foreword 2.1 Understanding AI 8 Foreword By The Deputy Prime Minister III 2.2 A Case for Change - AI in Modern Transformational World 11 Preface By The Minister of MOSTI IV 2.3 Ethical Issues in AI 13 2.4 Building Trust 14 Executive Summary V 2.5 Global Calls for Ethics of AI 16 Preamble VI 2.6 National Considerations for Responsible AI 18 2.7 Responsible AI 21 2.8 Human-Friendly AI (HF-AI) Platform 28 Section 1: Introduction Section 3: National Guidelines For Stakeholders 1.1 Introduction 2 Introduction: Summary of Part A, Part B and Part C 32 1.2 Key Considerations for National Guidelines Development 4 1.3 Objectives of The National Guidelines on AIGE 5 Part A: National Guidelines For End Users of AI A1.0 Introduction 36 A2.0 Responsible AI for the End Users 38 A3.0 What You Can and Cannot do as End Users 42 Aligning End Users Contributions with Responsible AI and A4.0 43 Sustainability I  Table of Contents Part B: National Guidelines For Policy Makers In Government, Agencies, Section 4: Benchmarking And Organizations And Institutions International Smart Partnership B1.0 Introduction 46 B2.0 Responsible AI for Policy Makers 4.1 Benchmarking and x International Smart Partnership 106 51 B3.0 What You Can Do as Policy Makers 61 B4.0 The Independent Advisory Body. 69 B5.0 Aligning Government Initiative with Responsible AI and SDG & ESG Principles 70 Section 5: Conclusion 5.1 Conclusion 110 5.2 Moving Forward 111 Part C: National Guidelines For Developers, Designers, Technology Providers And List of Abbreviations 114 Suppliers 72 List of Figures & Tables 115 C1.0 Introduction C2.0 Responsible Principles for Sector Players 74 List of Appendices 116 C3.0 What You Can Do as Sector Players 78 Acknowledgement 135 C4.0 Demonstration and Sectors Cases 93 C5.0 Code of Ethics 102 C6.0 Aligning Industrial Economic Contributions with 103 Responsible AI and SDG & ESG Principles II  Foreword By Deputy Prime Minister Artificial Intelligence (AI) is revolutionizing various aspects of society, including governance, work, and problem-solving. However, it is crucial to uphold ethical principles such as fairness, privacy, security, and transparency. To develop and use AI responsibly, we must work towards technological, societal, institutional, and legal approaches that align our AI systems with our values by actively prioritizing transparency, accountability, and fairness. The Malaysian government is committed to ensuring AI technologies are deployed responsibly, respecting human rights, privacy, and societal well-being. Hence, the National Guidelines of AI Governance & Ethics have been prepared to meet international standards for sustainable development and corporate responsibility. Consumer rights must be at the center of AI development and implementation. Policymakers must establish policies and pass laws and regulations that provide safe and consumer-centered technology, utilizing principles such as the right to object and receive an explanation, the right to have personal data deleted, and the right to collective redress. As we navigate a sustainable future, it is essential to align our approach to AI with global sustainability goals and environmental, social, and governance requirements. On behalf of the Government of Malaysia, I would like to thank all ministries and government agencies, particularly the Ministry of Science, Technology & Innovation (MOSTI), the Working Committee of the National Blockchain and Artificial Intelligence Committee (NBAIC), academia, namely, Universiti Teknologi Malaysia, industry partners, including associations, civil society organisations, and professional bodies, both local and international, who have shared valuable input and experiences. All YAB Dato’ Sri Haji Fadillah bin Haji Yusof have worked hand in hand to make these National Guidelines of AI Governance & Ethics possible. Congratulations! Deputy Prime Minister III  Preface By Minister, MOSTI The Ministry of Science, Technology and Innovation (MOSTI) is implementing the National AI Roadmap 2021-2025 to harness AI's power for economic growth and society's betterment. One of the Roadmap strategies include the establishment of AI Governance which emphasis on AI leadership, governance, ethics, cybersecurity, standard development, risk management, and assessing the impact of responsible AI. We also recognize the importance of establishing clear national guidelines for the responsible use of AI technologies and ethical risks. These guidelines promotes seven AI principle throughout the AI lifecycle. The Ministry and Ministry of Investment, Trade and Industry (MITI) are actively participating in ISO/IEC JTC 1/SC 42 AI international standardization efforts to develop technical standards for AI technologies internationally. Currently MOSTI aggressively working to create a thriving national AI innovation ecosystem that allows everyone (government, business and society) to capitalize on the benefits of AI in a secured and safe manner for economic prosperity and social well-being. Implementing our National AI Roadmap Action Plans represents a significant step forward in harnessing AI's transformative power for society's benefit. YB Tuan Chang Lih Kang Together, let us embark on this journey towards a more prosperous, equitable, and Minister of Science, Technology and Innovation sustainable future powered by AI innovation and embrace the transformative power of AI while ensuring its benefits are enjoyed by all and no one is left behind. I hope these National Guidelines on AIGE will provide a reference to build trust, help to mitigate ethical risk as well as enhance AI adoption while simultaneously accelerating the country’s competitiveness. IV Executive Summary Artificial Intelligence (AI) is the intelligence of machines or software as opposed to the intelligence of humans. The ultimate goal of AI is to create technology to enable machines to function in an intelligent manner, which includes learning, reasoning, understanding natural language, and perception, together with helping to enhance human capabilities and improve decision-making. Generative AI (Gen-AI) has recently become a focus of excitement and is a technology that involves training models to learn linguistic patterns and structures from large datasets and then using those learned patterns to generate new content. The tremendous contribution of AI to sustainable economic development in different industries is rapidly becoming evident, making it an instant focus of attention at industrial, academic, and even government levels. In August 2022, the Malaysian Ministry of Science, Technology, and Innovation (MOSTI) launched the Malaysian National Artificial Intelligence Roadmap 2021-2025 (AI-RMAP), which set out frameworks for the integration of Artificial Intelligence (AI) into the different sectors of the economy through policy initiatives. Malaysia will be considered a high-technology nation when cutting-edge technologies like AI become a critical driver of productivity and competitiveness for the whole economy, not just the technology sector. For this reason, Malaysia is actively participating in ISO Artificial Intelligence Standards development (ISO/IEC JTC 1/SC 42). The Malaysian government has proposed seven principles to guide the development of Trusted and Responsible AI, in alignment with principles established by UNESCO, the OECD, the European Commission, and others. These are: Fairness Reliability, Safety and Control Privacy and Security Inclusiveness Transparency Accountability Pursuit of human benefit and happiness The intention of the National Guidelines on AI Governance & Ethics (AIGE) is to develop and deploy AI in a safe, trustworthy, and ethical manner, taking an approach to developing and deploying AI from both an ethical and legal point of view in line with AI-RMAP. V  Preamble These National Guidelines AIGE (“the National Guidelines”) are intended to establish Responsible AI practices as the government aspires to enhance the development and deployment of Artificial intelligence (AI) through the Malaysian National Artificial Intelligence Roadmap 2021- 2025 (AI-RMAP). The government hopes that the responsible use of AI will benefit the three main stakeholders, namely (1) the society, consisting of users, workers, and consumers in general, (2) government agencies and