Table of contents: 1. The spirit of the Italian bill on artificial intelligence. – 2. AI: a new paradigm for human-centeredness. – 3. A different interpretation of anthropocentrism: toward a sustainable AI. – 4. Conclusion.
- The spirit of the Italian bill on artificial intelligence
On the 20th of May 2024 the Italian Government introduced a bill specifically aimed at setting principles and fundamental rules on the application of artificial intelligence.
The bill consists of 25 articles divided into five chapters. The first one, from Article 1 to Article 6, describes the “principles and purposes”, such as personal data protection, non-discrimination, proportionality, security and transparency. The second one, from Art. 7 to Art. 16, establishes the “sectoral provisions”, including health and disability, labor, public administration, judiciary and national cybersecurity. The third one, from Art. 17 to Art. 22, provides for the “national strategy, national authorities and promotion actions”, as it concerns Italian governance and actions to promote artificial intelligence. The fourth one, from Article 23 to Article 24, defines the “provisions for the protection of users and copyright”. The last chapter, contained in Article 25, describes the “criminal provisions”, amending the national Criminal Code through a legislative delegation.[1]
In essence, the bill aims to define, in compliance with the AI Act, a domestic regulatory framework that promotes responsible and transparent use of AI, with particular reference to those areas where that use could have a significant social and economic impact such as health care, public administration, justice, and professions. This will be achieved through a “dual” governance: on one hand, the National Cybersecurity Agency (ACN) will exercise inspection, supervisory and sanctioning powers over AI systems, as well as it will ensure cybersecurity protection related to AI development. On the other hand, the Digital Italy Agency (AgID) will be in charge of the execution of the national strategy and it will promote innovation and AI development, by evaluating and monitoring AI systems.[2]
The text tries to address some quibbles arising from the anthropocentric view[3] that places the fundamental rights of individuals at the center of protection, in order to unite governmental bodies and private entities, whose collaboration is strategic. This brief contribution will explore some critical features related to the regulation of artificial intelligence based on anthropocentrism.
It is widely believed by experts that artificial intelligence is a neutral tool,[4] whose nature is neither good nor bad: its expansive tendency of potentialities clearly corresponds to an equally broad expansion of threats. In other words, the stone in the hands of Michelangelo can be Pietà, in the hands of Cain it can kill Abel. Moreover, the concept of artificial intelligence is misleading. It seems erroneous to think that machines can be intelligent.[5] Nowadays, human beings have lost the tangible sense as well as the physical nature of the machine, and they are frightened by the mere thought of understanding where their processing capacity comes from.
Therefore, for there to be, in the future, an adaptation to the irreversible and unstoppable evolution of artificial intelligence by the community, it is necessary to work on skills,[6] on digital awareness and inclusion that enables an understanding of the risks involved. In essence, it is essential to prepare citizens for change. Education requires intervention to control, guide and correct. And it is precisely in the new bill on artificial intelligence that is found this effort to consolidate digital skills and awareness, particularly in the use of AI. The bill promotes broad AI literacy by considering multiple aspects, such as the role of national authorities, promotion actions, criminal punishability of deep-fakes, copyright protection, and so forth.
- AI: a new paradigm for human-centeredness
AI consists of a set of complex and diverse realities, which inevitably affect education, communication, customs, consumption and a great many other aspects of daily life, risking, at times, controlling and dominating the habits of individuals without their noticing, limiting their freedom of choice.[7] So, it does not seem possible to say a priori that an anthropocentric view, where it is not properly managed, despite being human-focused, can make a beneficial contribution to humanity. Such an outcome can be achieved through responsible action, with respect to transparency, prevention and safety, taking into consideration not only the individual per se, but the entire ecosystem. The preservation of the planet, the progress of humanity, and the protection of the individual are interdependent: a purely individualist interpretation of human rights no longer seems correct.[8]
In fact, in this turbulent historical period marked by global conflicts, artificial intelligence highlights the obsolescence of the anthropocentric view: human beings have always imposed themselves in a careless way through artificial invention by egocentrically trying to control and dominate living reality to meet their own needs. This has distorted our perception of reality, denying the inevitable interconnection between the collective and the individual and appealing to an alienated separateness between them. This probably stems from the psychological need of human beings to make the entire universe comprehensible and interpretable through an anthropocentric view, due to our inability to accept that the real reason for our existence is obscure.[9] Artificial intelligence, therefore, has a range of potential that can help us to replace our false beliefs with acceptance of the unknown, an intrinsic element of our existence. It therefore seems essential not to reduce artificial intelligence to a mere means of selfish power, but to use it to revolutionize our distorted view of reality.
- A different interpretation of anthropocentrism: toward a sustainable AI
By “anthropocentric artificial intelligence” is meant the attitude of identifying AI as a tool solely in the service of humans. In many documents, both national and European, state that technology works for people. To consider artificial intelligence as a mere tool serving only humans seems reductive, unless we want to generalize and think that all living things on the planet are tools serving humans.[10] How many times has the anthropocentric view been widely criticized where it has produced negative effects on the environment? Putting man at the center has involved setting aside the planet. In fact, man has yet to understand that the planet is alive, suffering, sensing and evolving all the time, which is why ecosystems have been compromised by our selfish logics of individual self-interest.[11] The planet is based on balances that human beings have upset through extreme and excessive consumption of resources. Therefore, an absolutely anthropocentric view of artificial intelligence could well produce negative effects. To address these issues, it would be desirable to place not the individual but all living beings and the planet in general at the center of protection in order to achieve sustainable AI.[12] This was also noted in the 2019 Communication No. 168 of the European Commission, where it emerges that AI should take into account its impact on the environment and other living beings, ensuring a habitable environment for them, in order to ensure reliability.[13] In essence, the intention to place the individual alone and not the entire “human family” (as was defined in the preamble of the Universal Declaration of Human Rights)[14] at the center of artificial intelligence, respecting their dignity, including the needs of the most vulnerable, seems wrong.
