\n \n \n Image generated with stable diffusion model inspired by Alice in Wonderland\n \n
","metadata":{}},{"cell_type":"markdown","source":"\n# 1. Introduction\n\n\nThis essay is prepared for the 2023 Kaggle AI Report competition, focusing on the topic of **AI ethics**. In this essay, we will delve into **the enigma of AI ethics**, explore **the global perception of AI ethics**, examine **organizational perspectives**, discuss **AI ethics tools**, and highlight **notable initiatives** in the field. By addressing these key areas, our aim is to provide a comprehensive understanding of the advancements, challenges, and ongoing efforts in the field of AI ethics especially in the last two years.\n\nThe landscape of AI ethics, once a mere ripple in the vast ocean of discourse, has swelled into a formidable wave, crashing onto the shores of mainstream consciousness. This once esoteric realm has unfurled, revealing its profound implications on the tapestry of society, and has thus been thrust into the limelight. The world's gaze has shifted, recognizing the intricate dance between AI and ethics, and the necessity of choreographing this dance with careful consideration. This evolution is mirrored in the burgeoning corpus of scholarly articles, a testament to the intellectual curiosity ignited in academia, and the expanding frontiers of knowledge in this field.","metadata":{}},{"cell_type":"markdown","source":"
\n \n \n Scholarly Works over time. Source : Lens\n \n
","metadata":{}},{"cell_type":"markdown","source":"\n# 2. The Enigma of AI Ethics\n\nThe labyrinth of AI Ethics is a research sphere that scrutinizes the ethical and moral repercussions of deploying artificial intelligence. As AI infiltrates various domains from healthcare to finance, education to transportation, it has escalated into a significant quandary.\nThe conundrum of AI Ethics grapples with notions such as **transparency**, **justice**, **non-discrimination**, **accountability**, and **privacy**. For instance, how do we ensure AI systems are transparent, and users comprehend the decision-making process? How do we ascertain that AI systems are equitable and do not marginalize certain societal groups? Who bears the responsibility when AI systems err or inflict damage? How do we safeguard individual privacy when AI systems are increasingly adept at amassing and scrutinizing colossal amounts of personal data?\nAI Ethics endeavors to answer these questions and ensure AI respects human rights and fosters societal well-being. This involves formulating standards and regulations for AI development and use, promoting transparency and accountability in AI, and ensuring AI is used in a fair and non-discriminatory manner.\nHowever, AI Ethics is a complex and perpetually evolving field. With the swift progression of AI technology, new ethical questions and challenges continue to surface. Moreover, AI Ethics must consider a variety of perspectives and cultural values, as AI is deployed in various cultural and social contexts globally.\n\nIn the realm of AI Ethics, one intriguing challenge is the phenomenon of \"_strange errors_\" in AI systems, a concept explored in an article published in Cambridge Quarterly of Healthcare Ethics in 2023[1]. These are not your run-of-the-mill errors, but rather anomalies that baffle our understanding and defy prediction. They are like the rogue waves of the AI ocean, unexpected, unpredictable, and potentially devastating. These anomalies pose a significant challenge to the **trustworthiness** of AI systems, as they can lead to unforeseen and often unmanageable consequences. This underlines the pressing need for robust error detection and management strategies in AI systems, a safety net to catch these rogue waves and ensure ethical use of AI.\n\nAnother dimension of AI Ethics that calls for our scrutiny is the intertwined concept of **trust** and **reliability**. As emphasized in an article published in AI and Ethics in 2022 [2], placing trust in AI systems extends beyond their performance capabilities. It also encompasses their capacity to act in a manner that is congruent with our ethical and moral compass. This involves their potential to make decisions that are not only fair but also transparent and accountable. On the flip side, the reliability of AI systems pertains to their ability to execute tasks with consistency and predictability. These two aspects, trust and reliability, form the bedrock of AI Ethics, serving as the foundational pillars upon which the ethical use of AI is built.