policymakers, and (3) the industry, including developers and designers. AI is the application of technology to carry out intelligent behavior using data and input on a computer system. The ultimate goal of AI is to create technology to enable machines to function in an intelligent manner which includes learning, reasoning, understanding natural language, and perception, which together help to enhance human capabilities and improve decision-making. We can also define AI as the simulation of human intelligence in machines programmed to think like humans and imitate their actions. The field of AI brings together several approaches and techniques, including machine learning, natural language processing, robotics, and computer vision. In August 2022, the Malaysian Ministry of Science, Technology, and Innovation (MOSTI) launched the Malaysian National Artificial Intelligence Roadmap 2021-2025 (AI-RMAP), which set out frameworks for the integration of AI into the different sectors of the economy through policy initiatives. To achieve the objectives of the roadmap, two action plans have been initiated: the development of National Guidelines for AI Governance and Ethics and the Institutionalization of Responsible AI. These documents intend to develop and employ AI in a safe, trustworthy, and ethical manner, taking an approach to developing and deploying AI from both an ethical and a legal point of view. The Malaysian government recognizes that legal and cultural requirements for the use of AI may vary from one culture to another, especially in Malaysia. Using the National Guidelines, it aspires to promote the beneficial use of AI aligned with the national ethical principles embodied in our Code of Conduct. Acknowledging that AI is a dynamic field, the National Guidelines will be amended from time to time to reflect technological progress and changing ethical norms. VI  Section 1 INTRODUCTION 1.1 Introduction 2 1.2 Key Considerations for National Guidelines 4 Development 1.3 Objectives of The National Guidelines AIGE 5 Image: Storyset.com 1  1.1 INTRODUCTION AI plays a major and increasing role in modern society, and AI technology is appearing in various forms at all levels of our contact with society, from AI- powered devices in our pockets and handbags and small daily chatting intelligent robots to large industrial and government-level assisted offices where influential decisions are made at a national government or organizational level (Qin et al., 2023). AI is described as a system that uses the power of science and technology to create a system that can think and act like humans, and which can even surpass some precise algorithms of humans, having been applied and put into practice in many fields. AI-RMAP has defined AI as “a suite of technologies that enables machines to demonstrate intelligence, to adapt to new circumstances, and is used to amplify human ingenuity and intellectual capabilities across a broad range of challenges” (AI-RMAP, 2021). Generative AI (GenAI) has recently become a focus of excitement. Gen-AI is an AI technology that involves training models to learn patterns and structures from large datasets and then using those learned patterns to generate new content. This content can be manifested in various forms, ranging from textual representations in natural language, and images (including photographs, digital paintings, and cartoons) to videos, music, and software code, thus covering the full spectrum of symbolic expressions of human thought. AI is basically designed to enable humans to make the best decisions. Its tremendous contribution to sustainable economic development in different industries is rapidly becoming evident, making it an instant focus of attention at the industrial, academic and even government levels - Heylighen, 2017 - In comparison, less than As in blockchain, 9 out of 10 70% leading businesses have invested heavily in AI technologies, and 37% 15% of major transnational companies will of major companies in the same sectors from 4 of the largest of businesses and organizations in education, manufacturing, will have adopted at least one type of developing countries including India, Brazil, Mexico and retail, tourism, pharmaceuticals, IT, communication and AI technology in 5 major sectors or Indonesia are deploying AI. automobiles in developed countries are employing AI. areas, namely finance, education, retail, transport and automation These are among the biggest This expansion of AI will eliminate 85 million “traditional” by 2030. - McKinsey Global Institute - employers; and they intend to increase that number to 90% by 2030. jobs and create 97 million new job opportunities or categories of jobs by 2025 - Bughin et al. 2018 - - Deloitte Report, 2021 - This confirms Source: how AI is being NAIC, CSIRO, aggressively Australia & Nextused by all Move parties inConsulting Strategy carrying out their daily tasks. It is used to stay connected with new and returning customers through auto-reply emails, appointment reminders, and feedback surveys. In this regard, AI is capable of improving business practices. 2  Transition Towards Adoption of AI Technology And Its Economic Value Global economy contribution currently million expected to USD $400 increase to trillion by the year USD PwC Global, $15.7 2030 Global economic activity of around $13 trillion by 2030 Global AI market McKinsey Value is to reach USD $267 billion by 2027 ASEAN AI contribution to (Fortune Business Insights, 2021) GDP Malaysia USD 950 billion USD 115 billion by 2030 (14%) by 2030 (Kearney Analysis, 2020) 3  1.2 KEY CONSIDERATIONS FOR THE NATIONAL GUIDELINES DEVELOPMENT The following are taken into consideration in the development of the National Guidelines on AI Governance & Ethics for Responsible and Inclusive AI (National Guidelines on AIGE): The National Guidelines need to be inclusive and cater for a range of stakeholders, and must not create an AI for All exclusive space. This is to ensure that no one is left behind (AI for members of the public). The National Guidelines constitute a policy document for stakeholders, and showcase government intervention Building Trust and commitment to governing AI as part of Responsible Government, and to build trust, and obtain the trust of the nation. Comprehensible The National Guidelines need to be simple so that the stakeholders can understand them. Alignment with National, Regional The National Guidelines refer to documents at national, regional and global levels, ensuring their coherency and Global AI Practices worldwide, and also draw on local and indigenous knowledge. The National Guidelines reflect synergy among- government agencies and the stakeholders operationalize and Synergy institutionalize AI in implementing national policies and action plans. The National Guidelines serve as a reference and call to action for different sectors (healthcare, finance, etc) to Call to Action develop their implementation guidelines to serve their own sectors. Since AI is a dynamic technology, these National Guidelines will be reviewed regularly and updated to meet the Living Document requirements and expectations of stakeholders. The National Guidelines will be made available in both Malay and English. 4 1.3 OBJECTIVES OF THE NATIONAL GUIDELINES AIGE As mentioned in the Preamble section, these national guidelines are on a voluntary for the stakeholders. Nevertheless, these National Guidelines are intended to disseminate the importance of AI and Responsible AI to the entire nation with the theme of: AI for Malaysia, AI for All While Sections 1 and 2 are for general knowledge, Section 3 is divided into three separate parts according to the different stakeholders: National Guidelines National Guidelines for Policy Makers National Guidelines for Developers, for AI End Users in Government, Agencies, Designers, Technology Providers Organizations and Institutions and Suppliers Further explanation concerning the stakeholders are included in the Introduction to Section 3 The national guidelines are prepared with the stakeholders in mind. The objectives of these National Guidelines are: I. To support the implementation of the Malaysia National AI Roadmap 2021 - 2025 II. To facilitate the implementation of Responsible AI according to the 7 AI Principles III. To build trustworthiness in AI, which is emphasized by Responsible AI. IV. To manage risks caused by the development and deployment of AI technology. V. To maximize the benefits of AI to enhance the national productivity, economic growth and competitiveness. 