Preliminary report on the first draft of the Recommendation on the Ethics of Artificial Intelligence

Preliminary report on the first draft of the Recommendation on the Ethics of Artificial Intelligence

circular letter

Corporate author

  • UNESCO. Director-General, 2017- (Azoulay, A.)

Document code

  • CL/4327

Collation

  • 33 pages

Language

  • English

Also available in

Year of publication

  • 2020

Unrted Nations Educational, Scientific and Cultural Organlzatlon Réf. : CL/4327 Subject: Preliminary Report on thé first draft of thé Recommendation on thé Ethics ofArtificial Intelligence Dear Sir/Madam, At its 40th session in November 2019, UNESCO's Général Conférence decided to prépare an international standard-setting instrument on thé ethics ofartificial intelligence, in thé form ofa recommendation. This is an essentiel engagement for our Organization, on a topic that not only constitutes a technological and économie révolution, but also a major anthropological disruption. Since thé last Général Conférence, UNESCO has worked actively on this issue, by establishing a group of experts and launching wide-ranging consultations at ail levels. Thèse efforts hâve resulted in this Preliminary Report and draft Recommendation, which you will find enclosed herewith. In accordance with our Constitution and thé Rules of Procédure concerning recommendations, Member States are invited to forward their comments and observations on this text to thé UNESCO Secrétariat at least ten months before thé opening of thé 41st session of thé Général Conférence. l would therefore be grateful if you could send thèse contributions, no later than 31 December 2020, to Gabriela Ramos, Assistant Director-General for thé Social and Human Sciences Sector, at thé following émail address: [email protected]. Thanking you for your commitment to this essential issue, please accept, Sir/Madam, thé assurances of my highest considération. Audrey Azoulay Director-General Encs: 2 1. Preliminary report on thé first draft of thé Recommendation on thé Ethics ofArtificiat Intelligence 2. First draft of thé Recommendation on thé Ethics ofArtificial Intelligence Ce: National Commissions for UNESCO Permanent Délégations to UNESCO 7, place de Fontenoy 75352 Paris 07 SP, France Tél. :+33 (0)1 45681000 To Ministers in charge of relations with UNESCOPRELIMINARY REPORT ON THE FIRST DRAFT 0F THE RECOMMENDATION ON THE ETHICS 0F ARTIFICIAL INTELLIGENCE This preiiminary report was prepared in accordance with Article 10. 1 of thé Rules of Procédure concerning recommendations to Member States and international conventions covered by thé terms of Article IV, paragraph 4, of thé Constitution, on thé draft Recommendation on thé Ethics ofArtificial Intelligence. INTRODUCTION 1. Artificial intelligence1 (Al) is one of thé central issues of thé era of converging technologies with profound implications for humanity, cultures, societies and thé environment. Al is already having impact across ail sectors, and it is already transforming thé future of éducation, natural and social sciences, culture and communication. Thèse areas, along with thé higher goal of promoting thé respect of human rights and human dignity, along with a culture of peace, are core to UNESCO's mandate. 2. As with previous technological révolutions, Al has thé potential to transform thé future of humanity for thé better and in favour of sustainable development. However, it can also bring downside risks and challenges, in particular derived from malicious utilization of thé technology, that affects human rights, or from thé fact that highly complex Al Systems can widen substantially thé already existing high inequalities and divides. In fact, thèse technologies, and thé digital transformation hâve a "winner takes ail dynamic" that needs to be addressed. Thé impact of thé technology will dépend on thé way humanity frames it and masters it, and on thé way it prioritizes thé goal of leaving no one behind. This is where UNESCO's rôle in promoting social inclusion and fighting inequalities, is of paramount importance at thé global level. In order to sketch possible scénarios and unlock Al's potential to grasp development opportunities, while managing risks, it is important to develop a more comprehensive understanding of how societies are transformed by disruptive technologies, such as Al. 3. Digital transformation is happening at a moment where increased inequalities of incarne and opportunities are at its highest level in décades (within and between countries). Thèse technologies are already contributing to deepening such inequalities in a profound way, as few firms and a handful of countries own thé largest share of thèse technologies. This can change thé landscape beyond récognition. Core éléments in thé Al life cycle détermine who participâtes and who doesn't. This can be related to data governance, which is a key issue (including data collection, data ownership, data sharing and data access). Data is thé core 'raw material' that altows countries and firms to advance new solutions and products, but it is highly concentrated and there is no shared international governance framework. Thé skills necessary for data mining and thé development of models and algorithms are also highly concentrated. Therefore, there is an essential link between data and modelling work, and thé governance ofAI. 4. Thé inclusion of women in thé digital transformation, or thé lack of it, is especially pertinent, as thé development of thèse technologies is massively contributing to opening an additional gender gap. Women do not participate equally in IT disciplines (they participate even less in IT than in science, technology, engineering and mathematics (STEM); they are underrepresented in thé industry, and therefore in thé development of Al Systems. This opens a worrisome trend by which gender blases and stereotyping are being reproduced massively in thé technologies. If technologies are not developed and applied in a gender-responsive way, they are therefore likely to reproduce and reinforce existing discriminatory social norms and even more violence against women (which is already happening). More women, including from différent ethnie origins, must be involved in thé While there is no one définition of "Artificial Intelligence" (Al), debate tends to focus on machines capable of imitating certain functionalities of human intelligence, including such features as perception, leaming, reasoning, problem solving, language interaction, and even producing créative work.-2- design and deployment ofAI tools. Thé same can be said about vulnérable populations, and those that are left behind, or discriminated against. They will find it more difficult to cope and benefit from thé technological progress that Al technologies bring. 5. This is why thé work needs to be accompanied by an ethical reflection. It is clear that Al technologies are not value neutral, but inherently biased, among others, due to thé data on which they are trained, and thé choices made while training on thé data. It is also influenced by thé fact that Al machine décisions (particularly those based on Machine Learning) cannot be fully predictable or understood. Moreover, as Al is a distributed technology, whose current practical governance is spread across numerous institutions, organizations and companies, thé reflection on its good governance requires a pluralistic, multidisciplinary, multicultural and multi-stakeholder approach, opening up questions about what type of future we want for humanity. This reflection needs to address thé main challenges in thé development ofAI technologies related to thé blases embedded in algorithms, thé protection of people's privacy and personal data, thé risks of creating new forms of exclusion and inequalities, thé issues of just distribution of benefits and risks, accountability, responsibility, impacts on employment and thé future of work, human rights, security and risks of dualuse. 6. To leverage thé potential of Al for thé good of humanity, UNESCO hosted in Paris a séries of meetings in 2018, including; thé seminar on artificial intelli ence co-organized by thé Permanent Délégations of Brazil and Turkey to UNESCO (June 2018); a roundtable on "Artificial Intelli ence: Reflection on its com lexit and im act on our societ " (September 2018); an open discussion on "Harnessin Artificial Intelli ence to Foster Knowled e Societies and Good Governance", held at thé Mozilla Foundation (November2018), and thé debate on "Al for Human Ri hts and SDGs: Fosterin Multi-Stakeholder Inclusive and 0 en A roaches" held as part of thé Internet Governance Forum (November 2018). In coopération with UNESCO's partners in thé field, several events were held in différent régions, such as Thé First Forum on Al and Internet of Thin s in Smart Sustainable Cities (Buenos Aires, Argentins, May 2018). UNESCO's ongoing reflection on thé ethical global dialogue on Al, focusing on norms and standards, was also depicted by thé Forum on Al in Africa organized in December 2018 in Benguérir, Morocco, where thé participants called for thé development of an Al strategy for Africa for instance, as well as for an Al that focuses on thé human dimension. UNESCO's overall planned action on Al was presented to Member States at thé Information Meetin held on 22 January 2019, which was followed by thé international experts' debate on "Tech Futures: Ho e or Fear?". This reflection was further developed on 4 March 2019 in UNESCO Headquarters with a global conférence entitled "Princi tes for artificial intetli ence:towardsahumanistica roach?" and UNESCO's Mobile Learnin Week 2019 (4-8 March), that focused on Al and sustainable development. In May 2019, thé International Conférence on Al and Education was also held in Beijing, China. Furthermore, in December 2019, a second roundtable on "Chan in Relationshi between Artificial Intelli ence and Humans" was held at UNESCO Headquarters; and thé Re ional Forum on Artificial Intelli ence in Latin America and thé Caribbean was organized in Sao Paulo, Brazil. 7. Because of its profound social implications, many organizations and governments are concerned about thé ethical implications of Al. National, régional and other stratégies and frameworks on Al are being developed. There is a growing number of reports and guidelines on Al and ethics - such as those by thé Council of Europe, thé European Union, thé Institute of Electrical and Electronics Engineers (IEEE), thé Organisation for Economie Co-operation and Development (OECD), thé International Télécommunications Union, thé World Health Organization (WHO), and thé UN Secretary-General's High Level Panel on Digital Coopération - which hâve identified important principlesforthe design, development, and deploymentofAI. UNESCO has been following thèse discussions closely, e. g. as a Co-Champion on Al of thé UN Secretary-General's High Level Panel on Digital Coopération, an observer in thé European Commission High-Level Expert Group on Al and in thé Council of Europe Ad hoc Committee on Artificial Intelligence (CAHAI), a member of thé OECD Expert Group on Al (AIGO), a partner of thé ITU Al for Good Global Summit and a participant in other intergovernmental forums. UNESCO also hosted a high-level dialogue on thé ethical dimensions of Al during thé World Summit on thé Information Society (WSIS) Forum in April-3- 2019, and presented ongoing work on ethics of artificial intelligence at a high-level policy session during thé virtual WSIS Forum in July 2020. 8. UNESCO has a unique perspective to add to this debate given its strong comparative advantage thanks to its universality in membership and drawing on its multidisciplinary expertise. In this regard, thé Organization can truly provide a global and pluralistic platform for dialogue on thé ethics of Al, bringing together both developed and developing countries, différent cultural perspectives, as well as various stakeholders within thé public and private sphères. Therefore, in addition to thé many ethical guidetines and frameworks that are currently being developed by governments, companies, and societal organizations, UNESCO can contribute to thé development of Al for thé benefit of ail of humanity, sustainable development, and peace. For this purpose, UNESCO acts as a bridge between Member States -which hâve highlighted repeatedly their support for UNESCO's Al work - and civil society, thé technical community, academia, thé private sector, including thé cultural and créative industries, building on its record of multi-stakeholder consultation and consensus building. 9. Building on thé previous work of UNESCO's World Commission on thé Ethics of Scientific Knowledge and Technology (COMEST), in August 2018, thé Commission was requested to prépare a preliminary study on thé ethics ofAI to help inform UNESCO's reflection in this area. Thé work of COMEST on this topic provided thé background for thé preliminary study on thé technical and légal aspects relating to thé desirability of a standard-setting instrument on thé ethics of artificial intelligence (Al), which was submitted to thé Executive Board at its 206th session (document 206 EX/42), in accordance with Article 3 of thé Rules of Procédure concerning recommendations to Member States and international conventions covered by thé terms of Article IV, paragraph 4 of thé Constitution. 10. Subsequently, thé Executive Board adopted 206 EX/Decision 42 inscribing thé preliminary study on thé agenda of thé 40th session of thé Général Conférence. Thé Général Conférence examined thé preliminary study and thé comments and observations of thé Executive Board thereon (document 40 C/67). By 40 C/Resolution 37, thé Général Conférence recognized thé benefits of an international standard-setting instrument on thé ethics of artificial intelligence in thé form of a recommendation (thé "Recommendation"). Furthermore, MemberStates invited UNESCO to initiate thé process of elaborating thé Recommendation, and to submit to it at its 41 st session thé draft text of thé Recommendation. Thé mandate provided by thé Member States is highly important: there are many efforts trying to ensure governance of thé development of Al, while this is thé only standard- setting instrument that is clear on thé fact that this governance should be based on strong respect of human rights, human dignity, and ethical considérations. 