\n\nWithin the intricate labyrinth of AI Ethics, a critical notion is that of \"_blind spots_\" in AI ethics, as underscored by Thilo Hagendorff in an article published in AI and Ethics in 2021 [3]. Hagendorff argues that discussions on AI ethics often focus on principles such as **explainability**, **fairness**, and **privacy**, but frequently overlook critical aspects of research, alternative perspectives, and non-ideal approaches. He highlights ethical questions that are seldom or never discussed in AI ethics, such as the negative externalities of AI systems, the strict anthropocentrism of AI ethics, and the environmental impact of AI.\n\nAs we journey deeper into the multifaceted and ever-evolving field of AI Ethics, it becomes clear that continuous research and dialogue are necessary. As we continue to deploy AI across various domains, we must strive to tackle these ethical challenges and ensure that AI serves the best interests of humanity. This brings us to the next important discussion: How is AI Ethics perceived around the world?","metadata":{}},{"cell_type":"markdown","source":"\n# 3. Unraveling the Global Tapestry of AI Ethics: A Spectrum of Views\n\nAs we've traversed through this discourse, it's evident that the ethics of artificial intelligence (AI) has emerged as a topic of escalating resonance in recent years. A constellation of nations are actively participating in these dialogues, molding the standards and regulations that govern the realm of AI ethics. In the global symphony of discourse surrounding AI ethics, the United States emerges as the conductor, leading the ensemble with its technological prowess. Following closely in the orchestration are China and the United Kingdom, each playing their unique parts in this intricate composition. Yet, the melody would be incomplete without the harmonious contributions of other nations. The European Union, Canada, Australia, and India, each add their distinctive notes, enriching the vibrant composition of discussions that shape the landscape of AI ethics.","metadata":{}},{"cell_type":"markdown","source":"
\n \n \n Scholarly Works on AI ethics from January 2021 to July 2023 by country. Source : Lens\n \n
","metadata":{}},{"cell_type":"markdown","source":"To comprehend the global perception of AI Ethics, we must dive into the realm of AI's role in government services. AI technology has emerged as a revolutionary force reshaping government services across the globe, as underscored in a systematic literature review published in the Open Access Indonesia Journal of Social Sciences in 2023 [4]. This metamorphosis opens up new avenues for governments to amplify their effectiveness, efficiency, and responsiveness in delivering services to their citizens. However, it simultaneously ushers in new ethical conundrums such as data privacy, ethics, and public trust, which demand careful navigation. Therefore, the significance of AI Ethics is illuminated in ensuring that the integration of AI in government services is in harmony with our ethical and moral principles.\n\nDifferent countries have adopted diverse approaches to AI Ethics, reflecting their unique cultural, social, legal, and regulatory frameworks. For instance, in 2021, the European Union proposed the **AI Act**, a comprehensive legal framework that addresses the risks associated with AI and positions Europe as a global leader in this realm. The AI Act categorizes AI use into four levels of risk: unacceptable risk, high risk, limited risk, and minimal risk. Its primary objective is to foster trust in AI systems, ensuring their reliability and key role in enhancing Europe's global competitiveness [5].\n\nIn the United States, the White House Office outlined a \"**_Blueprint for an AI Bill of Rights_**\" in 2022 [7]. This blueprint incorporates five principles that guide the design, use, and deployment of automated systems. These principles encompass creating safe and effective systems, safeguarding against algorithmic discrimination, ensuring data privacy, providing notice and explanation, and offering human alternatives, consideration, and fallback options.\n\nIn a similar vein, Canada has unfurled the **AI and Data Act**, a legislative move aimed at bolstering the nation's private sector privacy law, laying down guidelines for responsible development and deployment of AI, and propelling the implementation of Canada's Digital Charter. The Act is a beacon guiding the use of AI in a manner that respects human rights, upholds ethical values, and fosters societal well-being.