5 “ The most urgent problem today is not to limit the development of AI or even to ensure it’s aligned with human values. It’s “ to ensure we have a plan to ensure it does what we want. - Stuart Russell- 6  Section 2 UNDERSTANDING AI 2.1 Understanding AI 8 2.2 A Case for Change - AI in Modern 11 Transformational World 2.3 Ethical Issues in AI 13 2.4 Building Trust 14 2.5 Global Calls on the Ethics of AI 16 2.6 National Considerations for Responsible AI 18 2.7 Responsible AI 21 2.8 Human-Friendly AI (HF-AI) Platform 28 7  2.1 UNDERSTANDING AI This introduction intends to introduce topics and explanations on the terminologies of AI in the hope that readers will find them useful and enlightening. AI is a rapidly evolving field of research and innovative technologies. Thus, to give readers, an idea of how AI has arrived at the current state of the art, this section traces its nature and evolutions over the decades. Artificial Intelligence (AI) The field of computer science that seeks to create intelligent machines that can replicate or exceed human intelligence. Artificial Intelligence (AI) Machine learning 1950s A subset of AI that incorporates aspects of computer science, mathematics, and coding. Focuses on developing algorithm and models that help machines Machine learning learn from data and predict trends and behaviours, without human assistance. 1959 Deep learning A function of AI that imitates the human brain by learning from how Deep learning it structures and processes information to make decisions. Instead of relying on an algorithm that can only perform one specific task, 2017 this subset of machine learning can learn from unstructured data without supervision. Generative AI Generative AI 2021 Models of algorithms that create new output such as text, video, photo, code, data, statistics from huge amounts of data that they are training of. Figure 1: The evolution of AI technology 8 2.1 UNDERSTANDING AI a. Traditional AI AI, a technology developed since the mid-20th century, has Advancements in artificial intelligence (AI) have yielded significant revolutionized various fields, including Humanities and Natural progress in solving complex problems through text analysis. This Science. Its development has been influenced by the need to replace approach involves identifying fundamental linguistic components such physical work, such as washing clothes, with AI, which has evolved as nouns, adjectives, and verbs. The Hidden Markov Model (HMM) to encompass various interpretations and applications. played a pivotal role in enabling linguists to comprehend how computers solve problems. This model facilitates the discovery of novel The has grown significantly from its roots in basic computation and solutions rather than relying solely on pre-programmed ones. possesses the remarkable capability to undertake cognitive tasks. This Furthermore, AI aspires to replicate human intuition, akin to the steering empowers machines to handle a wide range of activities, from actions employed by human drivers, a development proving crucial for mundane chores such as garment cleaning to complex mathematical the advancement of self-driving car technology. computations. However, unlike humans, machines do not attempt to replicate the intricate workings of the human brain. This fundamentally Artificial intelligence (AI) is revolutionizing facial recognition technology different approach results in their "artificial" intelligence. Paradoxically, by drawing inspiration from the brain's communication and problem- this non-human method can create a convincing illusion of sentience solving mechanisms. Through the integration of machine learning when computer programs utilize language processing. algorithms, AI strives to replicate the natural processes involved in In the 1960s, the ELIZA program emerged as a response to the Turing human face recognition. This approach holds the potential to not only test, a seminal concept designed to assess a machine's ability to improve facial recognition accuracy but also to deepen our scientific convincingly simulate human conversation. ELIZA's objective was to understanding of how neural networks within the brain communicate comprehend and respond to user input, thereby prompting inquiry into and facilitate intelligent tasks. the very nature of human intelligence. As advancements facilitated machines in mimicking human intelligence with increasing efficacy, CAPTCHAs were subsequently developed to distinguish between Image: Freepik.com human and machine interaction. 9  b. Generative AI c. AI and General Intelligence Artificial intelligence (AI) has significantly revolutionized the field of Recent research indicates that human intelligence is not a single ability, machine learning by leveraging extensive datasets. The vast amount of but a combination of multiple abilities. While humans excel in specific tasks data readily available on the internet fuels this progress, enabling the like kicking a ball, machines are designed for general intelligence (AGI), creation of novel machine-learning models capable of generating new which is expected to be achieved in the near future. AGI aims to build forms of data, including text, images, and audio files. This advancement machines capable of performing multiple tasks like humans. has demonstrably enhanced large-scale computing capabilities. Generative AI, a subfield of artificial intelligence, has garnered significant popularity in recent times. A prominent example is ChatGPT, which allows users to interact with seemingly intelligent machines. However, it is paramount to distinguish between machine intelligence and human intelligence. While machines are meticulously trained to emulate human behavior, they lack the fundamental characteristics of human-like consciousness. Attributing sentience or pseudo-consciousness to machines can lead to misconceptions. Since the generative AI approach was introduced in 2023, the Image by Vectorjuice from government has been exploring the policy, framework, opportunities, and Freepik risks including intellectual property matters. All these will be incorporated OpenAI, a leading AI development organization, aims to advance AGI, or in the next version of the AIGE. AI systems that are smarter than humans, benefiting everyone. However, the exact definition and definition remain unclear. Reliable sources like the Alan Turing Institute provide valuable insights into AI and AGI, introducing a moral aspect to AI research and technology. AI's current capabilities are being assessed, but potential dangers are also being recognized. The generative AI system recommending inappropriate ingredients underscores the need for a balance between recognizing Image: Storyset.com potential dangers and assessing AI's capabilities, as a machine's comprehension of its output is crucial. Image by Freepik 10  2.2 A CASE FOR CHANGE - AI IN A MODERN TRANSFORMATIONAL WORLD The following are some examples of AI and Generative AI in different sectors in Malaysia: Examples of Real-world AI and Business, Banking and Finance GenAI in Malaysia Sectors AI-driven chatbots are used in customer service, provide instant responses to inquiries, and improve the overall banking experience. Chat CIMB Bank was the first bank in Malaysia to launch an AI-driven chatbot, namely Enhanced Virtual Assistant (EVA). HSBC Bank launched a similar Chatbot Amy for its corporate banking clients. Amy understands both English and simplified Chinese, and can provide prompt responses to banking queries 24/7. RHB Bank and RinggitPlus, jointly launched My RHB Easy, a chatbot that gives users flexibility to apply for a personal loan, without the manual completion and submission of application forms. Hong Leong Bank Berhad has introduced a virtual assistant named HALI to improve the efficiency of bank operations. The chatbot developed for HLB is designed to assist two main core support functions, namely Human Resources and Branch Operations. It will take over staff queries on policies and procedures, previously undertaken by staff members from the two divisions 11  Examples of Real-world AI and GenAI in Malaysia The transportation sector The healthcare sector AirAsia has been at the forefront of using chatbots and artificial M3DICINE Technology Sdn Bhd, together with its strategic partner intelligence to enhance customer service and streamline their CREST (Collaborative Research in Engineering, Science and operations. Virtual personal assistant, AVA has been the first Technology) has launched Stethee Pro, the world’s first AI-enabled onboarding platform created in 2019 and designed to assist customers stethoscope system. with tasks such as flight bookings, seat selections, check-in procedures, and answering frequently asked questions in eleven languages. In 2023, The Stethee employs machine learning that can identify patterns ‘Ask Bo’, an AI-powered concierge, was launched in connection with and detailed analysis while monitoring the progress in the health of AirAsia’s commitment to transparent communications. Guests have individual patients and uncover new trends in the fight against heart more autonomy by being given the opportunity to talk live to human and lung disease globally (The Legal 500, 2021). agents and reveal live information during the Ask Bo interaction. 