We’re talking about ethics and artificial intelligence. The final version of the document The document was prepared by a High-Level Expert Group on Artificial Intelligence appointed by the European Commission.

Once the draft was issued, the second phase of the publication process of the final version was moved.

At this stage, the European Commission has involved and invited a number of accredited bodies at European level to send their technical contributions in order to integrate and improve the draft.

More than 500 comments, suggestions and additions have been screened and checked, which have enriched the official version. Therefore, the current version is the result of a demanding work of analysis, deepening and integration of these contributions.

Who is the author Piero Poccianti is president of the Italian Association for Artificial Intelligence (AIxIA), a non-profit scientific association founded in 1988 with the aim of promoting the research and dissemination of techniques proper to artificial intelligence. Considered the most authoritative academic body in Italy in the field of artificial intelligence, today it has more than 1,000 members among researchers, university professors and society. It focuses on the dissemination of artificial intelligence intended as training, research and development of new paradigms and technology in favour of companies and the Country System.

The purpose of the guidelines is to promote the implementation of a reliable Artificial Intelligence. To be considered as such, there are three components which must be:

it must comply with the legislation,

must follow ethical principles and values,

It should be well structured both from a technical and social point of view, as even with the best intentions, an IA system could cause side effects.

Each of the above components is necessary but not sufficient to achieve the desired reliability. Ideally, the three components should work harmoniously and integrate into their operation. If tensions between these elements arise at a practical level, it will be society itself that will work and act to resolve this tension.

Ethics and artificial intelligence in three chapters

We will analyse in detail the document which consists of three chapters.

It must be borne in mind, however, that the guide in question is based on three levels of abstraction, from the most abstract, that is, Chapter I, in which fundamental rights, principles and values are discussed, to the most concrete, that is, Chapter III, where a In particular, Chapter I refers to the ways in which the ethical purpose of the IA is ensured, defining the fundamental rights, principles and values it should respect. Starting from these principles, Chapter II provides a guide to the implementation of a reliable IA, taking into account both ethical and technical robustness.

To this end, the requirements for a correct Finally, Chapter III makes these requirements operational, providing a concrete but not exhaustive checklist for the assessment of the reliability of This list is then adapted to specific use cases.

Unlike other documents concerning the ethics of the AI, these guidelines do not aim to list again the values and fundamental principles for the AI, but rather to provide guidance on how to implement and make operational these principles in the AI systems.

To learn more:

Aixia’s contribution to the European Code of Ethics

What the first Italian forum on artificial intelligence has expressed

Chapter 1

Based on the fundamental human rights, the first chapter identifies the ethical principles and the corresponding values that must be respected in developing, providing and using the IA systems.

The key indications described and given in this chapter concern:

respect for human autonomy, prevention of harm, fairness and clarity. Particular attention should be paid to situations involving the most vulnerable groups such as children, people with disabilities and others who have historically been disadvantaged or are at risk of exclusion. Other reports are also to be considered as those characterised by asymmetries of power or information, such as the relationships between employers and workers or between businesses and consumers. The recognition that, while bringing substantial benefits to individuals and society, artificial intelligence systems can still cause some risks and have such negative impacts, difficult to predict, identify or measure (e.g. democracy, the rule of law and distributional justice, The adoption of appropriate measures to mitigate these risks when appropriate and in proportion to the extent of the risk.

Chapter 2

Based on the considerations raised in Chapter 1, Chapter 2 provides a guideline on how IA systems should be implemented by complying with the following 7 requirements:

human supervision, technical soundness and security, privacy and data governance, transparency, diversity, non-discrimination and equity, environmental and social welfare, responsibility.

However, significant contrasts and tensions between the different principles and requirements could arise. This is why it is essential to continuously identify, evaluate, document and communicate these trade-offs and their solutions.

Chapter 3

This article provides a concrete, though not exhaustive list of points to consider to implement the requirements highlighted in the previous chapter.

The key indications emerging from Chapter 3 are:

A list for the assessment of AI systems should be adopted to ensure that they are reliable. This assessment is to be carried out during the development, implementation or use and must be adapted to the specific use being made of the AI technology under consideration The assessment list provided will never be exhaustive. In fact, in order to guarantee a reliable AI, a continuous process is needed that includes: identification and implementation of requirements, evaluation of solutions, guaranteeing to improve the results throughout the life cycle of the AI system and the involvement of all stakeholders.

A final section of the document aims to make some of the issues addressed in the document even more concrete. Examples are given in which the use of AI solutions could lead to positive impacts and benefits, as well as criticalities and negative impacts that AI systems, if implemented, could cause and should be carefully considered.

Among the benefits that could be derived from the application of artificial intelligence are mentioned: the reduction of pollution resulting in a reduction of global warming, the optimisation of the use of energy resources and resources in general (think about water). The possible uses of the AI to improve the life of the world population are also mentioned, with particular reference to the elderly, who could be provided with support tools for healthy ageing.

And more: systems that can improve education and culture, support citizens in the process of transformation in progress, making them active and not passive objects.

While among the critical issues highlighted are: the identification of behaviour of individual subjects with discriminatory purposes, the possibility of exploiting IA systems by masking their digital identity (i.e. making them appear as a human being), the measurement of the faculties of people

Ethics and opportunities

So, the document provides an extensive view of the opportunities that we could seize and the critical issues that we will face following the implementation of artificial intelligence systems. Nevertheless, it must be considered as a starting point and not as an exhaustive instrument, the adoption of which can be decisive in order to direct the IA for the good of humanity and the planet.

The European roadmap: an interpretation

Taking note of this fact, the European Commission indicates a roadmap for adoption consisting of several steps:

Companies and public bodies interested in contributing to the process can register on the European Community website and propose themselves as candidates to carry out pilot activities.

and public bodies interested in contributing to the process can register on the European Community website and propose themselves as candidates to carry out pilot activities. The feedback from these activities will provide the basis for the development of the guidelines and in particular the tools for joining the recommendations in 2020.

In the area of ethics, while recognising the validity and good approach and implementation of this document, we must ask ourselves whether such a framework really has the capacity to be meaningful and to be adopted on a large scale so as to obtain an IA time for the well-being of people who reduce

Although the indications are properly combined, we must ask ourselves how compatible these recommendations are with the globally extended modern society and how much, in reality, some of them do not enter into a collision course with the economic, political and cohabitation systems of the environment representative of that geo- That is, the era in which Homo Sapiens has a strong impact on the balance of the planet.

As many economists and scholars of various disciplines have now denounced, if our ultimate desire is to increase profit and GDP growth, it is not necessarily that the improvement of these indicators will lead to well-being.

If we continue to adopt economic models of two centuries ago, where capital and labour are scarce resources, while the environment is considered immutable we will not be able to achieve those ethical aspects that the document correctly highlights.

Artificial intelligence systems are powerful tools, capable of fulfilling our desires, so we must pay attention to how we express them and above all not to express the wrong ones.

Leave a Reply

Your email address will not be published.

You May Also Like