\n\nThese contrasting strategies underscore the rich tapestry of perspectives on AI Ethics across the globe, highlighting the critical need for international cooperation in carving out common standards and regulations for AI Ethics.\n\nIn this scenario, the Ethical, Legal, and Social Implications (**ELSI**) framework emerges as a crucial actor. Originally crafted for the Human Genome Project, this framework has widened its scope to envelop a multitude of fields, including the sphere of artificial intelligence. Its paramount mission is to predict and wrestle with the ethical, legal, and social obstacles that sprout alongside the advancement of science and technology.\n**Privacy**, **consent**, **fairness in access and use of technologies**, **governance**, **regulation**, and **impacts on social values and norms** are all facets contemplated within the ELSI framework. By adopting a forward-thinking approach, the ELSI framework not only identifies and manages potential risks but also steers technology development to maximize benefits while mitigating adverse impacts. A 2022 study by Hagendorff [6] offers an in-depth dissection of ELSI segmentation related to AI in Japan, the United States, and Germany.\n\nMoreover, in a 2023 article titled \"_Honor Ethics: The Challenge of Globalizing Value Alignment in AI_\" the authors emphasize the need to incorporate **non-WEIRD** (Western, Educated, Industrialized, Rich, Democratic) perspectives in AI ethics [8]. They introduce the concept of honor as a means to diversify the discussion on AI ethics beyond the normative values of WEIRD societies. The authors argue that integrating **honor** into the discourse on value alignment for AI can elucidate the complexity of various moral systems and underscore the necessity for a truly global AI ethics framework that considers non-WEIRD perspectives and acknowledges deeply rooted values and principles outside of liberalized Western societies.\n\nWhile gaining insight into the global perspective on AI Ethics provides a broad overview of ethical considerations and regulatory frameworks across countries, it is equally important to delve into the organizational perspective. This perspective focuses on how individual organizations, be they private corporations or non-profit entities, approach and implement AI Ethics within their operations. The subsequent section will explore organizational perspectives, examining how these entities navigate the complex ethical landscape of AI, develop their ethical guidelines, and implement these principles in their AI systems and processes. This granular understanding of the practical application of AI Ethics in real-world settings will shed light on the multifaceted nature of ethical considerations in AI.\n","metadata":{}},{"cell_type":"markdown","source":"\n# 4. Organizational Perspectives: Navigating the Ethical Landscape\n\nProminent organizations such as **UNESCO**, **IEEE**, and **WHO** have expressed their perspectives on AI Ethics, recognizing its significance and endeavoring to establish standards and guidelines for ethical AI development and usage.\n\nFor instance, **UNESCO** produced the \"_Recommendation on the Ethics of Artificial Intelligence_\" in 2021, the first-ever global standard on AI ethics. This framework, adopted by all 193 Member States, centers on protecting **human rights** and **dignity**. It advances fundamental principles such as **transparency**, **fairness**, and the crucial **role of human oversight in AI systems**. The Recommendation's extensive Policy Action Areas facilitate policymakers in translating core values and principles into actionable steps across various domains, including **data governance**, **environment and ecosystems**, **gender**, **education and research**, and **health and social well-being** [9].","metadata":{}},{"cell_type":"markdown","source":"
\n \n \n Key policy area defined by UNESCO. Source : UNESCO\n \n