12 2.3 ETHICAL ISSUES IN AI The internet has created amazing opportunities, but it has also given both good and malicious people access to them. However, the advancement of AI (e.g. Generative AI) has increased the risk of spreading lies and enabling harmful actions, as machines are not trained for truth or ethical values. Some examples of ethical issues in AI are: Over-reliance on machine-led decisions without disclosure in the banking sector Limiting access/discriminatory pricing of products/services due to consumers’ race/gender Citizens object to using Customers demand reasoning/ facial recognition clarification for decision taken by an AI technology by police for algorithm to deny credit/ transaction mass surveillance Collecting and processing personal data in AI algorithms without individual consent Ethical consideration of colliding with wild animals or Citizens object to the Crewless Ships and self other vehicles. Autonomous Processing data in AI collection and use of their directed drone swarms; using Vehicle can not make a algorithms for purposes other data such as biometrics, by off-the-shelf tech for the next decision to avoid a than for which it was collected an AI system. stage of human conflict. collision. 13  2.4 BUILDING TRUST It is essential to build trust in AI to overcome resistance and pave the way for its widespread adoption. This Explains why a study was conducted from 2021 to 2022 to ascertain stakeholder perceptions of AI. Its findings are illustrated in Figure 2 below. In 2023, relevant government agencies introduced several actions and developed interventions to address the concerns of stakeholders identified in the study. Figure 2: Stakeholder perception of AI Dedicated task force/ committee for planning, Question: Al applications/ systems are secured Question: implementing and managing Al initiatives. Government Actions and Interventions 13% Government Actions and Interventions 16% Introducing and implementing Establish AI Governance the 7 AI Principles of ethics to structure 84% support Responsible AI 87% Perception: Perception: 84% Disagree 87% Disagree Safeguarded transparency in protecting the Study, review and update related Al policies and Question: Question: privacy of users and consumers regulations to accelerate Al development Government Actions and Interventions 8% Government Actions and Interventions 16% Preparing guidelines which are Introducing and implementing expected to be launched in March 84% the 7 AI Principles of ethics 92% 2024 Perception: Perception: 84% Disagree 92% Disagree Agree Disagree Respondent: 250 government staff, industry and associations totalling of 800 respondents in the Klang Valley 14 2.4.1 Trust and Benefits As illustrated in the study, many people were worried about the safety of AI technology. This document serves as a bridge to gain stakeholders’ trust through the implementation of responsible AI. Trust is essential for protecting stakeholders in their use of emerging technologies. Digital Trust, as defined by various expert organizations, is the expectations of individuals that “digital technologies and services – and the people, processes, and organizations providing them – will protect all stakeholders’ interests and ensure a secure digital world” (ISACA, WEF Digital Trust Initiative, 2022). The National Guidelines address the need to have AI systems governed, developed, and deployed while adhering to the 7 principles. The following are some potential benefits of AI when used responsibly: Improving the Healthcare and Societal Well Enhancing AI Technology Development 4 Being 1 In the healthcare sector, AI technologies are providing more Leadership in AI technology development can create the ability accurate diagnostics, personalized medicine, and predictive to set global norms and standards for the ethical, safe, and analytics for patient care, leading to better health outcomes and equitable use of AI. efficiency in healthcare systems. At the same time, AI has the potential to address major societal challenges, managing large- scale public health crises by analyzing data and creating predictive models, for example during the last COVID pandemic. 2 Increasing Economic Growth and Innovation AI drives innovation in industry, creating new products and 5 Improving Talent For Job Opportunities services, improving efficiency and productivity, and As AI changes the job landscape, countries see the fostering economic growth. Many companies invest in AI to importance of investing in AI to create new job opportunities, gain a competitive edge on the global stage.. upskill and reskill the workforce in line with the future needs of the industry, and build trust in AI. 3 Ensuring Security and Safety Enhancing The Quality of Life AI plays a critical role in national security and safety, from 6 AI may also improve the quality of life by enhancing services such as transportation, communication, and cybersecurity to the analysis of intelligence data and entertainment, making them more personalized, efficient, autonomous weaponry. and accessible. Given all these benefits, countries tend to regard AI as a strategic asset that is central to their future development, security, and competitiveness, prompting significant investment and policy focus in this area. As AI technology continues to evolve, its importance in national agendas is likely to grow even further. 15  2.5 GLOBAL CALLS ON THE ETHICS OF AI The first ethical code for AI systems was introduced by the famed With the evolution of the technology and advancement of the modern science fiction writer Isaac Asimov, who presented his Three Laws of world, the discussion of AI ethics started at an international platform in Robotics in Runaround (Asimov 1942). These three were later 2019, when the Organization for Economic Co-operation and supplemented by a fourth law, called the Zeroth Law of Robotics, in Development (OECD) introduced its Principles on AI. Robots and Empire (Asimov 1986). However, the first global wide discussion on the Ethics of AI was formally The four laws are as follows: started by UNESCO with the participation of its 193 member states in A robot may not injure a human being or, through inaction, allow November 2019 at the 40th General Conference of UNESCO, when they a human being to be harmed; adopted 40 C/Resolution 37, which mandated the Director-General “to A robot must obey the orders given to it by human beings except prepare an international standard-setting instrument on the ethics of where such orders would conflict with the first law; artificial intelligence (AI) in the form of a recommendation”. This was later A robot must protect its existence as long as such protection presented in 2021 as a Recommendation on the Ethics of AI. does not conflict with the first or second law; A robot may not harm humanity or, by inaction, allow humanity to suffer harm. UNESCO Recommendation on the Ethics of AI - Nov 2021 The Recommendation on the Ethics of AI (UNESCO, 2021) which is the first-ever global standard for AI Ethics was adopted by all 193 member states on 23 November 2021. The recommendation emphasises 10 AI Principles The protection of human rights and dignity is the cornerstone of the Recommendation, based on the advancement of fundamental principles such as transparency and fairness. This recommendation enables policymakers to translate the core values and principles into action to data governance, environment and ecosystems, gender, education and research, health and social well-being, etc. UNESCO encouraged each of the countries to develop AI governance for their country. 