Search

Over the last two years, The Alan Turing Institute and the Information Commissioner’sOffice (ICO) have been working together todiscover ways to tackle thedifficult issues surrounding explainable AI. Theultimate product of this joint endeavour, Explaining decisions made with AI, published in May2020, is the most comprehensive practical guidance on AI explanation produced anywhere todate. We have put together this workbook to help support the uptake of that guidance.The goal of the workbook is to summarise some of mainthemes from Explaining decisionsmade with AI and then to provide the materials for a workshop exercise that has been builtaround a use case created to help you gain a flavour of how to put the guidance into practice.In the first three sections, we run through the basics of Explaining decisions made with AI. Weprovide a precis of the four principles of AI explainability, the typology of AI explanations,and the tasks involved in the explanation-aware design, development, and use of AI/MLsystems. We then provide some reflection questions, which are intended to be a launchingpad for group discussion, and a starting point for the case-study-based exercise that we haveincluded as Appendix B. In Appendix A, we go into more detailed suggestions about how toorganise the workshop. These recommendations are based on two workshops we had theprivilege of co-hosting with our colleagues from the ICO and Manchester MetropolitanUniversity in January 2021. The participants of these workshops came from both the privateand public sectors, and we are extremely grateful to them for their energy, enthusiasm, andtremendous insight. This workbook would simply not exist without the commitment andkeenness of all our collaborators and workshop participants.

Over the last two years, The Alan Turing Institute and the Information Commissioner’sOffice (ICO) have been working together todiscover ways to tackle thedifficult issues surrounding explainable AI. Theultimate product of this joint endeavour, Explaining decisions made with AI, published in May2020, is the most comprehensive practical guidance on AI explanation produced anywhere todate. We have put together this workbook to help support the uptake of that guidance.The goal of the workbook is to summarise some of mainthemes from Explaining decisionsmade with AI and then to provide the materials for a workshop exercise that has been builtaround a use case created to help you gain a flavour of how to put the guidance into practice.In the first three sections, we run through the basics of Explaining decisions made with AI. Weprovide a precis of the four principles of AI explainability, the typology of AI explanations,and the tasks involved in the explanation-aware design, development, and use of AI/MLsystems. We then provide some reflection questions, which are intended to be a launchingpad for group discussion, and a starting point for the case-study-based exercise that we haveincluded as Appendix B. In Appendix A, we go into more detailed suggestions about how toorganise the workshop. These recommendations are based on two workshops we had theprivilege of co-hosting with our colleagues from the ICO and Manchester MetropolitanUniversity in January 2021. The participants of these workshops came from both the privateand public sectors, and we are extremely grateful to them for their energy, enthusiasm, andtremendous insight. This workbook would simply not exist without the commitment andkeenness of all our collaborators and workshop participants.