11. Thé COVID-19 pandemic has exemplified how Al technologies can help, but also how much we need a global framework that informs thé way we use them and thé way we deploy them. To control thé pandemic, we are giving access to our data and our activities, raising concerns regarding privacy and data protection and thé temporal scope of extended powers, which ail require reconceptualization to ensure respect for thé rule of law. 12. This international normative instrument will thus help thé world reach an ethical consensus on Al, which is not just a technological turning point but also an anthropological disruption. This is a necessary step to more proactive thinking beyond thé traditional légal approaches, which lag behind. Thé proposed Recommendation should become an ethical guiding compass and a normative bedrock allowing to build a strong respect for thé rule of law in thé digital world. THE PROCESS 13. In line with thé roadmap presented in Annex III of document 40 C/67, in March 2020, an Ad Hoc Ex ert Grou (AhlEG) (category VI meeting) was established to elaborate a first draft of thé Recommendation. It is composed of24 independent experts, appointed on a geographical balance basis from ail UNESCO régions. From March to May 2020, due to thé COVID-19 pandemic, thé-4- AHEG conducted its work using online/virtual modalities over a period of six weeks and produced a first version of thé draft Recommendation. 14. A multi-stakeholder consultation process on thé first version of thé draft Recommendation was conducted from June to August 2020. In light of thé challenges posed by thé pandemic, artners from around thé world came together with UNESCO to support this consultation process based on three components: (i) public online consultation; (ii) régional and sub-regional virtual consultations co-organized with host countries/institutions in ail of UNESCO's régions; and (iii) open, multi- stakeholder, and citizen délibération workshops organized by partners. Thé public online consultation received completed responses from more than 600 pari:icipants, generating more than 50,000 comments. Furthermore, eleven régional and sub-regional virtual consultations, including two with youth, were held in ail of UNESCO's régions, involving more than 500 participants. Thé majority of thé consultations were opened at a high level, and included thé Assistant Director- Général for SHS. Finally, thé open, multi-stakeholder, and citizen délibération workshops drew approximately 500 participants. 15. Thé consultations generated a rich conversation that raised awareness, triggered régional and sub-regional debates, and willingness of diverse stakeholders to collaborate with UNESCO in this endeavour. In at least three régions or sub-regions, participants hâve expressed thé willingness to establish régional and sub-regional networks of différent disciplines to continue discussions on thé ethics ofAI. 16. Thé consultations hâve clearly shown thé appetite to hâve an ethical framework on Al. Some régions hâve called for co-creation practices along thé whole Al life cycle, ensuring broader understanding of éducation, ensuring gender equality, preserving thé value of thé native cultures beyond their languages, strengthen thé need for human rights framework. They hâve also called for peace to be added as thé only way to live in harmony and underlined thé need for thé document to be intergenerational in terms of values. There is évident strong acceptance of thé environmental dimension of Al technologies. Moreover, there is support for its comprehensiveness in terms of including ail thé ecosystems and departing from a purely anthropocentric view. 17. Consultation on thé first version of thé draft Recommendation was also held with thé UN System through thé intersessional meeting of thé High-Level Committee on Programmes (HLCP) on 29 July 2020. Ail entities expressed their full support and recognized thé importance of UNESCO's work to develop this Recommendation and its relevance to thé UN System, Member States and other actors. In général, it was emphasized that thé text can and should be more ambitious by underlining more its innovative features and by proposing stronger language and far reaching recommendations when it cornes to issues such as governance, human rights, and implementation after its adoption. This is necessary in orderto effectively ensure ethical approaches to Al worldwide, to stand out from other documents in thé field and to help position UNESCO as a référence for capacity building for Member States in thé area of ethics of Al. Thé comments of UN partners were transmitted to thé AHEG accordingly. 18. Thé AHEG then revised thé first version of thé draft Recommendation from mid-August until early September 2020, based on thé feedback of thé consultations and on its further discussions. As such, thé first draft of thé Recommendation is transmitted to Member States together with this preliminary report. THE FIRST DRAFT 0F THE RECOIVIMENDATION 19. When preparing thé first version of thé draft Recommendation, thé AhlEG's discussions concentrated primarily around thé risks and challenges inhérent to thé rapid development of Al technologies that hâve been outlined at thé outset of this report. Spécifie attention has been paid to how much At technologies are already massively impacting our world, and to an urgent need for a common framework and narrative. Thé world is not fair and just, on thé contrary, and we are often receding from our long-standing achievements. Thé COVID-19 pandemic is exacerbating thé-5- inequality trends by hitting hardest thé most vulnérable population. Thé fundamental dichotomy is either thé technologies contribute to worsen thé problems of exclusion, inequalities and division, or they are used for noble purposes. Thus, thé AHEG identified and conceptualized thé issues that need to be corrected and thé outcomes that they want to achieve with thé proposed ethical framework. 20. While recognizing thé potential rôle of Al in contributing to solving complex challenges and bringing immense économie value, thé AHEG stressed thé concerns that Al créâtes regarding gender equality, diversity and inclusion, biases, privacy and data protection, increase of disinformation, digital exposure by populations at risk, just to name a few. Thé AHEG's délibérations showed that today, at thé global level, there is a need for a général universal ethical guidance in terms of core values that must underpin thé development of Al Systems. Due to their transnational character, durable solutions can be found only at thé global level. Such an instrument should provide a global mechanism to document thé social-cultural changes brought about by thé rapid, non-linear development ofAI and related ethical issues. 21. Having agreed that Al technologies are not value neutral, thé AHEG took thé approach of identifying foundational values and closely interlinked principles, which inform further outlined policy actions. Foundational values hâve a rôle of necessary preconditions or prerequisites for principles and policy actions to work. Taken together they form a cohérent framework to ensure ethical Al. Thé flrst version of thé draft Recommendation has identified thé values of human dignity, human rights and fundamental freedoms, leaving no one behind, living in harmony, trustworthiness, and protection of thé environment. Thé latter has been purposefully elevated to thé level of values because thé analysis of various documents proposing principles for Al has shown that thé protection of thé environment reçoives little attention or is overlooked. 22. Thé AHEG has also identified principles, which are associated with thé human-AI Systems interaction and those reflecting characteristics associated with thé properties of Al Systems themselves. Thé AHEG has recognized that there is an emerging convergence around thé importance of certain principles that allow ensuring ethical approaches within thé Al lifecycle. Therefore, thé first version of thé draft Recommendation includes such commonly recognized principles as human oversight, privacy, fairness, transparency and explainability, safety and security, responsibility and accountability, and others. Apart from that, thé AHEG proposed new principles, which hâve not been previously seen elsewhere, such as proportionality, which helps ensuring that Al methods used correspond to thé aims, do not bear excessive risks and are appropriate to thé context. Nevertheless, thé variety of source materials showed that both thé notions and substance of principles may vary significantly, and thus thé most challenging task for thé AHEG was not identification of principles but clarifying their meaning and proposing policy actions that would ensure their implementation in practice. 23. In view of implementing values and principles, thé AHEG has proposed a number of concrète policy actions, which span from promoting diversity and inclusiveness to mitigating négative impacts on society, culture and thé environment. Thé AHEG attempted to address thé identified problems also by introducing innovative approaches, e. g. in thé areas of governance, impact assessment and capacity building. To support that, thé scope and addressees of thé Recommendation hâve been expanded to include, among others, thé private sector. 24. Thé consultations hâve shown that thé draft Recommendation can be bolder in terms of substance, proposed language and courses of action. This is especially so in terms of addressing gender issues, and broader inclusion, justice and integrity. Greater emphasis is needed to reflect thé differential access to and impact of Al across subgroups and individuals, especially women, girls, older persans, youth and children, which should run through thé entire document. It was suggested for thé draft Recommendation to address more thé issues of diversity of country groups as well as différences within countries in terms of level and diffusion ofAI technologies. It was recognized that thé draft Recommendation highlights thé trade-offs, which are not always taken into account in other documents, and it was suggested to elaborate on those further in a more practical way-6- 25. Suggestions were made to further elaborate on stratégies to be followed in helping thé low and middle incarne countries, including but not limited to countries in Africa, Latin America and thé Caribbean, and Central Asia, as well as Small Island Developing States to access Al Systems in various fields. There is a need for principles, actions and/or other recommendations aimed specifically at bridging gaps, capacities, resources, and power imbalances between countries and régions with regard to thé prolifération of Al and capacity to regulate between Al-producing and Al- consuming countries. 26. Suggestions were made in terms of thé structure of thé document. In particular, thé policy actions hâve been perceived as rather complex and prolonged, which require furtherwork. Similarly, aims and objectives of thé document should become clearer and more concrète. This is important in order to keep thé recommendation simple and pragmatic so that it can be operationalized quickly. Consultations also underscored that thé additional value added of thé UNESCO instrument is linked to thé more action-oriented sections, advising on "how" to foster ethical Al developments, besides thé "what" related to thé définitions ofwhat is ethical. 27. Last but not least, it was suggested for thé draft Recommendation to be more ambitious. This includes making bold proposais and be more assertive in suggesting that a stronger international légal framework is needed. It has been proposed to be more explicit about thé innovative features that are reflected in this draft text in order for it to stand out among other documents. Additional innovative ideas and proposais were highly welcomed to appear in thé final first draft of thé Recommendation. 28. Thé comments received during thé consultation process hâve been taken into account and hâve been reflected in thé revised first draft of thé Recommendation that is annexed to this preliminary report. NEXT STEPS 29. It is envisagea that thé Recommendation, to be effective, would need to be accompanied by strong policy and capacity building support in its implementation and transformation into légal frameworks, and Al spécifie practices in thé public and private sector. As such, UNESCO will be bringing its multidisciplinary compétences to assist Member States in fulfilling thé provisions outlined in thé Recommendation once adopted. Thé Recommendation will also provide a solid foundation for thé entire UN System in its collective response to thé ethical challenges posed by Al technologies in various fields, with UNESCO playing a leading rôle in this area. In this regard, UNESCO has over 25 years of expérience in developing and implementing normative work through policy advice and capacity building activities in bioethics and ethics of science and technology. 30. Member States are now invited to submit their comments and observations on thé first draft of thé Recommendation annexed hereto no later than 31 December 2020. Responses should be transmitted to Gabriela Ramos, Assistant Director-General for thé Social and Human Sciences at thé following émail address: ai-ethics unesco. or . A final report containing one or more draft texts of thé Recommendation will be prepared on thé basis of those comments and observations and communicated to Member States by April 2021. Thé final report shall be submitted to thé intergovernmental meetings of a spécial committee (category II meeting) in April and June 2021 (exact dates to be determined) for finalization and eventual adoption by UNESCO's Général Conférence at its 41 st session.FIRST DRAFT 0F THE RECOMMENDATION ON THE ETHICS 0F ARTIFICIAL INTELLIGENCE PREAMBLE Thé Général Conférence of thé United Nations Educational, Scientific and Cultural Organization (UNESCO), meeting in Paris from xx to xx, at its xx session, Recognizing thé profound and dynamic impact of artificial intelligence (Al) on societies, ecosystems, and human lives, including thé human mind, in part because of thé new ways in which it influences human thinking, interaction and decision-making, and affects éducation, human, social and natural sciences, culture, and communication and information, Recalling that, by thé terms of its Constitution, UNESCO seeks to contribute to peace and security by promoting collaboration among nations through éducation, thé sciences, culture, and communication and information, in order to further universal respect for justice, for thé rule of law and for thé human rights and fundamental freedoms which are affirmed for thé peoples of thé world, Conv/'ncecf that thé standard-setting instrument presented hère, based on international law and on a global normative approach, focusing on human dignity and human rights, as well as gender equality, social and économie justice, physical and mental well-being, diversity, interconnectedness, inclusiveness, and environmental and ecosystem protection can guide Al technologies in a responsible direction, Considering that Al technologies can be of gréât service to humanity but also raise fundamental ethical concerns, for instance regarding thé biases they can embed