16  Malaysia Alignment with Global Calls : The AI Roadmap - August 2022 To support the UNESCO recommendation on AI ethics, the Malaysian government introduced 7 AI Principles in August 2022 as These National Guidelines for AI explained in section 2.7.1 Governance & Ethics for The principles are included in the Malaysian National Artificial Responsible and Inclusive AI Intelligence Roadmap 2021-2025 (AI-RMAP) Malaysia 7 AI (National Guideline AIGE) are Principles The Malaysian government has also taken the initiative to develop aligned to international best these National Guidelines to enhance AI technology development practices and and adoption. standards in principles. There are currently varying approaches to AI governance and in ASEAN AI Governance and Ethics Guidelines - February 2024 line with the emerging global consensus, Malaysia supports a risk-based approach that will allow 2022 2024 for alignment with international approaches, which will support national competitiveness. At the regional level, ASEAN Member States approved the ASEAN Guide on AI Governance and Ethics in February 2024 to empower organizations and governments in the region to use AI responsibly and increase users’ confidence in AI. This document serves as a practical guide for organizations in the region that wish to develop and deploy AI technologies. 17  2.6 NATIONAL CONSIDERATIONS FOR RESPONSIBLE AI 2.6.1 Rukun Negara (National Principles) Generally, any ethical guidelines or framework that supports values must be consistent with the constitution - a constitutional ethic’ comprising principles and values, particularly inalienable human rights and civil liberties that line up with the aspirations of the framers of the Federal Constitution (Sadhu Singh, 2023). In addition, other national values are laid down in the Constitution Establishing Islam as the official religion of the Federation while allowing freedom to practice other faiths and the Constitutional Monarch, Yang di-Pertuan Agong (‘YDPA’), as the supreme head of the Federation, and the supremacy of the Constitution. Conceived by the National Consultative Council, a diverse body tasked with rebuilding harmony, the Rukun Negara was officially proclaimed on 31 August 1970. Rukun Negara is not just a set of principles, but constitutes a declaration of shared values, a common ground upon which all Malaysians can unite, regardless of ethnicity, religion, or background. Its five core principles -Belief in God, Loyalty to King and Country, Supremacy of the Constitution, Rule of Law, and Courtesy and Morality -aimed to bridge the divides and foster a sense of shared identity. The Rukun Negara and the spirit of the Federal Constitution of Malaysia are intimately linked. Both documents enshrine core values that guide the nation's development. Both emphasize the importance of a pluralistic society based on mutual respect and understanding. The Rukun Negara can be seen as a complementary text that strengthens and clarifies the spirit of the Federal Constitution. The fundamental or core values of a nation-state are found in its constitution, comprising political, religious, moral, cultural and economic values (Faruqi, 2019). These values must naturally form an ethical framework adopted in Malaysia, namely the Responsible AI Principles. With the supremacy of the Constitution Preserved in Article 4(1) of the Federal Constitution, which declares that the ‘Constitution is the supreme law of the Federation and that any law passed after Merdeka Day which is inconsistent with this Constitution shall, to the extent of the inconsistency be void’, the Constitution provides a bedrock of values which when distilled operate alongside the Responsible AI Principles. 18 2.6.2 Rukun Negara and Responsible AI Building a responsible AI ecosystem in Malaysia necessitates interweaving the nation’s core values with the 7 AI Principles of responsible AI. Belief in God and Fairness Loyalty to King and Country and 1 Adhering to this core value involves: 2 Transparency Aligning the development of AI with the 7 AI principles that This includes encompass diverse religious perspectives, and preventing Accountability and Public Trust which involves ensuring that there algorithmic bias based on religious beliefs, and is transparency in AI decision-making processes, upholding the Ensuring that the development is consistent with the Federal King's role as a symbol of unity and the Constitution's emphasis Constitution's spirit of equality and non-discrimination avoiding on accountability, and any form of discrimination on account of religion and any other The National Interest and Security which involves the protected characteristics. development of AI applications that safeguard national security and the public interest, aligning with the principle of loyalty while 3 Supremacy of the Constitution adhering to legal and ethical frameworks. & and Rule of Law Courtesy & Morality and Human These two core values include matters related to: 5 Dignity 4 Safety and Security to ensure that AI systems comply with legal and ethical frameworks, respecting the fundamental rights and This core value requires that the AI systems are designed to: liberties enshrined in the Federal Constitution. Respect individual privacy and dignity, aligning with the principle Inclusivity and Diversity which incorporate diverse perspectives in of courtesy and the Federal Constitution's protection of privacy. AI development and governance, reflecting the Malaysian Emphasize Social Responsibility: Utilizing AI to address societal multicultural nature of Malaysia and the Federal Constitution's challenges and promote social good, upholding the spirit of emphasis on equality. ethical conduct and the Federal Constitution's commitment to a Human Control and Oversight to maintain human control over AI just and equitable society. systems, preventing autonomous actions that could violate constitutional rights or lead to unintended consequences. The thoughtful integration of these values and principles makes it possible for Malaysia to harness the power of AI, and at the same upholds its national identity and commitment to justice, inclusivity, and human dignity. 19 2.6.3 Responsible AI from an Islamic Perspective Islamic principles guide the ethical development of AI, as Islam emphasizes social good for public interest and human welfare. The 7 AI principles align with Islamic practices and can be used as a foundation to analyze the implications of AI. Muslim participants believe that understanding and applying Islamic principles can contribute to a knowledgeable and ethical approach to adopting AI, as they align with the teachings of Allah and the Islamic principles of responsible AI. Islam, a way of life, guides all aspects of human existence, including the ethical adoption of AI technology. As viceregents (Khalifah) in the world, humans are responsible for the ethical consumption of resources provided by Allah. AI technologies have shown positive impacts on various aspects, such as education, business, and healthcare management. Islam accepts all technological progress, including AI, but requires ethical guidance to ensure it does not transgress Islamic boundaries and result in negative impacts. AI in decision-making processes must adhere to ethical principles, promoting social justice and avoiding unfair treatment. Islam emphasizes the importance of human authority over AI systems to prevent misapplication and violate ethical principles. AI can optimize resource utilization, reduce waste, and promote sustainable practices, contributing to environmental preservation. Muslims seek guidance from Islamic teachings to address moral challenges posed by AI, such as self-governing weapons and stewardship of resources and the environment. AI technology offers immense possibilities for Muslims, but it's crucial to respect human dignity, rights, and privacy. AI systems should promote inclusivity and diversity, avoiding disregard for human connection, compromise, and ethical judgment. Muslims believe AI should enhance human capabilities, not replace them, and upskilling and reskilling are essential for relevance in rapid technological changes. The Islamic perspective on AI emphasizes the importance of ethical regulation and innovation, aligning with the 7 AI Principles. AI can address societal challenges and benefit humanity, but it also requires careful consideration of ethical dilemmas and human values preservation. By actively engaging in AI development and regulation, Muslims can contribute to a future where technology and ethics align with Responsible AI, fostering collaboration and open dialogue. AI technology, while providing hardware and software, also plays a crucial role in spiritual values, providing the necessary heartware to drive activities forward. 