","metadata":{}},{"cell_type":"markdown","source":"In 2023, **IEEE** launched a program known as the IEEE Global Initiative to develop global socio-technical standards in AI Ethics and Governance. This program includes a series of ongoing standards projects, including the IEEE P7000™ series, which addresses specific issues at the intersection of technology and ethics. The mission of the IEEE Global Initiative is to equip all participants engaged in the creation and development of autonomous and intelligent systems with the necessary knowledge, training, and tools to place ethical considerations at the forefront, thereby propelling these technologies to serve the best interests of humanity [10].\n\nRecognizing the vital role of AI Ethics in the field of health, the **World Health Organization** published guidance on \"_Ethics and Governance of Artificial Intelligence for Health_\" in 2021. This guidance highlights AI's potential to improve diagnosis, treatment, health research, and public health functions. However, it emphasizes the need for ethics and human rights to be at the core of AI design, deployment, and use. The report identifies ethical challenges and risks associated with AI in health and proposes consensus principles to ensure that AI benefits all countries while being governed accountably and responsively [11].\n\nThese initiatives by UNESCO, IEEE, and WHO underscore the importance of ethical considerations in AI and the need for **transparency**, **fairness**, and **human oversight**. However, acknowledging these principles and translating them into practice are two distinct challenges. This realization leads us to the next phase of our discussion: the tools employed to effectively integrate these ethical considerations into AI systems. In the subsequent segment, we will delve into an array of tools and methodologies employed to bring AI ethics into action, spanning from AI auditing frameworks to ethical AI design principles and practices. These instrumentalities serve a pivotal role in connecting the dots between theoretical tenets and their real-world implementation, guaranteeing that AI systems are not only technically robust but also ethically accountable.","metadata":{}},{"cell_type":"markdown","source":"\n# 5. Operationalizing AI Ethics: Tools for Ethical Deployment\n\nAI Ethics is a rapidly evolving field, with numerous tools and methodologies developed to facilitate its implementation. These tools aim to ensure that AI systems are developed and used ethically, helping organizations navigate the complex landscape of AI ethics and enabling the practical implementation of ethical principles in AI development and utilization.\n\nIn a paper titled \"_Actionable Approaches to Promote Ethical AI in Libraries_\" in 2021 [12], the increasing use of AI in libraries is discussed, along with the ethical challenges it poses. The paper emphasizes the need for careful development, implementation, and maintenance of AI systems in libraries to address ethical issues encompassing **data**, **design**, **deployment**, **bias**, **fairness**, **accountability**, **transparency**, **responsibility**, **environmental impact**, and more. The paper suggests actionable approaches for promoting ethical AI, drawing from other contexts and adapting them to AI-powered information services throughout different stages of the software lifecycle.\n\nAnother paper, \"_Seeing Like a Toolkit: How Toolkits Envision the Work of AI Ethics_\" in 2023 [13], examines how AI ethics tools embody values and assumptions regarding the practice of ethics. By analyzing a corpus of 27 AI ethics tools, the researchers identify discourses on ethics, intended users, and work practices envisioned by these tools. They find that AI ethics tools predominantly frame ethics as technical work for individual practitioners, despite calling for involvement of a broader set of stakeholders to address social aspects of AI ethics. The tools tend to overlook organizational, labor, and political implications related to AI ethics in practice, revealing a disconnect between the stakeholders and work practices envisioned by the tools.\n\nAI ethics tools can be categorized into several distinct groups, each addressing specific aspects of AI ethics:\n\n+ **Tools for data privacy and secure model training**: **TensorFlow Privacy** [14] and **TensorFlow Federated** [15] are examples of tools enabling the training of machine learning models while preserving data privacy. These tools ensure the protection of sensitive individual information during model training, thus upholding privacy rights and building public trust in AI systems.\n\n+ **Tools for fairness and non-discrimination**: **AI Fairness 360** [16] and **Fairlearn** [17] are tools designed to identify and mitigate bias in machine learning models. They play a crucial role in ensuring that AI systems treat individuals fairly and do not discriminate based on protected characteristics such as race, gender, or age.