and exacerbate, potentially resulting in inequality, exclusion and a threatto cultural, social and ecological diversity and social or économie divides; thé need for transparency and understandability of thé workings of algorithms and thé data with which they hâve been trained; and their potential impact on human dignity, human rights, gender equality, privacy, freedom of expression, access to information, social, économie, political and cultural processes, scientific and engineering practices, animal welfare, and thé environment and ecosystems, Recognizing that Al technologies can deepen existing divides and inequalities in thé world, within and between countries, and that justice, trust and fairness must be upheld so that no one should be left behind, either in enjoying thé benefits of Al technologies or in thé protection against their négative implications, while recognizing thé différent circumstances of différent countries and thé désire ofsome people not to take part in ail technological developments, Consclous of thé fact that ail countries are facing an accélération of thé use of information and communication technologies and Al technologies, as well as an increasing need for média and information literacy, and that thé digital economy présents important societat, économie and environmental challenges and opportunities of benefits sharing, especially for low- and middle- income countries (LMICs), including but not limited to least developed countries (LDCs), landlocked developing countries (LLDCs) and small island developing States (SIDS), requiring thé récognition, protection and promotion ofendogenous cultures, values and knowledge in order to develop sustainable digital économies, Recognizing that Al technologies hâve thé potential to be bénéficiai to thé environment and ecosystems but in order for those benefits to be realized, fair access to thé technologies is-2- required without ignoring but instead addressing potential harms to and impact on thé environment and ecosystems, Noting that addressing risks and ethical concerns should not hamper innovation but rather provide new opportunities and stimulate new and responsible practices of research and innovation that anchor Al technologies in human rights, values and principles, and moral and ethical reflection, Recalling that in November 2019, thé Général Conférence of UNESCO, at its 40th session, adopted 40 C/Resolution 37, by which it mandated thé Director-General "to prépare an international standard-setting instrument on thé ethics ofartificial intelligence (Al) in thé form ofa recommendation", which is to be submitted to thé Général Conférence at its 41 st session in 2021, Recognizing that thé development ofAI technologies results in an increase of information which nécessitâtes a commensurate increase in média and information literacy as well as access to critical sources of information, Observing that a normative framework for Al technologies and its social implications finds its basis in ethics, as well as human rights, fundamental freedoms, access to data, information and knowledge, international and national légal frameworks, thé freedom of research and innovation, human and environmental and ecosystem well-being, and connects ethical values and principles to thé challenges and opportunities linked to Al technologies, based on common understanding and shared aims, Recognizing that ethical values and principles can powerfully shape thé development and implementation of rights-based policy measures and légal norms, by providing guidance where thé ambit of norms is unclear or where such norms are not yet in place due to thé fast pace of technological development combined with thé relatively slower pace of policy responses, Convinced that globally accepted ethical standards for Al technologies and international law, in particular human rights law, principles and standards can play a key rôle in harmonizing Al-related légal norms across thé globe, RecognizingVne Universal Déclaration ofhfuman Rights(1948), including Article 27 emphasizing thé right to share in scientific advancement and its benefits; thé instruments of thé international human rights framework, including thé International Convention on thé Elimination of Ail Forms of Racial Discrimination (1965), thé International Covenant on Civil and Political Rights (1966), thé International Covenant on Economie, Social and Cultural Rights (1966), thé United Nations Convention on thé Elimination of Ail Forms of Discrimination against Women (1979), thé United Nations Convention on thé Rights of thé Child (1989), and thé United Nations Convention on thé Rights of Persans with Disabilities (2006); thé UNESCO Convention on thé Protection and Promotion of thé Diversity of Cultural Expressions (2005), Noting thé UNESCO Déclaration on thé Responsibilities of thé Présent Générations Towards Future Générations (1997); thé United Nations Déclaration on thé Rights of Indigenous Peuples (2007); thé Report of thé United Nations Secretary-General on thé Follow-up to thé Second World Assembly on Ageing (A/66/173) of 2011, focusing on thé situation of thé human rights of older persans; thé Report of thé Spécial Représentative of thé United Nations Secretary-General on thé issue of human rights and transnational corporations and other business enterprises (A/HRC/17/31) of 2011, outlining thé "Guiding Principles on Business and Human Rights: Implementing United Nations 'Protect, Respect and Remedy' Framework"; thé United Nations-3- Général Assembly résolution on thé review of thé World Summit on thé Information Society (A/68/302); thé Human Rights Council's résolution on "Thé right to privacy in thé digital âge" (A/HRC/RES/42/15) adopted on 26 September 2019; thé Report of thé Spécial Rapporteur on thé promotion and protection of thé right to freedom of opinion and expression (A/73/348); thé UNESCO Recommendation on Science and Scientific Researchers (2017); thé UNESCO Internet Universality Indicators (endorsed by UNESCO's International Programme for thé Development of Communication in 2019), including thé R.O.A.M. principles (endorsed by UNESCO's Général Conférence in 2015); thé UNESCO Recommendation Concerning thé Préservation of, and Access to, Documentary Héritage Including in Digital Form (2015); thé Report of thé United Nations Secretary-General's High-level Panel on Digital Coopération on "Thé Age of Digital Interdependence" (2019), and thé United Nations Secretary-General's Roadmap for Digital Coopération (2020); thé Universal Déclaration on Bioethics and Human Rights (2005); thé UNESCO Déclaration on Ethical Principles in relation to Climate Change (2017); thé United Nations Global Puise initiative; and thé outcomes and reports of thé ITU's Al for Good Global Summits, Noting also existing frameworks related to thé ethics of Al of other intergovernmental organizations, such as thé relevant human rights and other légal instruments adopted by thé Council of Europe, and thé work of its Ad Hoc Committee on Al (CAHAI); thé work of thé European Union related to Al, and of thé European Commission's High-Level Expert Group on Al, including thé Ethical Guidelines for Trustworthy Al; thé work of OECD's first Group of Experts (AIGO) and its successor thé OECD Network of Experts on Al (ONE Al), thé OECD's Recommendation of thé Council on Al and thé OECD Al Policy Observatory (OECD. AI); thé G20 Al Principles, drawn therefrom, and outlined in thé G20 Ministerial Statement on Trade and Digital Economy; thé G7's Chartevoix Common Vision for thé Future of Al; thé work of thé African Union's Working Group on Al; and thé work of thé Arab League's Working Group on Al, Emphasizing that spécifie attention must be paid to LMICs, including but not limited to LDCs, LLDCs and SIDS, as they hâve their own capacity but hâve been underrepresented in thé Al ethics debate, which raises concerns about neglecting local knowledge, cultural and ethical pluralism, value Systems and thé demands of global fairness to deal with thé positive and négative impacts ofAI technologies, Conscious of thé many existing national policies and other frameworks related to thé ethics and régulation ofAI technologies, Conscious as we//ofthe many initiatives and frameworks related to thé ethics ofAI developed by thé private sector, professional organizations, and non-governmental organizations, such as thé IEEE'S Global Initiative on Ethics of Autonomous and Intelligent Systems and its work on Ethically Aligned Design; thé World Economie Forum's "Global Technology Governance: A Multistakeholder Approach"; thé UNI Global Union's "Top 10 Principles for Ethical Artificial Intelligence"; thé Montréal Déclaration for a Responsible Development of Al; thé Toronto Déclaration: Protecting thé rights to equality and non-discrimination in machine learning Systems; thé Harmonious Artificial Intelligence Principles (HAIP); and thé Tenets of thé Partnership on Al, Convinced that Al technologies can bring important benefits, but that achieving them can also amplify tension around innovation debt, asymmetric access to knowledge, barriers of rights to information and gaps in capacity of creativity in developing cycles, human and institutional capacities, barriers to access to technological innovation, and a lack of adéquate physical and digital infrastructure and regulatory frameworks regarding data,-4- Underlining that global coopération and solidarity are needed to address thé challenges that Al technologies bring in diversity and interconnectivity of cultures and ethical Systems, to mitigate potential misuse, and to ensure that Al stratégies and regulatory frameworks are not guided only by national and commercial interests and économie compétition, Taking fully into account that thé rapid development of Al technologies challenges their ethical implementation and governance, because of thé diversity of ethical orientations and cultures around thé world, thé lack of agility of thé law in relation to technology and knowledge societies, and thé risk that local and régional ethical standards and values be disrupted by Al technologies, 1. Adopts thé présent Recommendation on thé Ethics of Artificial Intelligence; 2. Recommends that Member States apply thé provisions ofthis Recommendation by taking appropriate steps, including whatever législative or other measures may be required, in conformity with thé constitutional practice and governing structures of each State, to give effect within their jurisdictions to thé principles and norms of thé Recommendation in conformity with international law, as well as constitutional practice; 3. Also recommends that Member States ensure assumption of responsibilities by ail stakeholders, including private sector companies in Al technologies, and bring thé Recommendation to thé attention of thé authorities, bodies, research and académie organizations, institutions and organizations in public, private and civil society sectors involved in Al technologies, in order to guarantee that thé development and use of Al technologies are guided by both sound scientific research as well as ethical analysis and évaluation. l. SCOPE 0F APPLICATION 1. This Recommendation addresses ethical issues related to Al. It approaches Al ethics as a systematic normative reflection, based on a holistic and evotving framework of interdependent values, principles and actions that can guide societies in dealing responsibly with thé known and unknown impacts of Al technologies on human beings, societies, and thé environment and ecosystems, and offers them a basis to accept or reject Al technologies. Rather than equating ethics to law, human rights, or a normative add-on to technologies, it considers ethics as a dynamic basis for thé normative évaluation and guidance of Al technologies, referring to human dignity, well-being and thé prévention of harm as a compass and rooted in thé ethics of science and technology. 2. This Recommendation does not hâve thé ambition to provide one single définition of Al, since such a définition would need to change over time, in accordance with technological developments. Rather, its ambition is to address those features of Al Systems that are of central ethical relevance and on which there is large international consensus. Therefore, this Recommendation approaches Al Systems as technological Systems which hâve thé capacity to process information in a way that resembles intelligent behaviour, and typically includes aspects of reasoning, learning, perception, prédiction, planning or control. Three éléments hâve a central place in this approach: (a) Al Systems are information-processing technologies that embody models and algorithms that produce a capacity to learn and to perform cognitive tasks leading to outcomes such as prédiction and decision-making in real and virtual environments. Al Systems are designed to operate with some aspects of autonomy by means of-5- knowledge modelling and représentation and by exploiting data and calculating corrélations. Al Systems may include several methods, such as but not limited to: (i) machine learning, including deep learning and reinforcement learning, (ii) machine reasoning, including planning, scheduling, knowledge représentation and reasoning, search, and optimization, and (iii) cyber-physical Systems, including thé Internet-of-Things, robotic Systems, social robotics, and human-computer interfaces which involve control, perception, thé processing of data collected by sensors, and thé opération of actuators in thé environment in which Al Systems work. (b) Ethical questions regarding Al Systems pertain to ail stages of thé Al System life cycle, understood hère to range from research, design, and development to deployment and use, including maintenance, opération, trade, financing, monitoring and évaluation, validation, end-of-use, disassembly, and termination. In addition, Al actors can be defined as any actor involved in at least one stage of thé Al life cycle, and can refer both to natural and légal persans, such as researchers, programmers, engineers, data scientists, end-users, large technology companies, small and médium enterprises, start-ups, universities, public entities, among others. (e) Al Systems raise new types of ethical issues that include, but are not limited to, their impact on decision-making, employment and labour, social interaction, health care, éducation, média, freedom of expression, accessto information, privacy, democracy, discrimination, and weaponization. Furthermore, new ethical challenges are created by thé potential of Al algorithms to reproduce biases, for instance regarding gender, ethnicity, and âge, and thus to exacerbate already existing forms of discrimination, identity préjudice and stereotyping. Some of thèse issues are related to thé capacity ofAI Systems to perform tasks which previously only living beings could do, and which were in some cases even limited to human beings only. Thèse characteristics give Al Systems a profound, new rôle in human practices and society, as welt as in their relationship with thé environment and ecosystems, creating a new context for children and young people to grow up in, develop an understanding of thé world and themselves, critically understand média and information, and learn to make décisions. In thé long term, Al Systems could challenge human's spécial sensé of expérience and agency, raising additional concerns about human self-understanding, social, cultural and environmental interaction, autonomy, agency, worth and dignity. 