20 2.7 RESPONSIBLE AI 2.7.1 AI Principles The aspiration for these National Guidelines is for the voluntary adoption To ensure comprehensive and consistent application, the operational of the seven AI Principles by developers and deployers of AI alongside definitions for the 7 AI Principles for Responsible AI in Malaysia will be existing laws to ensure that the adoption of AI technology is undertaken developed as the nation progresses on the trajectory of AI adoption in consideration of both the potential benefits to the nation as a whole as and evolution. This will ensure that the application of these principles is well as minimizing any harms and risks posed to the users. These in line with converging values found in existing ethical frameworks national guidelines should be operationalized as part of a self- worldwide. As an illustration, the first principle, Fairness is often governance mechanism alongside the use of AI assurance techniques associated with values such as justice, which includes respect for when assessing and auditing AI systems. diversity, inclusion, and equality, as well as avenues for redress and remedy to appeal, review, or challenge decisions. Similarly, the To comply with all of the 7 principles themselves, stakeholders need to seventh principle, Transparency, can be interpreted to include efforts to understand the main component of each principle and how it is applied increase explainability, interpretability, or other acts of communication to them, which to a certain extent, needs consideration on its own. While and disclosure. the national guidelines (as specified in Section 3) do provide general This is an approach to develop and deploy AI from both an ethical and guidance, it is important to note that each stakeholder group may have a legal point of view. The goal of responsible AI is to employ AI in a its interpretation of the principles, according to its sector and industry to safe, trustworthy, and ethical manner. There should be a clear which the technology applies. It is noted that the complexity of the way understanding of who is responsible for the outcomes of AI systems, the AI system works and adhering to the national guidelines may be far and that these systems are transparent and understandable to the end more intricate. However, the principles remain for the benefit of all users. stakeholders. There will be both convergence and divergence in the definition of these principles as these National Guidelines continue to iterate and evolve alongside other standards being developed for regional and global adoption. Malaysia 7 AI Principles 21  2.7.2 Morals, Principles and Ethics When you read about AI, or listen to podcasts on AI you will frequently come across references to ethics and principles, which have become increasingly important as AI has developed. Less frequently, you will see references to morals, especially in the phrase moral principles. The three words ethics, principles, and morals are very close in meaning, and in many contexts, they can replace each other without much changing the meaning. However, they come from different sources, and there are slight differences in the way they are appropriately used. Explaining the use of these terms is taken as the point of departure for this section. If 7 AI principles are considered important, it is essential to ask why this is the case, which is the topic of the section entitled “The need for ethical principles”. This completes the essential background for the discussion of the seven AI principles adopted by the MOSTI, which forms the focus of this whole section on ethics, principles, and morals Morals and Moral Principles Morals are rules based on socially acceptable behavior, and they come from the Latin word for customs. Principles are general rules of conduct, and moral principles often focus on prohibiting socially unacceptable actions like stealing or assaulting other people. These principles help us determine what is right and wrong, and they are connected to the Golden Rule which involves treating others as we would like to be treated. Moral values are shaped by social customs and can change as society evolves. For instance, a hundred years ago, concerns focused more on sexual behavior, but racism became a significant issue by the end of the century. Racism is considered immoral because it goes against the Golden Rule. Our understanding of principles can also shift over time. Stealing used to refer mainly to taking physical possessions like money or jewelry, but in recent times, it includes the unauthorized use of ideas or creative work, known as intellectual property. This poses challenges for AI companies, especially with generative AI, as the training data may contain unauthorized intellectual property, Image: Reshot.com impacting the output. 22  “ If we develop AI without any understanding of ethics, it will magnify our biases. If we develop AI without concern for its alignment with the human values, it will lack principles. It is up to us to “ ensure the technology is designed and used responsibly. - Max Tegmark- 23  The Need For Ethical Principles In an ideal world, people would behave following moral and ethical principles. That is alas not the kind of world in which AI has been developed. In the real world, AI is widely perceived to have the potential to bring with it catastrophic dangers, including in the extreme case the destruction of the entire human race. AI machines make use of new technologies, and many real dangers associated with AI belong to the long history of machines and new technologies. Although these dangers are real, they are not caused specifically by AI, and so belong to a different group than dangers specific to AI. The following subsections deal in turn with alleged catastrophic dangers, machines and technologies, and dangers associated with AI. Alleged Catastrophic Dangers Image: Freepik.com There are on the internet many warnings claiming that AI is leading to catastrophe and presents an existential threat to humanity (‘a danger that threatens the very existence of human beings). While there are real dangers that have to be taken very seriously, it is important not to invent imaginary dangers. Much confusion is caused by referring to machines as though they were human, and using expressions such as machine “learning”, “training” and “intelligence”. These are metaphors and they have to be understood as metaphors, which capture some resemblance between two entities without implying that they are the same thing. There is of course no foot at the foot of a mountain, and no head at the head of a valley. The development of machines is like learning and training, and the output is like intelligence; but these resemblances do not make them identical. If the distinction between resemblance and identity is disregarded, Image: Freepik.com and if we assume without evidence that machines will one day turn hostile and compete with humans for mastery of the world, the development of “superintelligent” machines is indeed frightening. The outlook is even more bleak if predictions that machines will become billions of times more intelligent than humans are to be believed. However, there are real dangers that we know will have to be faced. 