Artificial Intelligence (AI) is making significant progress in recent times and is gaining a strong foothold in business. Currently, there is no generally accepted scholarly framework for the governance of AI-based information technologies within corporate environments. Boards of directors who have the responsibility of overseeing corporate operations need to know how best to govern AI technologies within their companies. In response, this dissertation aims to understand the key elements that can assist boards in the governance of AI-based information technologies. Further, it attempts to understand how AI governance elements dynamically interact within a holistic system. As AI governance is a novel phenomenon, an exploratory investigation was conducted via a qualitative approach. Specifically, the study adopted a grounded theory methodology, within the constructivist paradigm, with the intent of generating theory instead of validating existing theory. Data collection included in-depth interviews with key experts in AI research, development, management, and governance processes in corporate and academic settings. Data were further supplemented with data received from conference presentations given by AI experts. Findings from this dissertation elicited a theoretical model of AI governance that shows various AI governance areas and constituting elements, their dynamic interaction, as well as the impact of these elements in enhancing the organizational performance of AI-based projects and reducing the risks associated with those projects. This dissertation provides a scholarly contribution by comparing governance elements within the IT governance domain and the new AI governance domain. In addition to theoretical contributions, this study provides practical contributions for the benefit of the boards of directors. These include a holistic AI governance framework that pictorially represents twenty-two AI governance elements that boards can use to build their own custom AI governance frameworks. In addition, recommendations are provided to assist boards in starting or enhancing their AI governance journeys. Thesis Doctor of Philosophy (PhD) Artificial Intelligence (AI) refers to a set of technologies that seek to perform cognitive functions associated with human minds, such as learning, planning, and problem-solving. AI brings abundant opportunities as well as substantial risks. Major companies are trying to figure out how best to benefit from AI technologies. Boards of directors, with the responsibility of overseeing company operations, need to know how best to govern such technologies. In response, this study was conducted to uncover key AI governance elements that can assist boards in the governance of AI. Data were collected through in-depth interviews with AI experts and by attending AI conference presentations. Findings yield a theoretical model of AI governance that can assist scholars in enhancing their understanding of this emerging governance area. Findings also provide a holistic framework of AI governance that boards can use as a practical tool to enhance their effectiveness of the AI governance process.

The steadily increasing capabilities in AI systems can have tremendous beneficial impacts on society. However, it is important to simultaneously tackle possible risks that these developments are accompanied by. Therefore, the relatively young field of AI safety has gained international relevance. In parallel, popular media were commenting on whether society should ascribe motifs such as fear or enthusiasm to AI. However, in order to assess the landscape of AI risks and opportunities, it is instead first and foremost of relevance not to be afraid, not to be enthusiastic, but to understand as similarly suggested by Spinoza in the 17th century. In this vein, in this thesis, a transdisciplinary examination is performed to understand how to address possible instantiations of AI risks with the aid of scientifically grounded hybrid cognitive-affective strategies. The identified strategies are “hybrid" due to the fact that AI systems cannot be analyzed in isolation and the nature of human entities as well as the properties of human-machine interactions have to be taken into account within a socio-technological framework and not only addressing unintentional failures but also intentional malice. Consequently, the attribute “cognitive-affective" refers to the inherently affective nature of human cognition. We consider two disjunct sets of systems: Type I and Type II systems. Type II systems are systems that are able to consciously create and understand explanatory knowledge. Conversely, Type I systems are all systems that do not exhibit this ability. All current AIs are of Type I. However, even if Type II AI is non-existent nowadays, its implementation is not physically impossible. Overall, we identify the following non-exhaustive set of 10 tailored hybrid cognitive-affective strategical clusters for AI safety 1) international (meta-)goals, 2) transdisciplinary Type I/II AI safety and related education, 3) socio-technological feedback-loop, 4) integration of affective, dyadic and social information, 5) security measures and ethical adversarial examples research, 6) virtual reality frameworks, 7) orthogonality-based disentanglement of responsibilities, 8) augmented utilitarianism and ethical goal functions, 9) AI self-awareness and 10) artificial creativity augmentation research. In the thesis, we also introduce the so-called AI safety paradox stating, figuratively speaking, that value alignment and control represent conjugate requirements. In theory, with a Type II AI, a mutual value alignment might be achievable via a co- construction of novel values, however, at the cost of its predictability. Conversely, it is possible to build Type I AI systems that are controllable and predictable, but they would not exhibit a sufficient understanding of human morality. Nevertheless, AI safety can be addressed by a cybersecurity oriented and risk-centered approach reformulating AI safety as a discipline which proactively addresses AI risks and reactively responds to occurring instantiations of AI risks. In a nutshell, future AI safety requires transdisciplinarily conceived and scientifically grounded dynamics combining proactive error-prediction and reactive error-correction within a socio-technological feedback-loop together with the cognizance that it is first of relevance not to be afraid, not to be enthusiastic, but to understand – that the price of security is eternal creativity.

