Ethics and AI | Part 3

Since AI begins in the minds of men, it is in the minds of men that the defences of ethics must be constructed

1. UNESCO Recommendation on the Ethics of Artificial Intelligence

In November 2021, UNESCO adopted the Recommendation on the Ethics of Artificial Intelligence, marking its first global standard on AI ethics.¹ This recommendation is applicable to all its 194 member states and emphasises the protection of human rights and dignity as fundamental principles. It advocates for transparency, fairness, and the necessity of human oversight of AI systems. The recommendation outlines policy action areas that could guide policymakers in translating these core values into actionable frameworks across diverse sectors, including data governance, education, health, and social wellbeing.

One of the merits of UNESCO’s recommendation is that it describes the specific challenges brought by AI to ethics in addition to what other technologies may have brought. Indeed, the document enlists a comprehensive list of new types of ethical issues that AI systems raise: impact on decision-making, employment and labour, social interaction, health care, education, media, access to information, digital divide, personal data and consumer protection, environment, democracy, rule of law, security and policing, dual use, and human rights and fundamental freedoms, including freedom of expression, privacy and non-discrimination. 

According to UNESCO, less visible threats come from the potential of AI algorithms to reproduce and reinforce existing biases, and thus to exacerbate already existing forms of discrimination, prejudice and stereotyping. The impact on labour is also mentioned as a result of the capacity of AI systems to perform tasks which previously only living beings could do, and which were in some cases even limited to human beings only.²

UNESCO’s Recommendation outlines several key principles aimed at ensuring the ethical development and deployment of AI technologies throughout the AI lifecycle.

Those key principles could be clustered into two categories.

Impact on human rights and fundamental freedoms as defined by international law:

• Human rights and dignity: The protection of human rights and dignity is fundamental. All AI systems must respect, protect, and promote these rights throughout their lifecycle.
• Fairness and non-discrimination: AI actors must ensure fairness and non-discrimination in AI systems, actively working to minimize biases and ensure equitable access to AI benefits for all individuals.
• Privacy protection: Privacy must be safeguarded throughout the lifecycle of AI systems, with robust data protection frameworks established to prevent misuse of personal data.
• Necessity and proportionality: The use of AI systems should be governed by the principle of necessity and proportionality, ensuring that AI applications are justified and not used for harmful purposes such as social scoring or mass surveillance.

Impact on other aspects of social and economic life:

• Promotion of diversity: the importance of respecting and promoting diversity and inclusiveness in all aspects of AI development and application.
• Inclusive governance: Governance mechanisms for AI must be inclusive, transparent, multidisciplinary, multilateral, and involve diverse stakeholders to ensure comprehensive oversight.
• Continuous impact assessment: There should be ongoing assessments of the human, social, cultural, economic, and environmental impacts of AI technologies to ensure alignment with sustainable development goals.
• Public engagement and education: Promoting public understanding of AI through education, civic engagement, and training in digital skills.
• Transparency and explainability: AI systems should be transparent and explainable, allowing users to understand how decisions are made. 
• Accountability: There must be clear accountability mechanisms in place for AI systems, ensuring that ultimate responsibility remains with human actors rather than being displaced by technology.
• Safety and security: Any threats posed by these systems must be addressed to safeguard human well-being and environmental health.

Notably, UNESCO also launched a Global AI Ethics and Governance Observatory, which claims to be “a tool for ethical impact assessment.”. It is a platform that aims to provide resources for interested stakeholders to navigate amidst the ethical challenges posed by AI technologies. As it is the fruit of the UNESCO recommendation, the Observatory promotes collaboration, knowledge sharing, and capacity building. 

More interesting, the Observatory brings to attention case studies „to expose AI systems and tools […] released to users without clear and transparent analysis of the potential risks and how they might be mitigated, even when such risks were foreseeable”³. Such an example is the study entitled “Foundation models such as ChatGPT through the prism of the UNESCO Recommendation on the Ethics of Artificial Intelligence”⁴ which signals the danger of those generative AI tools, branded as “experimental” by their developers, which propose large language models that have routinely generated inaccurate, misleading, or discriminatory content.

UNESCO’s Recommendation prompted us to enter into the work of the United Nations system. Not surprisingly, this does not mean one single door because various entities in the system do not coalesce for a single vision but keep running on different tracks, at the risk of duplication and overlapping.

[Part 3 of 6-part series]

---

1. The recommendation was published under the code SHS/BIOI/PI/2021/1, 43 pages and illustrations and is available online. ↩︎
2. UNESCO Recommendation on the Ethics of Artificial Intelligence, section 1, p. 10. ↩︎
3. https://www.unesco.org/ethics-ai/en/eia ↩︎
4. https://unesdoc.unesco.org/ark:/48223/pf0000385629, doc. SHS/2023/PI/H/12. ↩︎

---

Dr Petru Dumitriu was a member of the Joint Inspection Unit (JIU) of the UN system and former ambassador of the Council of Europe to the United Nations Office at Geneva. He is the author of the JIU reports on ‘Knowledge Management in the United Nations System’, ‘The United Nations – Private Sector Partnership Arrangements in the Context of the 2030 Agenda’, ‘Strengthening Policy Research Uptake’, “Cloud Computing in the United Nations System”, and “Policies and Platforms in Support of Learning”. He received the Knowledge Management Award in 2017 and the Sustainable Development Award in 2019 for his reports. He is also the author of the Multilateral Diplomacy online course at DiploFoundation.