\n\n+ **Tools for model transparency and explainability**: **Model Card Toolkit** (MCT) [18], **DALEX** [19], and **XAI** [20] are tools aimed at understanding the behavior of machine learning models. They contribute to the transparency of AI systems by facilitating explanations of their decision-making processes, particularly in sensitive contexts like healthcare and justice. The significance of Explainable AI (XAI) has been highlighted in AI ethics, as it refers to AI systems that can be interpreted and understood by humans. However, the field of AI ethics and XAI lacks a unified framework and conceptualization, calling for further exploration [21].\n\n+ **Tools for data validation**: **TensorFlow Data Validation** (TFDV) [22] is a tool that analyzes and validates data used for machine learning. This ensures the quality and representativeness of the training data, minimizing errors and biases in the model's predictions.\n\n+ **Tools for protection against unauthorized tracking**: **Fawkes** [23] is a tool that helps protect individuals from unauthorized tracking by altering their publicly accessible photos. This contributes to safeguarding individuals' privacy rights in an era where data collection and usage occur without explicit consent.\n\n+ **Tools for data augmentation and training**: **TextAttack** [24] is an example of a tool that augments data for machine learning and aids in training models. These tools improve model performance, enhance reliability, and ensure robustness to diverse inputs. Data augmentation also helps improve the representativeness of training data, reducing biases in model predictions.\n\nThese tools and ongoing research demonstrate the increasing importance of AI ethics and Explainable AI in the development and use of AI systems.\n\nIn a paper titled \"_Ethics as a Service: A Pragmatic Operationalization of AI Ethics_\" published in 2021 [25], authors explore the practical utility of AI ethics for AI practitioners. They highlight the limitations of existing translation tools and methods in effectively operationalizing AI ethics, noting their susceptibility to ethics washing or being too strict and context-unresponsive. To overcome these limitations, the authors propose the concept of \"**_Ethics as a Service_**\" which aims to provide a theoretical foundation for making AI ethics practical and useful for AI practitioners.\n\nAs we wrap up our journey through the tools and methodologies that enable the ethical application of AI, it becomes clear that the practical execution of ethical principles in AI development and use is a multifaceted yet vital endeavor. The array of tools we've examined, from those safeguarding data privacy and secure model training to those advocating for fairness, non-discrimination, and transparency, all contribute to the ethical panorama of AI. However, the ethical application of AI is not merely a technical hurdle. It's also a human challenge, necessitating the active engagement and dedication of diverse individuals and groups.\n\nIn the ensuing section, we will explore the initiatives that have been set in motion to cultivate diversity and inclusion in the realm of AI ethics. These initiatives acknowledge the priceless contributions of underrepresented groups, particularly women, in molding the future of AI. They also underscore the significance of an interdisciplinary approach to AI ethics, which is crucial for grasping the comprehensive impact of ethical technologies. As we segue into this discussion, let's bear in mind that the ethical application of AI is not just about the right tools, but also about the right people and the right approach.","metadata":{}},{"cell_type":"markdown","source":"\n# 6. Initiatives in AI Ethics: Nurturing Diversity and Inclusion\n\nAcknowledging the critical role of diversity and inclusion in the tech landscape, the Women in AI Ethics™ (WAIE) initiative was set in motion to recognize and amplify the work of women in the realm of AI ethics. In 2023, the initiative unveiled a list of 100 remarkable and inspiring women who have made noteworthy contributions to this field [26]. This initiative underscores the influence of diversity in sculpting the future of AI and brings to light the invaluable contributions of women across various disciplines such as law, human rights advocacy, anthropology, and research. It nurtures an interdisciplinary approach to AI ethics, encapsulating the full impact of ethical technologies and enabling lesser-known voices to rise above the tech industry's cult of personality.