3. This Recommendation pays spécifie attention to thé broader ethical implications of Al Systems in relation to thé central domains of UNESCO: éducation, science, culture, and communication and information, as explored in thé 2019 Preliminary Study on thé Ethics of Artificial Intelligence by thé UNESCO World Commission on Ethics ofScientific Knowledge and Technology (COMEST): (a) Education, because living in digitalizing societies requires new educational practices, thé need for ethical reflection, critical thinking, responsible design practices, and new skills, given thé implications for thé labour market and employability. (b) Science, in thé broadest sensé and including ail académie fields from thé natural sciences and médical sciences to thé social sciences and humanities, as Al-6- technologies bring new research capacities, hâve implications for our concepts of scientific understanding and explanation, and create a new basis for decision-making. (e) Cultural identity and diversity, as Al technologies can enrich cultural and créative industries, but can also lead to an increased concentration of supply of cultural content, data, markets, and incarne in thé hands of only a few actors, with potential négative implications for thé diversity and pluralism of languages, média, cultural expressions, participation and equality. (d) Communication and information, as Al technologies play an increasingly important rôle in thé processing, structuring and provision of information, and thé issues of automated journalism and thé algorithmic provision of news and modération and curation of content on social média and search engines are just a few examples raising issues related to access to information, disinformation, misinformation, misunderstanding, thé émergence of new forms ofsocietal narratives, discrimination, freedom of expression, privacy, and média and information literacy, among others. 4. This Recommendation is addressed to States, both as Al actors and as responsible for developing légal and regulatory frameworks throughout thé entire Al System life cycle, and for promoting business responsibility. It also provides ethical guidance to ail Al actors, including thé private sector, by providing a basis for an Ethical Impact Assessment ofAI Systems throughout their life cycle. II. AIMS AND OBJECTIVES 5. This Recommendation aims to provide a basis to make Al Systems work for thé good of humanity, individuals, societies, and thé environment and ecosystems; and to prevent harm. 6. In addition to thé ethical frameworks regarding Al that hâve already been developed by various organizations ail aver thé world, this Recommendation aims to bring a globally accepted normative instrument that does not only focus on thé articulation of values and principles, but also on their practical realization, via concrète policy recommendations, with a strong emphasis on issues of gender equality and protection of thé environment and ecosystems. 7. Because thé complexity of thé ethical issues surrounding Al nécessitâtes thé coopération of multiple stakeholders across thé various levels and sectors of international, régional and national communities, this Recommendation aims to enable stakeholders to take shared responsibility based on a global and intercultural dialogue. 8. Thé objectives ofthis Recommendation are: (a) to provide a universal framework of values, principles and actions to guide States in thé formulation oftheir législation, policies or other instruments regarding Al; (b) to guide thé actions of individuels, groups, communities, institutions and private sector companies to ensure thé embedding ofethics in ail stages of thé Al System life cycle; (e) to promote respect for human dignity and gender equality, to safeguard thé interests of présent and future générations, and to protect human rights, fundamental freedoms, and thé environment and ecosystems in ail stages of thé Al System life cycle;- 7 - (d) to foster multi-stakeholder, multidisciplinary and pluralistic dialogue about ethical issues reiating to Al Systems; and (e) to promote équitable access to developments and knowledge in thé field of Al and thé sharing of benefits, with particular attention to thé needs and contributions of LMICs, including LDCs, LLDCs and SIDS. III. VALUES AND PRINCIPLES 9. Thé values and principles included below should be respected by ail actors in thé Al System life cycle, in thé first place, and be promoted through amendments to existing and élaboration of new législation, régulations and business guidelines. This must comply with international law as well as with international human rights law, principles and standards, and should be in line with social, political, environmental, educational, scientific and économie sustainability objectives. 10. Values play a powerful rôle as motivating ideals in shaping policy measures and légal norms. While thé set of values outlined below thus inspires désirable behaviour and represents thé foundations of principles, thé principles unpack thé values underlying them more concretely so that thé values can be more easily operationalized in policy statements and actions. 11. While ail thé values and principles outlined below are désirable per se, in any practical context there are inévitable trade-offs among them, requiring complex choices to be made about contextual prioritization, without compromising other principles or values in thé process. Trade- offs should take account of concerns related to proportionality and legitimate purpose. To navigate such scénarios judiciously will typically require engagement with a broad range of appropriate stakeholders guided by international human rights law, standards and principles, making use of social dialogue, as well as ethical délibération, due diligence, and impact assessment. 12. Thé trustworthiness and integrity of thé life cycle of Al Systems, if achieved, work for thé good of humanity, individuals, societies, and thé environment and ecosystems, and embody thé values and principles set out in this Recommendation. People should hâve good reason to trust that A! Systems bring shared benefits, while adéquate measures are taken to mitigate risks. An essential requirement for trustworthiness is that, throughout their life cycle, Al Systems are subject to monitoring by governments, private sector companies, independent civil society and other stakeholders. As trustworthiness is an outcome of thé operationalization of thé principles in this document, thé policy actions proposed in this Recommendation are ail directed at promoting trustworthiness in ail stages of thé Al life cycle. 111. 1 VALUES Respect, protection and promotion of human dignity, human rights and fundamental freedoms 13. Thé dignity of every human persan constitutes a foundation for thé indivisible System of human rights and fundamental freedoms and is essential throughout thé life cycle ofAI Systems. hluman dignity relates to thé récognition of thé intrinsic worth ofeach individual human being and thus dignity is not tied to sex, gender, language, religion, political or other opinion, national, ethnie, indigenous or social origin, sexual orientation and gender identity, property, birth, disability, âge or other status.-8- 14. No human being should be harmed physically, economically, socially, politically, or mentally during any phase of thé life cycle ofAI Systems. Throughoutthe life cycle ofAI Systems thé quality of life of every human being should be enhanced, while thé définition of "quality of life" should be left open to individuals or groups, as long as there is no violation or abuse of human rights, or thé dignity of humans in terms ofthis définition. 15. Persans may interact with Al Systems throughout their life cycle and receive assistance from them such as care for vulnérable people, including but not limited to children, older persans, persons with disabilities or thé ill. Within such interactions, persans should never be objectified, nor should their dignity be undermined, or human rights violated or abused. 16. Human rights and fundamental freedoms must be respected, protected, and promoted throughout thé life cycle of Al Systems. Governments, private sector, civil society, international organizations, technical communities, and academia must respect human rights instruments and frameworks in their interventions in thé processes surrounding thé life cycle of Al Systems. New technologies need to provide new means to advocate, défend and exercise human rights and not to infringe them. Environment and ecosystem flourishing 17. Environmental and ecosystem flourishing should be recognized and promoted through thé life cycle ofAI Systems. Furthermore, environment and ecosystems are thé existential necessity for humanity and other living beings to be able to enjoy thé benefits of advances in Al. 18. Ail actors involved in thé life cycle ofAI Systems must follow relevant international law and domestic législation, standards and practices, such as précaution, designed for environmental and ecosystem protection and restoration, and sustainable development. They should reduce thé environmental impact ofAI Systems, including but not limited to, its carbon footprint, to ensure thé minimization of climate change and environmental risk factors, and prevent thé unsustainable exploitation, use and transformation of natural resources contributing to thé détérioration of thé environment and thé dégradation of ecosystems. Ensuring diversity and inclusiveness 19. Respect, protection and promotion of diversity and inclusiveness should be ensured throughout thé life cycle of Al Systems, at a minimum consistent with international human rights law, standards and principles, as well as démographie, cultural, gender and social diversity and inclusiveness. This may be done by promoting active participation of ail individuals or groups based on sex, gender, language, religion, political or other opinion, national, ethnie, indigenous or social origin, sexual orientation and gender identity, property, birth, disability, âge or other status, in thé life cycle of Al Systems. Any homogenizing tendency should be monitored and addressed. 20. Thé scope of lifestyle choices, beliefs, opinions, expressions or personal expériences, including thé optional use of Al Systems and thé co-design of thèse architectures should not be restricted in any way during any phase of thé life cycle ofAI Systems. 21. Furthermore, efforts should be made to overcome, and never exploit, thé lack of necessary technological infrastructure, éducation and skills, as well as légal frameworks, in some communities, and particularly in LMICs, LDCs, LLDCs and SIDS.-9- Living in harmony and peace 22. Al actors should play an enabling rôle for harmonious and peaceful life, which is to ensure an interconnected future ensuring thé benefit of ail. Thé value of living in harmony and peace points to thé potential of Al Systems to contribute throughout their life cycle to thé interconnectedness of ail living créatures with each other and with thé natural environment. 23. Thé notion of humans being interconnected is based on thé knowledge that every human belongs to a greater whole, which is diminished when others are diminished in any way. Living in harmony and peace requires an organic, immédiate, uncalculated bond of solidarity, characterized by a permanent search for non-conflictual, peaceful relations, tending towards consensus with others and harmony with thé natural environment in thé broadest sensé of thé te rm. 24. This value demands that peace should be promoted throughout thé life cycle ofAI Systems, in so far as thé processes of thé life cycle of Al Systems should not segregate, objectify, or undermine thé safety of human beings, divide and turn individuals and groups against each other, or threaten thé harmonious coexistence between humans, non-humans, and thé natural environment, as this would negatively impact on humankind as a collective. 111.2 PRINCIPLES Proportionality and do no harm 25. It should be recognized that Al technologies do not necessarily, per se, ensure human and environmental and ecosystem flourishing. Furthermore, none of thé processes related to thé Al System life cycle shall exceed what is necessary to achieve legitimate aims or objectives and should be appropriate to thé context. In thé event of possible occurrence of any harm to human beings or thé environment and ecosystems, thé implementation of procédures for risk assessment and thé adoption of measures in order to preclude thé occurrence of such harm should be ensured. 26. Thé choice of an Al method should be justified in thé following ways: (a) Thé Al method chosen should be désirable and proportional to achieve a given legitimate aim; (b) Thé Al method chosen should not hâve a négative infringement on thé foundational values captured in this document; (e) Thé Al method should be appropriate to thé context and should be based on rigorous scientific foundations. In scénarios that involve life and death décisions, final human détermination should apply. Safety and security 27. Unwanted harms (safety risks) and vulnerabilities to attacks (security risks) should be avoided throughout thé life cycle of Al Systems to ensure human and environmental and ecosystem safety and security. Safe and secure Al will be enabled by thé development of sustainable, privacy-protective data access frameworks that foster better training of Al models utilizing quality data. Fairness and non-discrimination 28. Al actors should promote social justice, by respecting fairness. Fairness implies sharing benefits of Al technologies at local, national and international levels, while taking into-10- considération thé spécifie needs of différent âge groups, cultural Systems, différent language groups, persans with disabiiities, giris and women, and disadvantaged, marginalized and vulnérable populations. At thé local level, it is a matter ofworking to give communities access to Al Systems in thé languages oftheir choice and respecting différent cultures. At thé national level, governments are obliged to demonstrate equity between rural and urban areas, and among ail persans without distinction as to sex, gender, language, religion, political or other opinion, national, ethnie, indigenous or social origin, sexual orientation and gender identity, property, birth, disability, âge or other status, in terms of access to and participation in thé Al System life cycle. At thé international level, thé most technologically advanced countries hâve an obligation of solidarity with thé least advanced to ensure that thé benefits of Al technologies are shared such that access to and participation in thé Al System life cycle for thé latter contributes to a fairer world order with regard to information, communication, culture, éducation, research, and socio- économie and political stability. 