Digital platforms, including social media, are not just reforming the economy, but their extreme sociopolitical upheavals, ex. Cambridge Analytica and similar cases, highlight a seriously transformative phenomenon which needs to be confronted.The research assembles a unique set of empirical evidence from Iran, China, and the European Union. These are about the political strategy adopted to take control of those platforms and to ensure the governance maintenance of the status quo.

This conceptual paper addresses the issues of transparency as linked to artificial intelligence (AI) from socio-legal and computer scientific perspectives. Firstly, we discuss the conceptual distinction between transparency in AI and algorithmic transparency, and argue for the wider concept ‘in AI’, as a partly contested albeit useful notion in relation to transparency. Secondly, we show that transparency as a general concept is multifaceted, and of widespread theoretical use in multiple disciplines over time, particularly since the 1990s. Still, it has had a resurgence in contemporary notions of AI governance, such as in the multitude of recently published ethics guidelines on AI. Thirdly, we discuss and show the relevance of the fact that transparency expresses a conceptual metaphor of more general significance, linked to knowing, bringing positive connotations that may have normative effects to regulatory debates. Finally, we draw a possible categorisation of aspects related to transparency in AI, or what we interchangeably call AI transparency, and argue for the need of developing a multidisciplinary understanding, in order to contribute to the governance of AI as applied on markets and in society.

Summary The European Union (EU) has a strong reputation and track record for the development of guidelines for the ethical use of artificial intelligence (AI) generally. In this paper, we discuss the development of an AI and ethical framework by the European Insurance and Occupational Pensions Authority (EIOPA), for the European insurance market. EIOPA's earlier report on big data analytics (EIOPA, 2019) provided a foundation to analyze the complex range of issues associated with AI being deployed in insurance, such as behavioral insurance, parametric products, novel pricing and risk assessment algorithms, e-service, and claims management. The paper presents an overview of AI in insurance applications throughout the insurance value chain. A general discussion of ethics, AI, and insurance is provided, and a new hierarchical model is presented that describes insurance as a complex system that can be analyzed by taking a layered, multi-level approach that maps ethical issues directly to specific level(s)., Highlights • A socio-ethical framing of BDA, AI, and ML applications in the insurance pipeline • Analysis of AI in insurance operations regarding customer behavioral data, The bigger picture Big data analytics and increasingly sophisticated forms of artificial intelligence (AI) and machine learning applications are now part of the catalog of tools and methods that inform the insurance industry's back-end and front-end operations. Insurance products and services are important to support economic and societal development. From commercial property insurance, vehicle insurance, and personal health insurance, availability and access to affordable insurance products are key to society and citizens. That said, insurance products and services can only be accessible if a sustainable balance is achieved between the pricing of policies, operational costs, and claims. AI and machine learning have already proved valuable to the insurance industry in sustaining this balance. However, with the increased use of AI and machine learning, new risks and ethical questions are introduced to the insurance pipeline. Accordingly, there is a need to investigate the ethical contexts and questions of data use in insurance innovation and highlight the emerging tensions and concerns where possible. The paper combines the expertise of insurance experts and an AI ethics expert to assess and better understand some of the ethical ground relating to the tensions concerning how the insurance industry uses big data analytics, AI, and machine learning methods to sustain its commercial operations., The research aims to provide a deep dive into the increasing ethical questions and contexts relating to the insurance industry's sophisticated use of big data analytics, AI, and machine learning to sustain and develop its products and services. While the commercial opportunities are clear, there are questions of what costs the commercial benefits have in terms of how data are used and if the use leads to further tensions regarding privacy, surveillance, and profiling.