\n\n\nAdditionally, UNESCO launched the Women 4 Ethical AI platform to address the reproduction and amplification of gender inequalities in AI development and deployment [27]. This platform aims to promote the participation and safety of women and girls online, particularly in response to the rise of tech-enabled violence, gender-based misinformation, and fake news.\n\nDespite these pioneering efforts, there remains a mountain of work to be done to address diversity and inclusion challenges in the tech industry. Active engagement and commitment from decision-makers are indispensable to champion diversity and inclusion. Tangible steps are needed to dismantle barriers to entry in the male-dominated AI/tech arena and to prioritize the cultivation of technical skills.\n\nIn conclusion, initiatives like Women in AI Ethics™ and Women 4 Ethical AI play vital roles in fostering diversity and inclusion in the field of AI ethics. They emphasize the important contributions of women and advocate for an interdisciplinary approach to address AI ethics challenges. However, ongoing efforts are needed to promote diversity, inclusion, and ethical responsibility, ensuring that AI systems are developed and used in ways that align with ethical principles and serve the best interests of humanity.","metadata":{}},{"cell_type":"markdown","source":"\n# 7. Conclusion\n\nIn conclusion, the ethics of artificial intelligence (AI) pose a complex and multi-dimensional puzzle. Organizational viewpoints, as expressed by UNESCO, IEEE, and WHO, highlight the importance of formulating standards and guidelines for the ethical creation and implementation of AI. These initiatives emphasize **transparency**, **fairness**, and **human supervision** as fundamental principles to guide the deployment of AI.\n\nInstruments and methodologies to operationalize AI ethics also play a pivotal role. They aid organizations in traversing the intricate ethical terrain of AI and implementing ethical principles in the creation and utilization of AI systems. Tools for **data protection**, **fairness**, **explainability**, and validation contribute to ensuring that AI systems respect human rights, evade discriminatory biases, and function in a transparent and accountable manner.\n\nMoreover, initiatives such as Women in AI Ethics™ and Women 4 Ethical AI cast a spotlight on the critical role of **diversity** and **inclusion** within the sphere of AI ethics. They acknowledge the indispensable contributions of women in the advocacy for ethical AI and stress the necessity for an **interdisciplinary** approach to thoroughly address ethical quandaries in AI.\n\nNonetheless, there remains an extensive amount of work to be done to navigate the ethical challenges of AI. It is vital for stakeholders to actively pledge to endorse **diversity**, **inclusion**, and **equity** within the technology industry. Lists and initiatives are indeed a commendable initial stride, but concrete actions must be undertaken to break down barriers to entry in a sector that is predominantly male-dominated and to foster a culture steeped in ethics and responsibility.\n\nTo truly guarantee that AI serves the welfare of humanity, it is essential to persist in promoting **comprehensive research**, **fostering open dialogues**, and encouraging **international cooperation** in the realm of AI ethics. Only a comprehensive approach based on **robust ethical principles**, **diverse perspectives**, and adequate **regulation** will enable us to address the complex ethical challenges posed by AI and ensure its responsible and beneficial use for society at large.\n\nUltimately, by weaving ethical principles into the development and use of AI, we can sculpt a future where AI is a potent tool that respects **human values**, contributing to a more **equitable**, **inclusive**, and **prosperous society**.","metadata":{}},{"cell_type":"markdown","source":"\n# 8. References\n\n## Written material\n* [1] [\"Learning to Live with Strange Error: Beyond Trustworthiness in Artificial Intelligence Ethics\"](https://doi.org/10.1017/S0963180122000688), Cambridge Quarterly of Healthcare Ethics, 2023.\n* [2] [\"Trust and Reliability in Artificial Intelligence\"](https://doi.org/10.1007/s43681-022-00200-5), AI and Ethics, 2022.\n* [3] [\"Blind Spots in Artificial Intelligence\"](https://link.springer.com/article/10.1007/s43681-021-00122-8), AI and Ethics, 2023.\n* [4] [\"The Role of Artificial Intelligence in Government Services: A Systematic Literature Review\"](https://doi.org/10.37275/oaijss.v6i3.163), Open Access Indonesia Journal of Social Sciences, 2023.