29. Al actors should make ail efforts to minimize and avoid reinforcing or perpetuating inappropriate socio-technical blases based on identity préjudice, throughout thé life cycle of thé Al System to ensure fairness of such Systems. There should be a possibility to hâve a remedy against unfair algorithmic détermination and discrimination. 30. Furthermore, discrimination, digital and knowledge divides, and global inequalities need to be addressed throughout an Al system life cycle, including in terms ofaccess to technology, data, connectivity, knowledge and skills, and participation of thé affected communities as part of thé design phase, such that every person is treated equitably. Sustainability 31. Thé development of sustainable societies relies on thé achievement of a complex set of objectives on a continuum of social, cultural, économie and environmental dimensions. Thé advent of Al technologies can either benefit sustainability objectives or hinder their realization, depending on how they are applied across countries with varying levels of development. Thé continuous assessment of thé social, cultural, économie and environmental impact of Al technologies should therefore be carried out with full cognizance of thé implications of Al technologies for sustainability as a set ofconstantly evolving goals across a range of dimensions, such as currently identified in thé United Nations Sustainable Development Goals (SDGs). Privacy 32. Privacy, a right essential to thé protection of human dignity, human autonomy and human agency, must be respected, protected and promoted throughout thé life cycle ofAt Systems both at thé personal and collective level. tt is crucial that data for Al is being collected, used, shared, archived and deleted in ways that are consistent with thé values and principles set forth in this Recommendation. 33. Adéquate data protection frameworks and governance mechanisms should be established by regulatory agencies, at national or supranational level, protected by judicial Systems, and ensured throughout thé life cycle of Al Systems. This protection framework and mechanisms concern thé collection, control aver, and use of data and exercise oftheir rights by data subjects and of thé right for individuals to hâve Personal data erased, ensuring a legitimate aim and a valid légal basis for thé processing of personal data as well as for thé personalization, and de- and re- personalization of data, transparency, appropriate safeguards for sensitive data, and effective independent oversight.-11 - 34. Algorithmic Systems require thorough privacy impact assessments which also include societal and ethical considérations of their use and an innovative use of thé privacy by design approach. Human oversight and détermination 35. It must always be possible to attribute ethical and légal responsibility for any stage of thé life cycle of Al Systems to physical persons or to existing légal entities. Human oversight refers thus not only to individual human oversight, but to public oversight, as appropriate. 36. It may be thé case that sometimes humans would hâve to rely on Al Systems for reasons of efficacy, but thé décision to cède control in limited contexts remains that of humans, as humans can resort to Al Systems in decision-making and acting, but an Al system can never replace ultimate human responsibility and accountability. Transparency and explainability 37. Thé transparency of Al Systems is often a crucial precondition to ensure that fundamental human rights and ethical principles are respected, protected and promoted. Transparency is necessary for relevant national and international liability législation to work effectively. 38. While efforts need to be made to increase transparency and explainability of Al Systems throughout their life cycle to support démocratie governance, thé level of transparency and explainability should always be appropriate to thé context, as some trade-offs exist between transparency and explainability and other principles such as safety and security. People hâve thé right to be aware when a décision is being made on thé basis of Al algorithms, and in those circumstances require or request explanatory information from private sector companies or public sector institutions. 39. From a socio-technical lens, greater transparency contributes to more peaceful, just and inclusive societies. It allows for public scrutiny that can decrease corruption and discrimination, and can also help detect and prevent négative impacts on human rights. Transparency may contribute to trust from humans for Al Systems. Spécifie to thé Al System, transparency can enable people to understand how each stage of an Al System is put in place, appropriate to thé context and sensitivity of thé Al System. It may also include insight into factors that impact a spécifie prédiction or décision, and whether or not appropriate assurances (such as safety or fairness measures) are in place. In cases where serious adverse human rights impacts are foreseen, transparency may also require thé sharing of spécifie code or datasets. 40. Explainability refers to making intelligible and providing insight into thé outcome of Al Systems. Thé explainability ofAI Systems also refers to thé understandability of thé input, output and behaviour of each algorithmic building block and how it contributes to thé outcome of thé Systems. Thus, explainability is closely related to transparency, as outcomes and sub-processes leading to outcomes should be understandable and traceable, appropriate to thé use context. 41. Transparency and explainability relate closely to adéquate responsibility and accountability measures, as well as to thé trustworthiness ofAI Systems.-12- Responsibility and accountability 42. Al actors should respect, protect and promote human rights and promote thé protection of thé environment and ecosystems, assuming ethical and légal responsibility in accordance with extant national and international law, in particular international human rights law, principles and standards, and ethical guidance throughout thé life cycle ofAI Systems. Thé ethical responsibility and liability for thé décisions and actions based in any way on an Al system should always ultimately be attributable to Al actors. 43. Appropriate oversight, impact assessment, and due diligence mechanisms should be developed to ensure accountability for Al Systems and their impact throughout their life cycle. Both technical and institutional designs should ensure auditability and traceability of (thé working of) Al Systems in particular to address any conflicts with human rights and threats to environmental and ecosystem well-being. Awareness and literacy 44. Public awareness and understanding of Al technologies and thé value of data should be promoted through open and accessible éducation, civic engagement, digital skills and Al ethics training, média and information literacy and training led jointly by governments, intergovernmental organizations, civil society, academia, thé média, community leaders and thé private sector, and considering thé existing linguistic, social and cultural diversity, to ensure effective public participation so that ail members of society can take informed décisions about their use of Al Systems and be protected from undue influence. 45. Learning about thé impact of Al Systems should include learning about, through and for human rights, meaning that thé approach and understanding of Al Systems should be grounded by their impact on human rights and access to rights. Multi-stakeholder and adaptive governance and collaboration 46. International law and sovereignty should be respected in thé use of data. Data sovereignty means that States, complying with international law, regulate thé data generated within or passing through their territories, and take measures towards effective régulation of data based on respect for thé right to privacy and other human rights. 47. Participation of différent stakeholders throughout thé Al System life cycle is necessary for inclusive Al governance, sharing of benefits of Al, and fair technological advancement and its contribution to development goals. Stakeholders include but are not limited to governments, intergovernmental organizations, thé technical community, civil society, researchers and academia, média, éducation, policy-makers, private sector companies, human rights institutions and equality bodies, anti-discrimination monitoring bodies, and groups for youth and children. Thé adoption of open standards and interoperability to facilitate collaboration must be in place. Measures must be adopted to take into account shifts in technologies, thé émergence of new groups of stakeholders, and to allow for meaningful intervention by marginalized groups, communities and individuals. IV. AREAS 0F POLICY ACTION 48. Thé policy actions described in thé following policy areas operationalize thé values and principles set out in this Recommendation. Thé main action is for Member States to put in place-13- policy frameworks or mechanisms and to ensure that other stakeholders, such as private sector companies, académie and research institutions, and civil society, adhère to them by, among other actions, assisting ail stakeholders to develop ethical impact assessment and due diligence tools. Thé process for developing such policies or mechanisms should be inclusive of ail stakeholders and should take into account thé circumstances and priorities of each Member State. UNESCO can be a partner and support Member States in thé development as well as monitoring and évaluation of policy mechanisms. 49. UNESCO recognizes that Member States will be at différent stages of readiness to implement this Recommendation, in terms of scientific, technological, économie, educational, légal, regulatory, infrastructural, societal, cultural and other dimensions. It is noted that "readiness" hère is a dynamic status. In order to enable thé effective implementation of this Recommendation, UNESCO will therefore: (1) develop a readiness assessment methodology to assist Member States in identifying their status at spécifie moments of their readiness trajectory along a continuum of dimensions; and (2) ensure support for Member States in terms of developing a globally accepted methodology for Ethical Impact Assessment (EIA) of Al technologies, sharing of best practices, assessment guidelines and other mechanisms and analytical work. POLICY AREA 1: ETHICAL IMPACT ASSESSMENT 50. Member States should introduce impact assessments to identify and assess benefits, concerns and risks ofAI Systems, as well as risk prévention, mitigation and monitoring measures. Thé ethical impact assessment should identify impacts on human rights, in particular but not limited to thé rights of vulnérable groups, labour rights, thé environment and ecosystems, and ethical and social implications in line with thé principles set forth in this Recommendation. 51. Member States and private sector companies should develop due diligence and oversight mechanisms to identify, prevent, mitigate and account for how they address thé impact of Al Systems on human rights, rule of law and inclusive societies. Member States should also be able to assess thé socio-economic impact of Al Systems on poverty and ensure that thé gap between people living in wealth and poverty, as well as thé digital divide among and within countries are not increased with thé massive adoption of Al technologies at présent and in thé future. In order to do this, enforceable transparency protocols should be implemented, corresponding to thé right of access to information, including information of public interest held by private entities. 52. Member States and private sector companies should implement proper measures to monitor ail phases of an Al System life cycle, including thé behaviour of algorithms used for decision-making, thé data, as well as Al actors involved in thé process, especially in public services and where direct end-user interaction is needed. 53. Governments should adopt a regulatory framework that sets out a procédure, particularly for public authorities, to carry out ethical impact assessments on Al Systems to predict conséquences, mitigate risks, avoid harmful conséquences, facilitate citizen participation and address societal challenges. Thé assessment should also establish appropriate oversight mechanisms, including auditability, traceability and explainabilitywhich enable thé assessment of algorithms, data and design processes, as well as include external review ofAI Systems. Ethical impact assessments carried out by public authorities should be transparent and open to thé public. Such assessments should also be multidisciplinary, multi-stakeholder, multicultural, pluralistic and inclusive. Member States are encouraged to put in place mechanisms and tools, for exemple regulatory sandboxes or testing centres, which would enable impact monitoring and-14- assessment in a multidisciplinary and multi-stakeholder fashion. Thé public authorities should be required to monitor thé Al Systems implemented and/or deployed by those authorities by introducing appropriate mechanisms and tools. 54. Member States should establish monitoring and évaluation mechanisms for initiatives and policies related to Al ethics. Possible mechanisms include: a repository covering human rights- compilant and ethical development of Al Systems; a lessons sharing mechanism for Member States to seek feedback from other Member States on their policies and initiatives; a guide for ail Al actors to assess their adhérence to policy recommendations mentioned in this document; and follow-up tools. International human rights law, standards and principles should form part of thé ethical aspects of Al System assessments. POLICY AREA 2: ETHICAL GOVERNANCE AND STEWARDSHIP 55. Member States should ensure that any Al governance mechanism is inclusive, transparent, multidisciplinary, multilatéral (this includes thé possibility of mitigation and redress of harm across borders), and multi-stakeholder. Governance should include aspects of anticipation, protection, monitoring of impact, enforcement and redressai. 