24 AI should be used responsibly, respecting individual rights and liberty, and adhering to legal obligations and policy requirements. It should identify and mitigate potential bias, be robustly tested for reliability and safety, benefit everyone, use explainable methods, and be accountable for its effects. AI systems should be safe and secure, benefit everyone, and comply with federal constitutions. Accountability should be established at each stage and across its lifecycle, including maintaining records. Human-centered values should be respected. These AI elements should be periodically reviewed to ensure continued purpose and resolution of issues. To enhance public trust in AI, AI systems must comply with the Principles for Responsible AI outlined in the Malaysian National AI Roadmap 2021-2025. The section on Responsible AI Principles presents variations in their meaning across sections, potentially confusing their fundamental nature. A suggested revision is to focus on a single, detailed explanation of each principle, not repeated elsewhere. Other sections can cross-refer to this section, and consistency in language and substance should be maintained. Modeling the language and descriptions from the UNESCO Recommendation on the Ethics of Artificial Intelligence could be a possible approach. The AI principle of Accountability requires clarity on the definition of AI model owners and actors, as well as the definition of success or failure. In a B2B context, if an AI model is developed with responsible practices but biased data is submitted, the model may work successfully but produce undesirable output. AI governance is crucial for responsible development and use of technology, considering the complexity of AI ecosystems. Each party in the value chain should be assigned risk-proportionate obligations based on their role and control over data and context. AI developers often create customizable tools with low-risk purposes, but customers often control data submission, configuration, and output usage. B2B companies like Salesforce, as data processors, may not own or control customers' data, making it difficult for them to know or control how customers use their data in AI systems. Therefore, a reasonable division of responsibilities should be established between entities at different points along the AI supply chain, including B2B companies like Salesforce. 25  2.7.3 Seven (7) AI Principles We would appreciate the clarification on how these additional values will be integrated into and/or how these values will interact with the 7 AI Principles. These sections all set out some version of the Responsible AI Principles but there are variations of what each principle means, from section to section. This may lead to a suggested revision is that a single, detailed, and clear explanation of each of the Responsible AI Principles be set out as the main focus of this document and that it is not repeated elsewhere. Other sections can instead cross-refer to this section. In addition, there should be consistency in language and substance of the detailed description. Modelling on the language and descriptions from the UNESCO Recommendation on the Ethics of Artificial Intelligence (https://unesdoc.unesco.org/ark:/48223/pf0000381137) is a proposed approach. lack of clarity on the fundamental nature of the principles, that should be the same across all use of AI. 1 The development of AI must be designed to avoid bias or 2 discrimination against the target users. All AI systems should not have one-size-fits-all approaches and should address the widest possible range of factors (e.g. age, gender, religion, and ethnicity). FAIRNESS This enables equitable asses and active participation of all RELIABILITY, SAFETY AND CONTROL stakeholders AI systems should be safe and secure, and perform as 3 AI systems or solutions must be robustly tested to be reliable, safe intended, and resist being compromised by unauthorised and fail-safe by default so that users can trust and depend on the AI parties. solution. To ensure proper access, control and protection of critical situations, AI systems should work reliably, consistently and operate Developers should where necessary obtain consent from in a real world under normal circumstances and unexpected individuals before using or disclosing personal data for AI conditions. In order to prevent or mitigate negative outcomes, AI development and deployment. The systems must systems should be able to response quickly according to the guarantee privacy and data protection throughout the intended purposes. entire life cycle of the AI system. Information and data PRIVACY collected from users must not be used in an unlawful or Developers and end-users are encouraged to carry out relevant AND discriminatory way against them. Developers also should testing and certification as well as risk and impact assessments to incorporate security-by-design and privacy-by-design SECURITY prevent any potential harm and mitigate risks. Autonomous systems principles and refer to international standards related to must have safeguards to ensure ultimate controllability by humans, information security and privacy when implementing the AI particularly when it comes to the applications of AI in high-risk areas system. such as autonomous vehicles, military applications, and when human life is at stake. 26 4 5 INCLUSIVENESS TRANSPARENCY AI must be inclusive for all stakeholders to avoid unequal AI algorithms should be transparent to ensure that any capabilities can be explained, access to AI as in the case of social clefts. both technical processes of an AI systems and related human decisions. This will allow stakeholders to evaluate the risks of AI and address issues that may arise. AI systems should benefit everyone and address national needs and experiences inclusively in compliance with the Transparency principles are mainly applicable to contexts where AI is being used as Federal Constitution and National Principles. This principle part of decision-making process. There are 5 element need to be adhered: full should be adhered to in 3 ways: using inclusive disclosure about the facts that an AI system is being used in decision making; the development techniques; strategically developing tools intended purpose of the AI systems; the training data (describing the data that was used that are sensitive to the specific needs of vulnerable in training, what historical and social biases exist in the data; what procedures were groups in society; and proactively ensuring diversity used to verify data quality); maintenance and assessment of AI systems; and the ability among AI developers and decision-makers. to challenge the decisions made by the AI systems. 6 7 THE PURSUIT OF HUMAN BENEFIT AND HAPPINESS AI systems should respect human-centered values, pursue human benefit for human society, ACCOUNTABILITY enhance the quality of life, and increase human happiness. The developers, owners of AI models, and AI actors AI systems should not be used for malicious purposes in making decisions. Understanding should be accountable for the success or failure of the the way users interact with AI systems and how it is perceived if there are any negative AI solutions. They should take responsibility for outcomes from the adoption of AI can improve happiness and human well-being. ensuring the proper functioning of AI systems in Human oversight helps to ensure that AI systems do not undermine human autonomy through compliance with AI Acts, governances, and ethical governance mechanisms such as human-in-the-loop (HITL), human-on-the-loop (HOTL), and principles. human-in-command (HIC). HITL refers to the capability for human intervention in every While designing AI systems, 4 elements (system decision cycle of the AI systems, while HOTL refers to the capability for human intervention purpose, technology capability, quality and reliability, during the design cycle of the system and monitoring the system’s operation. HIC refers to and sensitive users) need to be considered to avoid the capability to oversee the overall activity of the AI systems and the ability to decide when consequential harm. and how to use the AI system. 