\n* [5] [\"Proposal for a Regulation laying down harmonised rules on artificial intelligence\"](https://digital-strategy.ec.europa.eu/en/library/proposal-regulation-laying-down-harmonised-rules-artificial-intelligence), European Commission, 2021.\n* [6] [\"Segmentation of ethics, legal, and social issues (ELSI) related to AI in Japan, the United States, and Germany\"](https://link.springer.com/article/10.1007/s43681-022-00207-y), AI & Ethics, 2022.\n* [7] [\"Blueprint for an AI Bill of Rights | OSTP | The White House\"](https://www.whitehouse.gov/ostp/ai-bill-of-rights/), The White House Office of Science and Technology Policy, 2023.\n* [8] [\"Honor Ethics: The Challenge of Globalizing Value Alignment in AI\"](https://doi.org/10.1145/3593013.3594026), ACM Digital Library, 2023.\n* [9] [\"Ethics of Artificial Intelligence\"](https://www.unesco.org/en/artificial-intelligence/recommendation-ethics), UNESCO, 2021.\n* [10] [\"The IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems\"](https://standards.ieee.org/news/get-program-ai-ethics/), IEEE Standards Association, 2023.\n* [11] [\"Ethics and governance of artificial intelligence for health\"](https://www.who.int/publications/i/item/9789240029200), WHO, 2021.\n* [12] [\"Actionable Approaches to Promote Ethical AI in Libraries\"](https://doi.org/10.48550/arXiv.2109.09672), 84th Annual Meeting of the Association for Information Science and Technology, Volume 58, 2021.\n* [13] [\"Seeing Like a Toolkit: How Toolkits Envision the Work of AI Ethics\"](https://doi.org/10.1145/3579621), Proceedings of the ACM on Human-Computer Interaction, Volume 7, 2023.\n\n* [21] [The Role of Explainable AI in the Research Field of AI Ethics](https://doi.org/10.1145/3599974), ACM Transactions on Interactive Intelligent Systems, 2023.\n\n* [25] [Ethics as a Service: A Pragmatic Operationalisation of AI Ethics](https://doi.org/10.1007/s11023-021-09563-w), Minds and Machines, 2021.\n* [26] [Women in AI Ethics](https://womeninaiethics.org/), 2023. \n* [27] [\"UNESCO launches Women 4 Ethical AI Platform\"](https://www.unesco.org/en/articles/unesco-launches-women-4-ethical-ai-platform), 2023.\n\n## Tools \n* [14] [TensorFlow Privacy](https://github.com/tensorflow/privacy), Python library that provides implementations of TensorFlow optimizers for training machine learning models with differential privacy. The library includes tutorials and analysis tools for computing the privacy guarantees provided. \n* [15] [TensorFlow Federated](https://github.com/tensorflow/federated), open-source framework for machine learning and other computations on decentralized data.\n* [16] [AI Fairness 360](https://github.com/Trusted-AI/AIF360), open-source library containing techniques developed by the research community to help detect and mitigate bias in machine learning models throughout the AI application lifecycle.\n* [17] [Fairlearn](https://fairlearn.org/), open-source, community-driven project designed to help data scientists improve the fairness of AI systems.\n* [18] [Model Card Toolkit](https://github.com/tensorflow/model-card-toolkit), open-source project that streamlines and automates the generation of Model Cards, documents that provide context and transparency into a model's development and performance. \n* [19] [DALEX](https://github.com/ModelOriented/DALEX), open-source project that provides a collection of tools for the exploration and explanation of complex machine learning models. \n* [20] [XAI](https://github.com/EthicalML/xai), toolbox for machine learning providing various tools for the analysis and evaluation of data and models.\n* [22] [TensorFlow Data Validation](https://github.com/tensorflow/data-validation), open-source library developed by TensorFlow that helps developers understand, validate, and monitor their machine learning data at scale.\n* [23] [Fawkes](https://sandlab.cs.uchicago.edu/fawkes/), privacy-enhancing technology that provides individuals with a tool to protect their privacy against unauthorized and malicious uses of their photos. \n* [24] [TextAttack](https://github.com/QData/TextAttack), Python framework for adversarial attacks, data augmentation, and adversarial training in NLP.","metadata":{}},{"cell_type":"markdown","source":"# Tools used\nThe following tools were used to help translate from French to English, reformulate sentences and correct mistakes :\n+ GPT-3.5\n+ DeepL","metadata":{}}]}
![](\"https://i.imgur.com/beI3wjZ.png\")
\n ![](\"https://i.imgur.com/Nkt8DgY.png\")
\n ![](\"https://i.imgur.com/DCJcLNj.png\")
\n ![](\"https://i.imgur.com/DV4uHkM.png\")
\n