56. Member States should ensure that harms caused to users through Al Systems are investigated and redressed, by enacting strong enforcement mechanisms and remédiai actions, to make certain that human rights and thé rule of law are respected in thé digital world as it is in thé physical world. Such mechanisms and actions should include remediation mechanisms provided by private sector companies. Thé auditability and traceability of Al Systems should be promoted to this end. In addition, Member States should strengthen their institutional capacities to deliver on this duty of care and should collaborate with researchers and other stakeholders to investigate, prevent and mitigate any potentially malicious uses ofAI Systems. 57. Member States are encouraged to consider forms of soft governance such as a certification mechanism for Al Systems and thé mutual récognition of their certification, according to thé sensitivity of thé application domain and expected impact on human lives, thé environment and ecosystems, and other ethical considérations set forth in this Recommendation. Such a mechanism might include différent levels of audit of Systems, data, and adhérence to ethical guidelines, and should be validated by authorized parties in each country. At thé same time, such a mechanism must not hinder innovation or disadvantage small and médium enterprises or start- ups by requiring large amounts of paperwork. Thèse mechanisms would also include a regular monitoring component to ensure System robustness and continued integrity and adhérence to ethical guidelines aver thé entire life cycle of thé Al System, requiring re-certification if necessary. 58. Government and public authorities should be required to carry out self-assessment of existing and proposed Al Systems, which in particular, should include thé assessment whether thé adoption of Al is appropriate and, if so, should include further assessment to détermine what thé appropriate method is, as well as assessment as to whether such adoption transgresses any human rights law, standards and principtes. 59. Member States should encourage public entities, private sector companies and civil society organizations to involve différent stakeholders in their Al governance and to consider adding thé rôle of an independent Al Ethics Officer or some other mechanism to oversee ethical impact assessment, auditing and continuous monitoring efforts and ensure ethical guidance of Al Systems. Member States, private sector companies and civil society organizations, with thé-15- support of UNESCO, are encouraged to croate a network of indépendant Al Ethics Officers to give support to this process at national, régional and international levels. 60. Member States should foster thé development of, and access to, a digital ecosystem for ethical development of Al Systems at thé national level, while encouraging international collaboration. Such an ecosystem includes in pari:icular digital technologies and infrastructure, and mechanismsforsharing Al knowledge, as appropriate. In this regard, MemberStates should consider reviewing their policies and regulatory frameworks, including on access to information and open government to reflect Al-specific requirements and promoting mechanisms, such as open repositories for publicly-funded or publicly-held data and source code and data trusts, to support thé safe, fair, légal and ethical sharing of data, among others. 61. Member States should establish mechanisms, in collaboration with international organizations, transnational corporations, académie institutions and civil society, to ensure thé active participation of ail Member States, especially LMICs, in particular LDCs, LLDCs and SIDS, in international discussions concerning Al governance. This can be through thé provision offunds, ensuring equal régional participation, or any other mechanisms. Furthermore, in order to ensure thé inclusiveness ofAI fora, Member States should facilitate thé travel ofAI actors in and out of their territory, especially from LMICs, in particular LDCs, LLDCs and SIDS, for thé purpose of participating in thèse fora. 62. Amendments to existing or élaboration of new national législation addressing Al Systems must comply with international human rights law and promote human rights and fundamental freedoms throughout thé Al system life cycle. Promotion thereof should also take thé form of governance initiatives, good exemplars of collaborative practices regarding Al Systems, and national and international technical and methodological guidelines as Al technologies advance. Diverse sectors, including thé private sector, in their practices regarding Al Systems must respect, protect and promote human rights and fundamental freedoms using existing and new instruments in combination with this Recommendation. 63. Member States should provide mechanisms for human rights and for social and économie impact ofAI monitoring and oversight, and other governance mechanisms such as independent data protection authorities, sectoral oversight, public bodies for thé oversight of acquisition of Al Systems for human rights sensitive use cases, such as criminal justice, law enforcement, welfare, employment, health care, among others, and independentjudiciary Systems. 64. Member States should ensure that governments and multilatéral organizations play a leading rôle in guaranteeing thé safety and security of Al Systems. Specifically, Member States, international organizations and other relevant bodies should develop international standards that describe measurable, testable levels of safety and transparency, so that Systems can be objectively assessed and levels of compliance determined. Furthermore, Member States should continuously support stratégie research on potential safety and security risks of Al technologies and should encourage research into transparency and explainability by putting additional funding into those areas for différent domains and at différent levels, such as technical and natural language. 65. Member States should implement policies to ensure that thé actions of Al actors are consistent with international human rights law, standards and principles throughout thé life cycle of Al Systems, while demonstrating awareness and respect for thé current cultural and social diversities including local customs and religious traditions.-16- 66. Member States should put in place mechanisms to require Al actors to disclose and combat any kind of stereotyping in thé outcomes of Al Systems and data, whether by design or by négligence, and to ensure that training data sets for Al Systems do not foster cultural, économie or social inequalities, préjudice, thé spreading of non-reliable information or thé dissémination of anti-democratic ideas. Particular attention should be given to régions where thé data are scarce. 67. Member States should implement policies to promote and increase diversity in Al development teams and training datasets, and to ensure equal access to Al technologies and their benefits, particularly for marginalized groups, both from rural and urban zones. 68. Mennber States should develop, review and adapt, as appropriate, regutatory and légal frameworks to achieve accountability and responsibility for thé content and outcomes of Al Systems at thé différent phases of their life cycle. Member States should introduce liability frameworks or clarify thé interprétation of existing frameworks to ensure thé attribution of accountability for thé outcomes and behaviour of Al Systems. Furthermore, when developing regulatory frameworks, Member States should, in particular, take into account that ultimate responsibility and accountability must always lie with natural or légal persons and that Al Systems should not be given légal personality themselves. To ensure this, such regulatory frameworks should be consistent with thé principle of human oversight and establish a comprehensive approach focused on thé actors and thé technological processes involved across thé différent stages of thé Al Systems life cycle. 69. Member States should enhance thé capacity of thé judiciary to make décisions related to Al Systems as per thé rule of law and in line with international standards, including in thé use of At Systems in their délibérations, while ensuring that thé principle of human oversight is upheld. 70. In order to establish norms where thèse do not exist, or to adapt existing légal frameworks, Member States should involve ail Al actors (including, but not limited to, researchers, représentatives of civil society and law enforcement, insurers, investors, manufacturers, engineers, lawyers, and users). Thé norms can mature into best practices, taws and régulations. Member States are further encouraged to use mechanisms such as policy prototypes and regulatory sandboxes to accelerate thé development of laws, régulations and policies in line with thé rapid development of new technologies and ensure that laws and régulations can be tested in a safe environment before being officially adopted. Member States should support local governments in thé development of local policies, régulations, and laws in line with national and international légal frameworks. 71. Member States should set clear requirements for Al System transparency and explainability so as to help ensure thé trustworthiness of thé full Al system life cycle. Such requirements should involve thé design and implementation of impact mechanisms that take into considération thé nature of application domain (Is this a high-risk domain such as law enforcement, security, éducation, recruitment and health care?), intended use (What are thé risks in terms of transgression of safety and human rights?), target audience (Who is requesting thé information) and feasibility (Is thé algorithm explainable or not and what are thé trade-offs between accuracy and explainability?) ofeach particular Al System. POLICY AREA 3: DATA POLICY 72. Member States should work to develop data governance stratégies that ensure thé continuai évaluation of thé quality of training data for Al Systems including thé adequacy of thé data collection and sélection processes, proper security and data protection measures, as well-17- as feedback mechanisms to learn from mistakes and share best practices among ail Al actors. Striking a balance between thé collection of metadata and users' privacy should be an upfront goal for such a strategy. 73. Member States should put in place appropriate safeguards to recognize and protect individuals' fundamental right to privacy, including through thé adoption or thé enforcement of législative frameworks that provide appropriate protection, compliant with international law. Member States should strongly encourage ail Al actors, including private sector companies, to follow existing international standards and in particular to carry out privacy impact assessments, as part of ethical impact assessments, which take into account thé wider socio- économie impact of thé intended data processing and to apply privacy by design in their Systems. Privacy should be respected, protected and promoted throughout thé life cycle of Al Systems. 74. Member States should ensure that individuals retain rights over their personal data and are protected by a framework which notably foresees transparency, appropriate safeguards for thé processing of sensitive data, thé highest level of data security, effective and meaningful accountability schemes and mechanisms, thé full enjoyment of data subjects' rights, in particutar thé right to access and thé right to erasure of their personal data in Al Systems, an appropriate level of protection while data are being used for commercial purposes such as enabling micro- targeted advertising, transferred cross-border, and an effective independent oversight as part of a data governance mechanism which respects data sovereignty and balances this with thé benefits of a free flow of information internationally, including access to data. 75. Member States should establish their data policies or équivalent frameworks, or reinforce existing ones, to ensure increased security for personal data and sensitive data, which if disclosed, may cause exceptional damage, injury or hardship to a persan. Examples include data relating to offences, criminal proceedings and convictions, and related security measures; biometric and genetic data; Personal data relating to ethnie or social origin, political opinions, trade union membership, religious and other beliefs, health and sexual life. 76. Member States should use Al Systems to improve access to information and knowledge, including of their data holdings, and address gaps in access to thé Al System life cycle. This can include support to researchers and developers to enhance freedom of expression and access to information, and increased proactive disclosure of officiai data and information. Member States should also promote open data, including through developing open repositories for publicly- funded or publicly-held data and source code. 77. Member States should ensure thé overall quatity and robustness of thé dataset for Al, and exercise vigilance in overseeing their collection and use. This could, if possible and feasible, include investing in thé création of gold standard datasets, including open and trustworthy datasets, which are diverse, constructed on a valid légal basis, including consent of data subjects, when required by law. Standards for annotating datasets should be encouraged, so it can easily be determined how a dataset is gathered and what properties it has. 78. Member States, as also suggested in thé report of thé UNSG's hligh-level Panel on Digital Coopération, with thé support of thé United Nations and UNESCO, should adopt a Digital Gommons approach to data where appropriate, increase interoperability of tools and datasets and interfaces of Systems hosting data, and encourage private sector companies to share thé data they collect as appropriate for research or public benefits. They should also promote public-18- and private efforts to create collaborative platforms to share quality data in trusted and secured data spaces. POLICY AREA 4: DEVELOPMENT AND INTERNATIONAL COOPERATION 79. Member States and transnational corporations should prioritize Al ethics by including discussions of Al-related ethical issues into relevant international, intergovernmental and multi- stakeholder fora. 80. Member States should ensure that thé use of Al in areas of development such as health care, agriculture/food supply, éducation, média, culture, environment, water management, infrastructure management, économie planning and growth, and others, adhères to thé values and principles set forth in this Recommendation. 