I bokens tredje kapitel analyserar Stefan Larsson etiska riktlinjer som styrmedel för utveckling och användning av artificiell intelligens (ai). Författaren pekar på riktlinjernas samspel med rättsliga styrmedel och undersöker vad det är med AI-utvecklingen som särart som lett till att etiska frågor fått en så framskjutande plats med en sådan bred flora av riktlinjer. Genom att anlägga ett europeiskt perspektiv bidrar kapitlet till den pågående globala, såväl som nationella, diskussionen om styrningen av ai som ofta benämns ”AI governance”. Framför allt behandlar kapitlet de etiska riktlinjerna för s.k. tillförlitlig AI som EU-kommissionens AI-expertgrupp publicerat i april 2019 och som ingår i kommissionens AI-strategiska arbete.Med stöd i samhällsvetenskaplig teknikforskning analyserar Larsson särskilt den svårfångade definitionen av AI, och argumenterar för behovet av att se teknologierna i dess tillämpade kontext och i dess samspel med mänskliga värderingar, uttryck och samhällsstrukturer. Det är inte minst viktigt då AI och maskininlärningen är beroende av stora mängder data eller exempel att utgå från. I relation till rättslig styrning medför detta, enligt Larsson, ett särskilt stort behov av flervetenskaplig kompetens vid både forskningen och utvecklingen av AI. Larsson ser det som synnerligen välfunnet att den europeiska diskussionen om hur AI ska styras kommit att präglas så mycket av anspråket att tillgodose en tillförlitlig AI-utveckling i samklang med samhällets värderingar och normer i övrigt.Avslutningsvis konstaterar Larsson att det finns ett behov inom AU av att gå från framtagandet av principiella ställningstaganden till att utveckla processuella tillvägagångssätt inom styrningen av AI, både i relation till etiska riktlinjer såväl som vid rättslig styrning av AI.

AI Blindspots are oversights in a team’s workflow that can generate harmful unintended consequences. They can arise from our unconscious biases or structural inequalities embedded in society. Blindspots can occur at any point before, during, or after the development of a model. The consequences of blindspots are challenging to foresee, but they tend to have adverse effects on historically marginalized communities.

As technologies become more and more pervasive in our everyday life new questions arise, for example, about security, accountability, fairness and ethics. These concerns are about all the realities that are involved or committed in designing, implementing, deploying and using the technology. This document addresses such concerns by presenting a set of practical obligations and recommendations for the development of applications and systems based on Artificial Intelligence (AI) techniques. These are derived from a definition of rights resulting from principles and ethical values rooted in the foundational charters of our social organization., BioLaw Journal - Rivista di BioDiritto, N° 3 (2019)

Following on from the publication of itsFeasibility Studyin December 2020, the Council of Europe’s Ad Hoc Committee on Artificial Intelligence (and its subgroups) initiated efforts to formulate and draft itsPossible elements of a legal framework on artificial intelligence, based on the Council of Europe’s standards on human rights, democracy, and the rule of law. This document was ultimately adopted by the CAHAI plenary in December 2021. To support this effort, The Alan Turing Institute undertook a programme of research that explored the governance processes and practical tools needed to operationalise the integration of human right due diligence with the assurance of trustworthy AI innovation practices. The resulting output, Human Rights, Democracy, and the Rule of Law Assurance Framework for AI Systems: A proposal,was completed and submitted to the Council of Europe in September 2021. It presents an end-to-end approach to the assurance of AI project lifecycles that integrates context-based risk analysis and appropriate stakeholder engagement with comprehensive impact assessment, and transparent risk management, impact mitigation, and innovation assurance practices. Taken together, these interlocking processes constitute aHuman Rights, Democracy and the Rule of Law Assurance Framework(HUDERAF). The HUDERAF combines the procedural requirements for principles-based human rights due diligence with the governance mechanisms needed to set up technical and socio-technical guardrails for responsible and trustworthy AI innovation practices. Its purpose is to provide an accessible and user-friendly set of mechanisms for facilitating compliance with a binding legal framework on artificial intelligence, based on the Council of Europe’s standards on human rights, democracy, and the rule of law, and to ensure that AI innovation projects are carried out with appropriate levels of public accountability, transparency, and democratic governance.