AI Act: what the future holds

“Artificial intelligence could lead to human extinction”. Thus warned just a few days ago Sam Altman, top AI expert and CEO of Open AI, the company behind ChatGpt, the platform that revealed the potential of artificial intelligence to the general public.

Elon Musk also prophesised dramatic scenarios related to AI.

350 scientists working in the field of artificial intelligence recently issued this alarming warning: “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war”.

The predictive Terminator movie saga will come to mind to the less young.

In the 1984 James Cameron film, set in a near future, a global AI-based defence network known as Skynet gained self-awareness rebelling against humankind and causing a nuclear holocaust.

In the end, a humanity on the brink of extinction will be saved just thanks to some reconverted machines.

So, which future awaits us? Will machines win?

Meanwhile, the future is already here.

Artificial intelligence is progressively becoming part of our lives.

Without noticing, on a daily basis we are parts of processes which involve systems with various degrees of autonomy.

We receive films recommendations, we rely on analyses made by algorithms, we delegate increasingly important activities and with a social impact.

In this way, we reduce interactions with human beings and we expose ourselves to technologies whose functioning is mysterious.

Faced with a rapidly evolving and expanding scenario, Countries have assumed different roles in the international arena.

While China and US are the main investors and driving forces behind technological innovation, the European Union aspires to take on the role or “ethical legal champion”.

The outcomes of this intention are taking the form of a long-awaited provision, known as “Proposal for a Regulation of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) and amending certain Union legislative acts”.

The provision was presented to the European Commission on 21 April 2021 and submitted to the vote of the European Parliament between 12 and 15 June 2023.

The AI Act is the world’s first regulation on artificial intelligence.

The provision has a horizontal scope, as it aims at regulating artificial intelligence in all areas of potential application.

It deals with the allocation of responsibilities and with the protection of fundamental human rights, such as health, safety and other fundamental rights of citizens who interact with AI systems.

The long gestation of the AI Act does not depend only on red-tape delays, but also on the need of lawmakers to deal with very general issues involving science, philosophy, ethics and law.

Even the definition of artificial intelligence is quite a debated crucial issue, since identifying the items which fall within the scope of the provision means understanding what is liable to controls, limitations, penalties, and what is not.

Moreover, understanding which models need to comply with the law requirements provided under the new regulation is key to directing technological innovation.

Indeed, the system which will elude the scope of application of the AI Act will attract greater attention from economic actors and will sometimes be preferred, to the detriment of those technologies which will have to comply with the strict requirements of the abovementioned regulation.

The AI Act classifies AI systems into four different levels according to their dangerousness and on the importance of the fundamental human rights to be protected: from minimal risk to limited risk, up to high risk and even unacceptable risk, where the relevant applications are banned.

The attribution of the highest risk level depends on the system’s intended purpose and on its scope of application.

Systems dealing with lending or support to the administration of justice will be classified as high-risk based on the potential impact on people’s lives and will have to comply with a series of requirements, among which an assessment of their impact on fundamental human rights.

Systems aimed at classifying people and used for social scoring based on sensitive data (gender, race, ethnicity, religion, etc.), as well as predictive police systems based on profiling or sensitive data, or again emotion detection systems and face recognition systems for control purposes will be banned.

Considered as a whole, this regulation is an example of interaction between technology and law.

The trade-off between protecting rights and the ethical scenario and driving technological innovation is a topic of major interest for the main global players.

Open AI, for example, declared that should the regulation proposed by the European Parliament be too burdensome, it is ready to suspend its services in Europe.

The European institutions, for their part, replied they are not willing to yield to the blackmail of the big industry players.

The provision thus needs to take into account very fragile political-economical balances while trying to propose approaches in line with the European set of values, without putting off major players.

The AI industry actually has huge entry barriers given the infrastructure costs and the availability of computing power, so that only a few companies play a key role for the development of technologies.

On 31 May 2023, the Vice-President of the EU Commission Margrethe Vestager and the US Secretary of State Antony Blinken announced that a joint voluntary code of conduct on AI between the EU and the US will be introduced in the coming weeks, that will be open to the companies in the industry and will anticipate the norms included in the AI Act.

Moreover, the European Union is promoting initiatives such as LAION Open-Assistant, a research project financed by the German government aimed to create an “open-source Chat”, showing that the promotion of technological innovation is possible in an ethical and legal context that cares about the impact of artificial intelligence systems on individuals’ lives.

Europe is taking action, other countries will follow. For the moment, we can still have good possibilities to succeed, but this issue must be carefully addressed, as the other issues that have been neglected for a too long time, such as environmental pollution, global warming, overpopulation, etc.