- Conclusion
Referring to the importace of the skills mentioned at the outset of this contribution, the individual should assume a role of digital competence without overriding – as happened with the environment – the free digital development, and thus AI. A holistic approach should be taken that assumes in-depth knowledge of the digital tool and intervenes ex ante, that is, before it is brought to market and not after.[15] Intervention ex post risks blocking development processes and focusing solely on sanctioning the one who caused any irreversible damage. This is late and inefficient intervention,[16] especially in key areas such as health and education. In essence, the absence of absolute scientific certainty should not be an excuse for delaying corrective action. The purpose giving rise to preventive action is to anticipate precautionary (and thus, to some extent, preclusive or limiting) interventions in the face of potential irreparable harm, such as that caused to the environment or health.[17]
The intervention of the individual in the initial design phase is crucial in imposing the goals and objectives of the AI system, thus carrying out a guiding activity, in accordance with the precautionary principle, which presupposes caution and risk management. It is essential to bring to market AI systems that are trained in advance and rigorously tested, without waiting for the resulting risks to be scientifically proven. Rather, a comprehensive risk analysis prior to system implementation seems effective, whereby various usage scenarios and possible impacts can be considered, involving a broad spectrum of stakeholders, including users, experts, regulators, and civil society representatives, to ensure a comprehensive and inclusive understanding of the potential effects of the system. In this way, the importance of preventive interventions is evident.
An anthropocentric view involves multiple attempts that often cause significant damage, which forces the need for actions to remedy the situation by ex post corrective measures, as was the case with the environment. However, in most cases such corrections fail to reverse the effects that have already been produced and become irreversible. It is, therefore, essential to establish principles on which artificial intelligence must be based as it evolves. But it is also essential that these principles, inspired by those of prevention and precaution, be based on in-depth expertise and adequate knowledge of the subject matter, and be employed in a system focused on accountability, that is the empowerment of economic actors in the field, as more suitable for neutralizing risks and solving legal problems related to emerging technologies. Such a proactive approach promotes greater confidence in technological progress and ensures that AI makes a positive contribution to the community and avoids repeating past environmental mistakes by putting people, not resources, at the center.[18]
—
[1] Italian Senate, Bill no. 1146, XIX legislative term, “Disposizioni e delega al Governo in materia di intelligenza artificiale”, 20 May 2024.
[2] Art. 18 of the bill.
[3] Art. 1 of the bill.
[4] M. Fasoli, Contro lo strumentalismo tecnologico. Per una teoria analitica della prescrittività degli artefatti, in Sistemi intelligenti, in Rivista quadrimestrale di scienze cognitive e di intelligenza artificiale, 2, 2020, 223 ff.
[5] C. M. Signorelli, Can Computers Become Conscious and Overcome Humans?, in Frontiers and Robotics and AI, 121, 2018, 1 ff.
[6] V. Vitezic – M. Peric, The role of digital skills in the acceptance of artificial intelligence, in Journal of Business & Industrial Marketing, 12 February 2024.
[7] N. De Marcellis-Warin – F. Marty – E. Thelisson et al., Artificial intelligence and consumer manipulations: from consumer’s counter algorithms to firm’s self-regulation tools, in AI Ethics, 2, 2022, 260.
[8] This is because every right necessarily imposes a duty, which inevitably results in a sense of responsibility; G. M. Greco, On Accessibility as a Human Right, with an Application to Media Accessibility, in A. Matamala – P. Orero (ed.), Researching Audio Description. New Approaches, 2016, 3.
[9] L. Droze, Anthropocentrism as the scapegoat of the environmental crisis: a review, in Ethics in science and environmental politics, 22, 2022, 31.
[10] S. Tommasi, L’intelligenza artificiale antropocentrica: limiti e opportunità, in Jus Civile, 4, 2020, 859.
[11] G. Shkliarevsky, Living a Non-Anthropocentric Future, ssrn.com, 18 November 2021.
[12] A. Wynsberghe, Sustainable AI: AI for sustainability and the sustainability of AI, in AI and Ethics, 1, 2021, 213 ff.
[13] European Commission, Building Trust in Human-Centric Artificial Intelligence, COM(2019), 168 final.
[14] OHCHR, Universal Declaration of Human Rights, 1948.
[15] G. Malgieri – F. Pasquale, From Transparency to Justification: Toward Ex Ante Accountability for AI, Brussels Privacy Hub Working Paper, 33, 2022, 4 ff.
[16] G. Malgieri – F. Pasquale, Licensing high-risk artificial intelligence: Toward ex ante justification for a disruptive technology, in Computer Law & Security Review, 52, 2024, 2-3.
[17] H. Lin – W. Liu, Risks and prevention in the application of AI, in J. MacIntyre – J. Zhao – X. Ma (eds.), The 2020 International Conference on Machine Learning and Big Data Analytics for IoT Security and Privacy, Berlin, 2021, 700 ff.
[18] Ivi, at fn 18.