27  2.8 HUMAN FRIENDLY AI (HF-AI) PLATFORM Establishing an effective AI communication platform is crucial to ensure total engagement from all stakeholders in enhancing AI development and adoption in Malaysia. The Human-Friendly AI (HF-AI) Platform was introduced to provide immediate 24/7 national support, answering queries, solving problems, and providing information without delay or zero downtime. This HF-AI Platform can offer the following: Operational Efficiency Data-Driven Insights Streamline and automate routine tasks and inquiries, freeing up This platform can collect and analyze interaction data, human employees to focus on more complex and strategic providing valuable insights into customer behavior, tasks. This leads to higher productivity and often reduced preferences, and trends. operational costs.. Personalization at Scale Enhanced Accessibility AI can analyze users’ data and tailor interactions accordingly, The platform can overcome language barriers and provide providing personalized recommendations and support for services in many languages, making information and services each user. accessible to all. Scalability Risk Management The AI platform can easily scale up to handle increased load AI can identify and mitigate risks by monitoring without the need for a proportional increase in human communications for fraud, misinformation, and other resources. anomalies. Innovation and Competitive Edge Implementing cutting-edge AI communication technology can set a company apart from competitors, making it more attractive to users. 28  The existing platforms can be used to support the implementation of HF-AI Platform: Knowledge Resource for Science and Technology Excellence, Malaysia (KRSTE.my) AI untuk Rakyat is an online program aimed at KRSTE.my utilizes SPAF to offer advanced knowledge management raising public awareness about Artificial Intelligence and technology services, facilitating context-based search, informed (AI) among senior citizens, students, professionals, decision-making, and visual content display. and those interested in developing a digital-first (name) mindset, promoting a clear explanation of AI. The programme is divided into two sections: Malaysian Government Central Data Exchange (MyGDX) Section 1 AI Aware: The aim is to raise awareness about Al, and MyGDX facilitates the provision of online services through data includes defining the terms Artificial Intelligence and integration, enhancing coordination and efficiency across agencies. This Human Intelligence and explaining the differences service reduces infrastructure provision costs and system integration, between AI and non-AI machines, and addressing thereby enhancing the overall efficiency of online services. popular misconceptions about AI technology. (name) Section 2 AI Appreciation: This section includes describing the different Outreach Programmes-Malaysia Open Science domains of AI and their impact on various industries, Platform (MOSP) and explaining the principles of responsible AI and The MOSP, a strategic initiative in Malaysia, aims to enhance the STI AI ethics. Collaborative Ecosystems, ensuring research data accessibility and sharing aligned with national priorities. It is connected to Jabatan Digital Negara's Open Government Data portal and MASTIC's Raw Database for Research and Science. 29 30  Section 3 NATIONAL GUIDELINES FOR STAKEHOLDERS 3.1 Introduction: Summary of Part A, Part B and 32 Part C 31  INTRODUCTION: SUMMARY OF PART A, B, AND C AI System as Structure for National Guidelines Development The OECD defines AI systems as machine-based systems that, for explicit or implicit objectives, generate outputs like predictions, content, recommendations, or decisions, varying in autonomy and adaptiveness post-deployment. These national guidelines cover stakeholders within the systems and include policy makers to govern within the system components. Stakeholder scope and national guidelines for stakeholders (End Users, Policy Markers, Developers) are illustrated in Figure 3. While these three stakeholder groups may not be exhaustive, they are necessary and fairly sufficient to establish the fundamental building blocks of an AI system. With this, a clear shared responsibility framework is developed to detail different roles and responsibilities among the key stakeholders. The shared responsibility framework needs to ensure that there is a clear delineation of roles and responsibilities between groups of stakeholders. It is important to note that, in reality, a single entity or person can certainly play more than one role. For example, a specific stakeholder can hold multiple roles at the same time, depending on the particular use of AI. These categories of stakeholders can be mutually exclusive and collectively exhaustive. Each stakeholder can refer to parts A, B, or C according to their interest and requirements. PART A End Users End users are individuals or organizations who use AI products. Different groups of end users use AI products and services in different ways. Some use AI-powered virtual assistants, smart home devices, or AI applications on their smartphones. Others working in different agencies and industries use AI-generated applications: government agencies use AI for security, law enforcement, or public services; manufacturers use it for quality control and to optimize their supply chain; banking institutions rely on AI for fraud detection; and Figure 3: The scope of stakeholder and content creators use AI for video editing or recommendation systems. Since these consumers may or national guidelines covering may not be aware that they are adopting AI technologies, these guidelines may raise their awareness of responsible AI use, inform people about their rights and responsibilities, and empower them to make informed decisions regarding AI technologies and applications. 32 PART B Policy Makers of Government, Agencies, Organizations and Institutions The primary target audiences for the National Guidelines are the policymakers, planners, and managers responsible for AI workforce policy and planning at national and local levels. Government officials, policymakers, and regulatory agencies are individuals or agencies that are responsible for formulating and implementing policies and regulations concerning the use of AI, and for ensuring ethical and Responsible AI practices across different sectors. The National Guidelines provide a framework and practical recommendations to ensure the responsible and ethical use of AI within the jurisdictions of those responsible and guide them in ensuring compliance, consumer protection, and fair competition in AI-related activities. PART C Developers, Designers, Technology Providers and Suppliers Developers and designers are those involved in developing and designing AI products. They come from different fields of expertise, such as computer science, data science, engineering, and AI research, and they work for different AI users and industries. The National Guidelines outline best practices, ethical considerations, and technical standards for Responsible AI development and deployment. The Guidelines also enable these professionals to align their work with recognized benchmarks for better outcomes and to reduce or mitigate any AI ethical risks. 33  “ Despite helping to improve productivity, efficiency, the quality of life, and creativity of humans, modern and dynamic Generative AI technology also has a dark side that society needs to be aware of. Human values, data protection, and the responsible use of technology are all affected and given a new meaning by this technology. “ Therefore, all developers, researchers, data scientists and decision-makers to work together to develop and adopt AI- related technology Responsibly and Ethically. Dato’ Seri Anwar Ibrahim - The Prime Minister of Malaysia 34  PART A AI FOR END USERS A1.0 Introduction 36 A2.0 Responsible AI for The End Users 38 A3.0 What You Can and Cannot do as AI Users 42 A4.0 Aligning Users Contributions with Responsible 43 Image by Storyset AI and Sustainability Image: Storyset.com 35  PART A A1.0 INTRODUCTION 1.1 INTRODUCTION TO THE NATIONAL GUIDELINES FOR AI END USERS The National Guidelines for end users are essential because they mitigate the risk or harm, protect rights and values, build trust and confidence, enhance accountability, and transparency, as well as align and empower end users’ rights and societal values. As an illustration, the f