An unprecedented recommendation from UNESCO, but with limited impact

Social networks, GPS routing, bank loans… Artificial intelligence is at the center of our lives, sometimes without even realizing it. though, The dangers of uncontrolled or harmful use, but also of unification of thought are very real. This is where UNESCO comes in. On November 24, 2021, at the 41st General Conference, the 193 Member States adopted a 28-page recommendation on the ethics of artificial intelligence. The text is unprecedented in its breadth, but remains limited in its true scope.

Identify potential and identify risks

It is accelerating in all sectors. Medicine, economics, finance, research, defense… AI has the potential to help humanity be more productive and efficient. corn If the benefits of this technology are profitable, it also has many potential risks.

We are concerned because a lot of decisions are now being made on the basis of AI: who has access to credit, who can get into which university, who can get a job interview… Even in the judicial system we do a lot of sorting with AI.
Gabriela Ramos, Director-General of UNESCO’s Social and Human Sciences Pole

The first danger: bias. Artificial intelligence works through algorithms, which are generated using databases. However, most of the time this data is not very comprehensive, even discrimination. This is the case, for example, at the crop level. The digital divide between North and South countries creates imbalances in the data collected. And therefore, AI is biased and does not represent specific cultural practicesAnd even some languages. And biases are not only a result of the algorithms themselves, but also the result of the way they are thought, especially according to who thinks about them. In companies specializing in artificial intelligence innovation, Women are an extreme minority: they make up only 22% of tech professionalsAccording to UNESCO. Consequences: a significant risk of gender bias, confirmed by databases that reflect a patriarchal society. “The problem with AI is that if the database contains information only about men, the technology will display decisions that benefit men, forgetting all the needs of women. That is whyWe want databases to be representative and inclusive, so that minorities are represented“, confirms Mariagrazia Squichiarini, Head of the Social and Human Sciences Sector at UNESCO.

Another danger is the uncontrolled or malicious use of artificial intelligence. violation of privacy and disrespect for human rights, MonotheismAnd audience monitoring…if poorly mastered or used for nefarious purposes, AI could have worrying consequences. These expectations become even more worrisome when we look at the machine learning capability of artificial intelligence, or “machine learning.” According to Dafna Venholz, a staff member in the Department of Bioethics and Ethics of Science at UNESCO, “It is always necessary to have the possibility of knowing who the origin of the algorithm was: _The responsibility for every decision should always rest with the human being, not the machine_“.

Finally, it is also important to consider The danger posed by digital technology to the environment. And AI can help better understand climate issuesIts use can be hazardous to the environment. According to UNESCO, Data mining consumes nearly 10% of the world’s energy, and data centers could consume 20% of the world’s total electricity by 2025.

Promoting ethical artificial intelligence, which is the (non-resolute) goal of UNESCO

NS Ensure that AI is used for the public goodAnd so that everyone can take advantage of their advantages, UNESCO has taken up this topic. Since 2018, The Foundation has convened 24 independent international experts from all continents, 400 associated schools, 200 university chairs and more than 50,000 contributions from researchers, NGOs and civil society actors. To help him implement his recommendation.

Since our species stands at a critical crossroads, where the creation of artificial intelligence systems can bring about such tremendous progress as the creation of many possibilities of destruction, the UNESCO Recommendation on the Ethics of Artificial Intelligence makes it possible to place governments, specialists and all individuals on an equal footing. Make the decision to best lead our community.
Dr. David Leslie, Ethics Officer at the Alan Turing Institute

Consultation with these experts results in an unpublished text, the cars It is the first global agreement in this field. bone It is also a way for the institution to regain its importance in global diplomacyWhile the United States and Israel left their ranks. At the head of UNESCO since 2017, Re-elected this yearIts director, Audrey Azoulay, has staked everything on the AI ​​master file. cum in heartA topic that promises to become essentialThere is no doubt that the head of the United Nations is hoping to restore the institution to its glory.

The recommendation anyway has the merit of the list, and it holds universal values ​​and principles, in the hope of making AI more ethical. According to Audrey Azoulay, Director-General of UNESCO, The recommendation is based on Respect and protect human rights and human dignity Need to protect the environment and ecosystems, ensuring that Ensuring inclusion and diversity, and _Caring for peaceful, just and independent societies_“.

concretely, The text directs the states on how to act to enforce this ethical framework. In the field of fundamental freedoms, The Recommendation prohibits any recourse to social categorization and mass surveillance; In the field of data governance sets the rules so that every citizen can retain control over the data they provide, and can access it at any time and if they wish to delete it; The text also calls on states to do so Creation of independent bodies to which any citizen can contact for the purpose of requesting information, but also for the purpose of recourse to assert their rights”Lists of Audrey Azoulay.

Without restrictions, but rather a guiding text for states and citizens

On November 24 These principles and means of action have been agreed by the 193 UNESCO Member States. Although encouraging, this consensus must be kept in mind, because The text is just a recommendation, without any authority. “That’s right, if a country doesn’t want to act, it won’t act. We don’t have the means to force them, Gabriela Ramos admits, _But the most important thing is to say that states have a responsibility to protect their citizens._“ And The Foundation will require countries to submit regular reports on developments and practices in the field of artificial intelligenceMeasure is still limited. One might suspect, for example, that the Chinese government, At the origin of social rating scalesIt can claim certain ethics in its use of artificial intelligence, or it does not reflect its decisions. What is more, Israel and the United States, which focus a number of companies associated with the technological sector of artificial intelligence, are not located because they are outside UNESCO.

So far, companies can do whatever they want. With this recommendation, we give governments the tools and authority to set the rules. This is their role and responsibility. And then private companies will have no choice: they will have to follow the rules, that’s all.
Daphna Wenholz, Department of Bioethics and Ethics of Science at UNESCO

The UNESCO Recommendation is above all a tool used by governments. It should enable them to define a framework to guide their policies in this area. Encouragement point: About 20 countries indicated that they had already started taking measures when the text was adopted.

So it is a tool of the states, but above all, and this is their main interest of the citizens. This recommendation is also for them, and will allow them to get acquainted with the technology that is present in their everyday life. “The goal is for people to be better able to decide whether or not they want to provide their information., but also that programmers ask themselves a question about the ethics of what they do, and the impact of their work on the lives of others.”, Mariagrazia Squicciarini points. Knowing AI, how it works and its dangers is the only way to keep us in control of this technology. In order to keep the extension The ability to choose what happens to our personal data.