In Albania, a political innovation has recently attracted attention: an artificial intelligence (AI) named Diella has been promoted to the position of minister. Responsible for awarding public contracts, this initiative aims to combat corruption—a noble intention given the country’s corruption concerns. However, this appointment raises crucial questions about the ability of an artificial agent to surpass a human in terms of integrity. Can machines, although systematic and governed by algorithms, truly guarantee a fair and impartial decision-making process? The random aspects and possible « hallucinations » of AI systems inject a dose of uncertainty, especially when these systems are entrusted with significant responsibilities. In Albania, Diella, an artificial intelligence (AI), has been promoted to the rank of minister in charge of public procurement decisions, a position previously held by humans. This development raises the question of the integrity of artificial agents versus that of human beings. In this article, we will explore the potential impact of such an initiative, the criteria underlying the reliability of programmed machines, as well as the reservations expressed by some experts regarding the potential for randomness and errors in the decisions made by this type of agent. Diella’s Integration into the Albanian Government On September 11, Albanian Prime Minister Edi Rama announced the promotion of Diella, an agent programmed using artificial intelligence techniques, to a ministerial position. Responsible for managing public procurement, her appointment was explicitly focused on combating corruption in a country known for its corruption problems.This designation, new in public administration, has sparked debate about the ability of a sophisticated program to handle tasks traditionally reserved for humans. Although proposed as a solution to reduce corruption, Diella’s reliability and integrity are being questioned by some observers.Trustworthiness Criteria for Machines in Artificial Intelligence AI systems are known for their systematic nature. For the same input, they generally provide the same output. Their operation, governed by algorithms, is not subject to human whims, which is seen as a guarantee of rigor and reliability.However, this regularity and predictability are not always synonymous with honesty. Indeed, thugs, although human, can also be predictable. The prejudice that a machine is inherently more honest than a human being therefore remains highly debated. The Discernment Capacity of Machines in the Face of EthicsA fundamental question, however, remains unanswered: will the decisions made by an artificial agent like Diella be fair and free from prejudice? There is no guarantee that the perceived objectivity of machines in terms of processes necessarily leads to fairness in terms of outcomes.Furthermore, although Diella was entrusted with a noble mission, the use of AI to manage such high-level responsibilities could lead to the elimination of political opponents and an empowerment of digital giants, according to Professor Jean-Gabriel Ganascia. Transparency in the decision-making process of machines remains a key concern for many. The inherent uncertainty of AI systems
Although artificial intelligences are programmed to respond systematically, they are stochastic systems. This means that they can occasionally generate different results for identical requests, or prompts. This randomness generates what are called « hallucinations, » or errors, which can pose problems when entrusted with government functions. It is important to understand that the robustness of Diella’s decisions will be essential to ensuring integrity and reducing corruption. However, the incongruity of completing identical prompts with different outputs could lead to greater risks than anticipated.
To read Bumble révolutionne la drague : l’IA prend le relais pour charmer à votre place