81. Member States should work through international organizations to provide platforms for international coopération on Al for development, including by contributing expertise, funding, data, demain knowledge, infrastructure, and facilitating collaboration between technical and business experts to tackle challenging development problems, especially for LMICs, in particular LDCs, LLDCs and SIDS. 82. Member States should work to promote international collaboration on Al research and innovation, including research and innovation centres and networks that promote greater participation and leadership of researchers from LMICs and other régions, including LDCs, LLDCs and SIDS. 83. Member States should promote Al ethics research by international organizations and research institutions, as well as transnational corporations, that can be a basis for thé ethical use of At Systems by public and private entities, including research into thé applicability of spécifie ethical frameworks in spécifie cultures and contexts, and thé possibilities to match thèse frameworks to technologically feasible solutions. 84. Member States should encourage international coopération and collaboration in thé field of Al to bridge geo-technological lines. Technological exchanges/consultations should take place between Member States and their populations, between thé public and private sectors, and between and among Member States in thé Global North and Global South. 85. Member States should develop and implement an international légal framework to encourage international coopération between States and other stakeholders paying spécial attention to thé situation of LMICs, in particular LDCs, LLDCs and SIDS. POLICY AREA 5: ENVIRONMENT AND ECOSYSTEMS 86. Member States should assess thé direct and indirect environmental impact throughout thé Al System life cycle, including but not limited to, its carbon footprint, energy consumption, and thé environmental impact of raw material extraction for supporting thé manufacturing of Al technologies. They should ensure compliance of ail Al actors with environmental law, policies, and practices. 87. Member States should introduce incentives, when needed and appropriate, to ensure thé development and adoption of rights-based and ethical Al-powered solutions for disaster risk resilience; thé monitoring, protection and régénération of thé environment and ecosystems; and-19- thé préservation of thé planet. Thèse Al Systems should involve thé participation of local and indigenous communities throughout their life cycle and should support circular economy type approaches and sustainable consumption and production patterns. Some examples include using Al Systems, when needed and appropriate, to: (a) Support thé protection, monitoring, and management of natural resources. (b) Support thé prévention, control, and management of ctimate-related problems. (e) Support a more efficient and sustainable food ecosystem. (d) Support thé accélération of access to and mass adoption of sustainable energy. (e) Enable and promote thé mainstreaming of sustainable infrastructure, sustainable business models, and sustainable finance for sustainable development. (f) Detect pollutants or predict levels of pollution and thus help relevant stakeholders identify, plan and put in place targeted interventions to prevent and reduce pollution and exposure. 88. When choosing Al methods, given thé data-intensive or resource-intensive character of some ofthem and thé respective impact on thé environment, Member States should ensure that At actors, in line with thé principle of proportionality, favour data, energy and resource-efïïcient Al methods. Requirements should be developed to ensure that appropriate évidence is available showing that an Al application will hâve thé intended effect, or that safeguards accompanying an Al application can support thé justification. POLICYAREA 6: GENDER 89. Member States should ensure that digital technologies and artificial intelligence fully contribute to achieve gender equality; and that thé rights and fundamental freedoms of girls and women, including their safety and integrity are not violated at any stage of thé Al System life cycle. Moreover, Ethical Impact Assessments should inctude a transversal gender perspective. 90. Member States should hâve dedicated funds from thé public budgets linked to financing gender-related schemes, ensure that national digital policies include a gender action plan, and develop spécifie policies, e.g. on labour éducation, targeted at supporting girls and women to make sure girls and women are not left out of thé digital economy powered by Al. Spécial investment in providing targeted programmes and gender-specific language, to increase thé opportunities of participation of girls and women in science, technology, engineering, and mathematics (STEM), including information and communication technologies (ICT) disciplines, preparedness, employability, career development and professional growth of girls and women should be considered and implemented. 91. Member States should ensure that thé potential of Al Systems to improve gender equality is realized. They should guarantee that thèse technologies do not contribute to exacerbating thé already wide gender gaps existing in several fields in thé analogue world. This includes thé gender wage gap; thé représentation in certain professions and activities gap; thé lack of représentation at thé top management positions, boards of directors, or research teams in thé Al field; thé éducation gap; digital/AI access, adoption, usage and affordability gap; thé unequal distribution of unpaid work and of thé caring responsibitities in our societies.-20- 92. Member States should ensure that gender stereotyping, and discriminatory biases are not translated into thé Ai Systems. Efforts are necessary to avoid thé compounding négative effect of technotogical divides in achieving genderequality and avoiding violence against girls and women, and ail other types of gender identifies. 93. Member States should encourage female entrepreneurship, participation and engagement in ail stages otan Al System life cycle by offering and promoting économie, regulatory incentives, among other incentives and support schemes, as well as policies that aim at a balanced gender participation in Al research in academia, gender représentation on digital/AI companies top management positions, board ofdirectors, or research teams. Governments should ensure public funds (on innovation, research and technologies) are channelled to inclusive programmes and companies, with clear gender représentation, and that private funds are encouraged through affirmative action principles. Moreover, policies on harassment-free environments should be developed and enforced together with thé encouragement of thé transfer of best practices on how to promote diversity throughout thé Al System life cycle. 94. UNESCO can help form a repository of best practices for incentivizing thé participation of women and under-represented groups on ail stages of thé Al life cycle. POLICY AREA 7: CULTURE 95. Member States are encouraged to incorporate Al Systems where appropriate in thé préservation, enrichment, understanding, promotion and accessibility of tangible, documentary and intangible cultural héritage, including endangered languages as well as indigenous languages and knowledge, for example by introducing or updating educational programmes related to thé application ofAI Systems in thèse areas where appropriate and ensuring a participatory approach, targeted at institutions and thé public. 96. Member States are encouragea to examine and address thé cultural impact of Al Systems, especially Natural Language Processing applications such as automated translation and voice assistants on thé nuances ofhuman language and expression. Such assessments should provide input for thé design and implementation of stratégies that maximize thé benefits from thèse Systems by bridging cultural gaps and increasing human understanding, as well as négative implications such as thé réduction of use, which could lead to thé disappearance of endangered languages, local dialects, and tonal and cultural variations associated with human language and expression. 97. Member States should promote Al éducation and digital training for artists and créative professionals to assess thé suitability of Al technologies for use in their profession as Al technologies are being used to create, produce, distribute and broadcast a variety of cultural goods and services, bearing in mind thé importance of preserving cultural héritage, diversity and artistic freedom. 98. Member States should promote awareness and évaluation ofAI tools among local cultural industries and small and médium enterprises working in thé field of culture, to avoid thé risk of concentration in thé cultural market. 99. Member States should engage large technology companies and other stakeholders to promote a diverse supply and plural access to cultural expressions, and in particular to ensure that algorithmic recommendation enhances thé visibility and discoverability of local content.-21 - 100. Member States should foster new research at thé intersection between Al and intellectual property, for example to détermine who are thé rights-holders of thé works created by means of Al technologies among thé différent stakeholders throughout thé Al life cycle. 101. Member States should encourage muséums, galleries, libraries and archives at thé national level to develop and use Al Systems to highlight their collections, strengthen their databases and grant access to them for their users. POLICY AREA 8: EDUCATION AND RESEARCH 102. Member States should work with international organizations, private and non-governmental entities to provide adéquate Al literacy éducation to thé public in ail countries in order to empower people and reduce thé digital divide and digital access inequalities resulting from thé wide adoption ofAI Systems. 103. Member States should promote thé acquisition of "prerequisite skills" for Al éducation, such as basic literacy, numeracy, coding and digital skills, and média and information literacy, as well as critical thinking, teamwork, communication, socio-emotional, and Al ethics skills, especially in countries where there are notable gaps in thé éducation of thèse skills. 104. Member States should promote général awareness programmes about Al developments, including on thé opportunities and challenges brought about by Al technologies. Thèse programmes should be accessible to non-technical as wetl as technical groups. 105. Member States should encourage research initiatives on thé responsible use of Al technologies in teaching, teacher training and e-learning among other topics, in a way that enhances opportunities and mitigates thé challenges and risks involved in this area. Thé initiatives should be accompanied by an adéquate assessment of thé quality of éducation and of impact on students and teachers of thé use of Al technologies. Member States should also ensure that Al technologies empower students and teachers and enhance their expérience, bearing in mind that emotional and social aspects and thé value of traditional forms of éducation are vital in thé teacher-student and student-student relationships, and should be considered when discussing thé adoption ofAI technologies in éducation. 106. Member States should promote thé participation of girls and women, diverse ethnicities and cultures, and persans with disabilities, in Al éducation programmes at ail levels, as well as thé monitoring and sharing of best practices in this regard with other Member States. 107. Member States should develop, in accordance with their national éducation programmes and traditions, Al ethics curricula for ail levels, and promote cross-collaboration between Al technical skills éducation and humanistic, ethical and social aspects of Al éducation. Online courses and digital resources of Al ethics éducation should be developed in local languages, especially in accessible formats for persans with disabilities. 108. Member States should promote Al ethics research either through investing in such research or by creating incentives for thé public and private sectors to invest in this area. 109. Member States should ensure that Al researchers are trained in research ethics and require them to include ethicat considérations in their designs, products and publications, especially in thé analyses of thé datasets they use, how they are annotated and thé quality and thé scope of thé results.-22- 110. Member States should encourage private sector companies to facilitate thé access of scientific community to their data for research, especially in LMICs, in particular LDCs, LLDCs and SIDS. This access should not be at thé expense of privacy. 111. Member States should promote gender diversity in Al research in academia and industry by offering incentives to girls and women to enter thé field, putting in place mechanisms to fight gender stereotyping and harassment within thé Al research community, and encouraging académie and private entities to share best practices on how to enhance gender diversity. 112. To ensure a critical évaluation ofAI research and proper monitoring of potential misuses or adverse effects, Member States should ensure that any future developments with regards to Al technologies should be based on rigorous scientific research, and promote interdisciplinary Al research by including disciplines otherthan science, technology, engineering, and mathematics (STEM), such as cultural studies, éducation, ethics, international relations, law, linguistics, philosophy, and political science. 113. Recognizing that Al technologies présent gréât opportunities to help advance scientific knowledge and practice, especially in traditionally model-driven disciplines, Member States should encourage scientific communities to be aware of thé benefits, limits and risks oftheir use; this includes attempting to ensure that conclusions drawn from data-driven approaches are robust and sound. Furthermore, Member States should welcome and support thé rôle of thé scientific community in contributing to policy, and in cultivating awareness of thé strengths and weaknesses of Al technologies. POLICY AREA 9: ECONOMY AND LABOUR 114. Member States should assess and address thé impact ofAI Systems on labour markets and its implications for éducation requirements, in ail countries and with spécial emphasis on countries where thé economy is labour-intensive. This can include thé introduction ofa wider range of'core" and interdisciplinary skills at ail éducation levels to provide current workers and new générations a fair chance of finding jobs in a rapidly changing market and to ensure their awareness of thé ethical aspects of Al Systems. Skills such as "learning how to learn", communication, critical thinking, teamwork, empathy, and thé ability to transfer one's knowledge across domains, should be taught alongside specialist, technical skills, as well as low-skilled tasks such as labelling datasets. Being transparent about what skills are in demand and updating curricula around thèse are key. 115. Member States should support collaboration agreements among governments, académie institutions, industry, workers' organizations and civil society to bridge thé gap of skillset requirements to align training programmes and stratégies with thé implications of thé future of work and thé needs of industry. Project-based teaching and learning approaches for Al should be promoted, allowing for partnerships between private sector companies, universities and research centres. 