How should intelligent machines "behave"?

Monday, 31 december 2018 | Redacción CEU

"I, robot", "Blade Runner", "Eva",... All these films have a common feature, they present, with more or less depth, the topic of artificial intelligence. As always, when this matter is brought up, a discussion soon arises: where are the limits to the development of this technology and what can its real potential be. AI offers substantial benefits to both individuals and the whole society, which is why its development is in full swing. However, as this universe of fiction has predicted, depending on the use that is given to this technology, its impact may also be negative. Given the fact that this is not a minor issue, the European Union has started to work on a draft of ethical principles for artificial intelligence. What rules should guide the "behavior" of intelligent machines? What does the European Commission suggest?

Not long ago, on this blog, we wrote about the importance of putting an ethical approach ahead of the development of new technologies. The Artificial intelligence industry is one of the most innovative areas which has a great transforming force and advances at a faster pace. Brussels is aware of the fact that AI means a great opportunity to increase prosperity and growth, improve the life quality and well-being of citizens and it may even be useful to face global challenges such as those which were set out by SDGs. However, it also recognizes that the progress of this technology can involve multiple risks. For this reason, the European Commission has been working on the writing of its guiding principles for several months. These fundamentals will serve as a road map when it comes to facing the expansion of this technology.

Who is writing the guide to European AI?

There are fifty-two independent experts who come from the academic field, the civil society and the business world. They make up the group of AI specialists who is working on the first version of the ethical principles "for a trustworthy Artificial Intelligence" in the European Union. Among them, there is a Spanish woman, Cristina San José, Chief Data Strategy at Santander group.

Only a few days ago, the content of the draft on which these experts have been working for the last months came to light. It is a first approximation, not a definitive text. In fact, until January 18, comments and suggestions on this first outline will be accepted. Users who register in the AI Alliance will be able to access a virtual forum in which they can offer their point of view. These individuals will have the opportunity to participate in the discussions on the measures that have been proposed so far. As of the aforementioned date, the group of experts will return to work to finally publish the concluding version of the text in March.

What does the European draft say?

The provisional text bets clearly on a people-centered approach. According to it, the development of this technology should seek to maximize the benefits and minimize the risks. In addition, the use of AI should not be considered a means itself, but rather a technology aimed at increasing human well-being. For AI to be reliable, it will have to respect fundamental rights, applicable regulation and fundamental principles and values (always ensuring an "ethical purpose") and it will also be technically robust. The draft of the High-level Expert Group on Artificial Intelligence in Europe points out that even with good intentions the lack of technological control can cause unintentional damage.

The guide of the European Commission lists five principles and values in order to ensure that AI is developed in a human-centered way:

The principle of beneficence: "do good"
The principle of non-maleficence: "do not harm"
The principle of autonomy: "preserve human agency"
The principle of justice: "be fair"
The principle of the explicability: "operate transparently"

The draft also includes a list of issues that raises concerns and has generated debate among experts: identification without consent, covert artificial intelligence systems, mass citizen scoring without consent in deviation of Fundamental Rights and lethal autonomous weapons systems (LAWS). In turn, the document leaves room for scenarios that today may look like the plot of a science fiction movie, but, in case they are developed in the future, they may have a very high associated risk (for example, the development of some kind of artificial "consciousness").

The goal of this draft is not to set a regulation, but to contain a series of principles that serve as a basis for the future development of laws in this area. Europe has already become a world reference in the protection of private data, now it intends to do the same in the field of artificial intelligence. In comparison with countries such as China and the USA, Europe is still far from being a reference in this industry. The EU has outlined a strategy on AI to try to turn this situation around. Its objective is that investments, both public and private, reach € 20 billion by the end of 2020.

At The CEU IAM Business School, we are aware of how important it is to adopt an ethical approach within the framework of business activity. The development of new technologies is testing the innovative and creative capacity of professionals, but also the values and principles with which they work. Our Global MBA has been designed for the professionals who wish to manage a global scenario in constant change and which is marked by digital transformation, taking into account clear values and a deep sense of ethics.