Saïda BELOUALI is a professor of applied ethics at Mohammed Premier University in Oujda, Morocco. She specialises in ethics applied to artificial intelligence and is co-director of the House of Artificial Intelligence in Oujda. She is currently an expert at UNESCO for the application of the recommendation on the ethics of AI in Morocco.
Attached to the cultural dimension of ethics applied to AI, she regularly contributes to the reflections of the Global AI Ethics Institute and several other international think tanks and research bodies, and is involved in numerous initiatives to promote this subject on the African continent.
1. The ethics of AI seem to be a central concern. What do you think is the reason for this interest? Isn't this concept overused?
At the dawn of major civilizational changes and disruptive advances that break with the usual models, we need to review our moral copy, to check whether our ethical ramparts have been de-fixed or damaged. Transformational progress is forcing us to ask ourselves fundamental questions to check whether our moral safety nets are still working or whether we need to readjust and devise other ethical frameworks and regulations. So-called 'intelligent' advances, and above all their unbridled pace, are creating new realities and raising new issues that call on us to be responsible. While bringing real opportunities in many areas, these technologies can present high risks throughout their design cycle, from the sometimes deplorable conditions imposed on click workers to the collection and processing of data and the invasion of privacy and fundamental rights of individuals.
AI ethics is about our ability as humans to maintain control and be accountable for our actions. We remain moral agents, and the choices we make impact us and others, society, the environment, and so on. When we design, use, and deploy intelligent systems, we are making a commitment, and the action we take in the face of a situation presenting moral challenges makes us accountable and responsible. Ethics is undoubtedly an overused term these days, but it is still the science of behavior that controls our actions in specific situations when decisions need to be taken in accordance with a specific set of values and principles.
2. In your opinion, what are the appropriate mechanisms for managing the impact of these technologies? At what level should they be considered: local or international ?
AI expands the horizon of economic and societal possibilities, but it can also be the source of intrusive and discriminatory practices that should require restrictions. Not all countries have a law specifically dedicated to AI, and obligations in this area are covered by a variety of regulations. Legislation may apply to AI, specifically that which covers the spectrum of human rights. In this case, cyberlaws and, in particular, laws relating to the protection of personal data are useful and can respond to AI-related situations while waiting for a global framework dedicated to AI if countries so wish. In most cases today, the framework is provided by laws that already exist and are not necessarily specific to AI systems.
For AI innovation to remain "responsible", it is important that a set of principles is respected throughout the life cycle of the intelligent system, as recommended by several entities and organizations. It is also important to understand that AI is a phenomenon whose impacts recognize absolutely no borders. International efforts are crucial. In this sense, ethical and regulatory vigilance should be both local and international. To ensure that it is relevant, it is important to agree on a frame of reference.
Certain uses may be prohibited in one place but not considered responsible in another. Concepts are semantic aggregations and constructions. What is considered responsible or non-responsible innovation in a given community? A lexical analysis and a form of "particularization of meaning" are necessary to reach agreement on the different representations we are manipulating. If we are to preserve the universality we share, particularly in relation to such disruptive technologies, it is essential that we work on contextualization and adaptation.
Hence the need, it seems to us, for international collaboration in this area to avoid, in particular, the major disparities in regulation that could create 'legislative havens' that become popular destinations because they are less demanding in terms of the supervision of these technologies.
3. You are the lead expert for the Readiness Assessment Methodology in Morocco, which involves operationalizing UNESCO's Recommendation on the Ethics of Artificial Intelligence. Can you explain what this is all about?
In November 2021, 193 UNESCO countries adopted the Recommendation on the Ethics of Artificial Intelligence. This is a global initiative that aims to provide states with a normative framework for the supervision of AI while taking advantage of its potential opportunities. More than fifty countries are currently in the process of translating this Recommendation into institutional and regulatory frameworks.
To make the Recommendation operational and to ensure that the principles laid down become effective, UNESCO has developed specific tools, including the AI Readiness Assessment Methodology (RAM). The assessment exercise uses several qualitative and quantitative indicators to map the national ecosystem concerned in order to assess the country's level of preparedness for AI. The diagnosis makes it possible to identify potential and strengths, while also highlighting areas requiring improvement and development. The methodology used to assess UNESCO's state of readiness is made up of several indicators grouped into five different areas : legal, social and cultural, scientific and educational, economic, technical and infrastructural.
As part of a support project, this 360° diagnosis is being carried out by UNESCO in four pilot countries: Brazil, Chile, Morocco and Senegal in order to determine their various states of readiness, particularly institutional and regulatory, and to assess the development potential of AI in each of these countries. The results of these diagnoses will enable the countries concerned to devise new policies on the ethical governance of AI and UNESCO to devise appropriate capacity-building support mechanisms in keeping with the spirit of the Recommendation.
The experiences of the RAM exercise in the pilot countries were recently presented at the WorldForum on the Ethics of AI organized by the Slovenian government and UNESCO on 5and 6 February 2024.
4. What is the main lesson to be learned from this exercise?
Controlled and supervised AI could enable different communities to make definite progress in education, while adaptive and personalized pathways could lead to greater inclusion of the population; health and agriculture will also be areas that will be transformed by intelligent technologies. The various countries, particularly in the Deep South, must be proactive to avoid suffering the impacts of AI without benefiting from its opportunities.
Furthermore, major advances in AI must be contextual and situational. They must respond to local issues. The changes they will bring about must represent a real opportunity for local populations. Above all, AI must not be synonymous with"click work" for some and unprecedented opportunities for others.
RAM's experience confirms the idea that all nations can benefit from this unprecedented boom, provided that they are creative and dare to imagine new responses to new situations.
Further readings
Recommendation on the Ethics of Artificial Intelligence. UNESCO,2021. https://en.unesco.org/about-us/legal-affairs/recommendation-ethics-artificial-intelligence
Readiness assessment methodology: a tool of the Recommendation on the Ethics of Artificial Intelligence. UNESCO, 2023. https://unesdoc.unesco.org/ark:/48223/pf0000385198
Mahamadou Simpara. Morocco Among Top 5 African Countries in Terms of AI Readiness. Morocco World News, Dec. 26, 2023. https://www.moroccoworldnews.com/2023/12/359773/morocco-among-top-5-african-countries-in-terms-of-ai-readiness