116. Member States should work with private sector companies, civil society organizations and other stakeholders, including workers and unions to ensure a fair transition for at-risk employées. This includes putting in place upskilling and reskilling programmes, flnding effective mechanisms of retaining employées during those transition periods, and exploring "safety net" programmes for those who cannot be retrained. Member States should develop and implement programmes to research and address thé challenges identified that could include upskilling and reskilling, enhanced social protection, proactive industry policies and interventions, tax benefits, new-23- taxations forms, among others. Tax régimes and other relevant régulations should be carefully examined and changea if needed to counteract thé conséquences of unemployment caused by Al-based automation. 117. Member States should encourage and support: researchers to analyse thé impact of Al Systems on thé local labour environment in order to anticipate future trends and challenges. Thèse studies should investigate thé impact ofAI Systems on économie, social and géographie sectors, as well as on human-robot interactions and human-human relationships, in order to advise on reskilling and redeployment best practices. 118. Member States should devise mechanisms to prevent thé monopolization of Al Systems throughout their life cycle and thé resulting inequalities, whether thèse are data, research, technology, market or other monopolies. Member States should assess relevant markets, and regulate and intervene if such monopolies exist, taking into account that, due to a lack of infrastructure, human capacity and régulations, LMICs, in particular LDCs, LLDCs and SIDS are more exposed and vulnérable to exploitation by large technology companies. POLICY AREA 10: HEALTH AND SOCIAL WELL-BEING 119. Member States should endeavour to employ effective Al Systems for improving human health and protecting thé right to life, while building and maintaining international solidarity to tackle global health risks and uncertainties, and ensure that their deployment of Al Systems in health care be consistent with international law and international human rights law, standards and principles. Member States should ensure that actors involved in health care Al Systems take into considération thé importance of a patient's relationships with their family and with health care staff. 120. Member States should regulate thé development and deployment of Al Systems related to health in général and mental health in particular to ensure that they are safe, effective, efficient, scientifically and medically sound. Moreover, in thé related area of digital health interventions, Member States are strongly encouraged to actively involve patients and their représentatives in ail relevant steps of thé development of thé System. 121. Member States should pay particular attention in regulating prédiction, détection and treatment solutions for health care in Al applications by: (a) ensuring oversight to minimize bias; (b) ensuring that thé professional, thé patient, caregiver or service user is included as a "demain expert" in thé team when developing thé algorithms; (e) paying due attention to privacy because of thé potential need of being constantly monitored; (d) ensuring that those whose data is being analysed are aware of and provide informed consent to thé tracking and analysis oftheir data; and (e) ensuring thé human care and final décision of diagnosis and treatment are taken by humans while acknowledging that Al Systems can also assist in their work.-24- 122. Member States should establish research on thé effects and régulation of potential harms to mental health related to Al Systems, such as higher degrees of dépression, anxiety, social isolation, developing addiction, trafficking and radicalization, misinformation, among others. 123. Member States should develop guidelines for human-robot interactions and their impact on human-human relationships, based on research and directed at thé future development of robots, with spécial attention to thé mental and physical health of human beings, especially regarding robots in health care and thé care for older persans and persans with disabilities, and regarding educational robots, toy robots, chatbots, and companion robots for children and adults. Furthermore, assistance of Al technologies should be applied to increase thé safety and ergonomie use of robots, including in a human-robot working environment. 124. Member States should ensure that human-robot interactions comply with thé same values and principles that apply to any other Al Systems, including human rights, thé promotion of diversity in relationships, and thé protection of vulnérable groups. 125. Member States should protect thé right of users to easily identify whether they are interacting with a living being, orwith an Al System imitating human or animal characteristics. 126. Member States should implement policies to raise awareness about thé anthropomorphization of Al technologies, including in thé language used to mention them, and assess thé manifestations, ethical implications and possible limitations of such anthropomorphization in particular in thé context of robot-human interaction and especially when children are involved. 127. Member States should encourage and promote collaborative research into thé effects of long-term interaction of people with Al Systems, paying particular attention to thé psychological and cognitive impact that thèse Systems can hâve on children and young people. This should be done using multiple norms, principles, protocols, disciplinary approaches, and assessment of thé modification of behaviours and habits, as well as careful évaluation of thé downstream cultural and societal impacts. 128. Member States, as well as ail stakeholders, should put in place mechanisms to meaningfully engage children and young people in conversations, debates, and decision-making with regards to thé impact ofAI Systems on their lives and futures. 129. Member States should promote thé accountable use ofAI Systems to counter hâte speech in thé online demain and disinformation and also to ensure that Al Systems are not used to produce and spread such content, particularly in times of élections. 130. Member States should create enabling environments for média to hâve thé rights and resources to effectively report on thé benefits and harms ofAI Systems, and also to make use of Al Systems in their reporting. V. MONITORING AND EVALUATION 131. Member States should, according to their spécifie conditions, governing structures and constitutional provisions, credibly and transparently monitor and evaluate policies, programmes and mechanisms related to ethics of Al using a combination of quantitative and qualitative approaches. In support to Member States, UNESCO can contribute by:-25- (a) developing a globally accepted methodology for Ethical Impact Assessment (EIA) of Al technologies, including guidance for its implementation in ail stages of thé Al system life cycle, based on rigorous scientific research; (b) developing a readiness methodology to assist Member States in identifying their status at spécifie moments of their readiness trajectory along a continuum of dimensions; (e) developing a globally accepted methodology to evatuate ex ante and ex post thé effectiveness and efficiency of thé policies for Al ethics and incentives against defined objectives; (d) strengthening thé research- and evidence-based analysis of and reporting on policies regarding Al ethics, including thé publication of a comparative index; and (e) collecting and disseminating progress, innovations, research reports, scientific publications, data and statistics regarding policies for Al ethics, to support sharing best practices and mutual learning, and to advance thé implementation of this Recommendation. 132. Processes for monitoring and évaluation should ensure broad participation of relevant stakeholders, including, but not limited to, people of différent âge groups, girls and women, persons with disabilities, disadvantaged, marginatized and vulnérable populations, indigenous communities, as well as people from diverse socio-economic backgrounds. Social, cultural, and gender diversity must be ensured, with a view to improving learning processes and strengthening thé connections between findings, decision-making, transparency and accountability for results. 133. In thé interests of promoting best policies and practices related to ethics of Al, appropriate tools and indicators should be developed for assessing thé effectiveness and efficiency thereof against agreed standards, priorities and targets, including spécifie targets for persans belonging to disadvantaged, marginalized and vulnérable groups, as well as thé impact of Al Systems at individual and societal levels. Thé monitoring and assessment of thé impact of Al Systems and related Al ethics policies and practices should be carried out continuously in a systematic way. This should be based on internationally agreed frameworks and involve évaluations ofprivate and public institutions, providers and programmes, including self-evaluations, as well as tracer studies and thé development of sets of indicators. Data collection and processing should be conducted in accordance with national législation on data protection and data privacy. 134. Thé possible mechanisms for monitoring and évaluation may include an Al ethics observatory, or contributions to existing initiatives by addressing adhérence to ethical principles across UNESCO's areas of compétence, an experience-sharing mechanism for Member States to provide feedback on each other's initiatives, Al regulatory sandboxes, and an assessment guide for ail Al actors to evaluate their adhérence to policy recommendations mentioned in this document. VI. UTILIZATION AND EXPLOITATION 0F THE PRESENT RECOMMENDATION 135. Member States and ail other stakeholders as identified in this Recommendation must respect, promote and protect thé ethical principles and standards regarding Al that are identified in this Recommendation, and should take ail feasible steps to give effect to its policy recommendations.-26- 136. Member States should strive to extend and complément their own action in respect of this Recommendation, by cooperating with ail national and international governmental and non- governmental organizations, as well as transnational corporations and scientific organizations, whose activities fall within thé scope and objectives of this Recommendation. Thé development of a globally accepted Ethical Impact Assessment methodology and thé establishment of national commissions for thé ethics oftechnology can be important instruments forthis. VII. PROMOTION 0F THE PRESENT RECOMMENDATION 137. UNESCO has thé vocation to be thé principal United Nations agency to promote and disseminate this Recommendation, and accordingly shall work in collaboration with other United Nations entities, including but not limited to thé United Nations Secretary-General's hligh-level Panel on Digital Coopération, thé World Commission on thé Ethics of Scientific Knowledge and Technology (COMEST), thé International Bioethics Committee (IBC), thé Intergovernmental Bioethics Committee (IGBC), thé International Télécommunication Union (ITU), thé International Labour Organization (ILO), thé World Intellectual Property Organization (WIPO), thé United Nations Children's Fund (UNICEF), UN Women, thé United Nations Industrial Development Organization (UNIDO), thé World Trade Organization (WTO), and other relevant United Nations entities concerned with thé ethics ofAI. 138. UNESCO shall also work in collaboration with other international and régional organizations, including but not limited to thé African Union (AU), thé Alianza del Pacifico, thé Association ofAfrican Universities (AAU), thé Association of Southeast Asian Nations (ASEAN), thé Caribbean Community (CARICOM), thé Caribbean Télécommunications Union, thé Caribbean Public Services Association, thé Common Market for Eastern and Southern Africa (COMESA), thé Community of Latin American and Caribbean States (CELAC), thé Council of Europe (CoE), thé Economie Community of West African States (ECOWAS), thé Eurasian Economie Union (EAEU), thé European Union (EU), thé International Association of Universities (IAU), thé Organisation for Economie Co-operation and Development (OECD), thé Organization for Security and Co-operation in Europe (OSCE), thé South Asian Association for Régional Coopération (SAARC), thé Southern African Development Community (SADC), thé Southern Common Market (MERCOSUR), as well as thé Institute of Electrical and Electronic Engineers (IEEE), thé International Organization for Standardization (ISO), and international financing institutions such as thé World Bank, thé InterAmerican Development Bank, and thé African Development Bank. 139. Even though, within UNESCO, thé mandate to promote and protect falls within thé authority of governments and intergovernmental bodies, civil society will be an important actor to advocate for thé public sector's interests and therefore UNESCO needs to ensure and promote its legitimacy. VIII. FINAL PROVISIONS 140. This Recommendation needs to be understood as a whole, and thé foundational values and principles are to be understood as complementary and interrelated. 141. Nothing in this Recommendation may be interpreted as approval for any State, other social actor, group, or person to engage in any activity or perform any act contrary to human rights, fundamental freedoms, human dignity and concern for thé environment and ecosystems.

Epub Document
Source document
Record
Title
Preliminary report on the first draft of the Recommendation on the Ethics of Artificial Intelligence
Collation
33 pages
Material type
Year of publication
Document code
CL/4327
Imprint
Country of publication
France
Language
English
Also available in
https://unesdoc.unesco.org/ark:/48223/pf0000374266_fre
Media type
Electronic
Source
UNESCO
Catalog Number
0000374266