The digital revolution in Greece, led by the Gov.gr platform and the strategic vision of the Ministry of Digital Governance, has undeniably streamlined the daily interactions between citizens and the state. However, beneath the surface of this modernizing euphoria, a concerning reality is emerging: the penetration of Artificial Intelligence (AI) into the public sector is often occurring within an institutional vacuum, lacking adequate safeguards, transparency, or public scrutiny.
This issue is no longer theoretical. From algorithms used by the Independent Authority for Public Revenue (AADE) to detect tax evasion to data analysis systems in healthcare and social welfare, AI is making or influencing decisions that shape citizens' lives. The lack of a clear accountability framework creates the risk of 'algorithmic arbitrariness,' where citizens face decisions they can neither understand nor contest.
The Rapid Transition and the Oversight Gap
Greece has demonstrated remarkable speed in adopting digital tools. The introduction of 'mAigov,' the first AI-powered digital assistant for the public sector, marked a significant milestone. Despite the communication success, very little has been disclosed regarding the training of these models, their data sources, and, most importantly, the measures taken to ensure they do not replicate biases or disseminate misinformation.
The problem is exacerbated by the fact that many of these solutions are developed by private companies under contracts often shielded by confidentiality clauses. This creates the so-called 'black box' of governance: public agencies rely on private algorithms whose internal mechanisms are unknown even to the state officials themselves. Who bears responsibility when an AI system rejects a benefit application or imposes a fine based on flawed pattern analysis?
Transparency and Accountability: The Black Box Problem
Transparency is not merely an ethical requirement; it is a legal necessity within the framework of the rule of law. According to international organizations, the use of algorithms in public administration must be accompanied by 'Fundamental Rights Impact Assessments.' In the Greek reality, such studies are rare or remain undisclosed.
- Lack of a public registry for algorithms utilized by the state.
- Inability of citizens to seek explanations regarding the logic of an automated system.
- Risk of digital exclusion for population groups lacking necessary digital literacy.
Furthermore, the use of AI in security and surveillance raises serious questions about personal data protection. The deployment of biometric recognition or predictive policing systems, while often presented as a means to enhance public safety, can easily slide into mass surveillance practices, undermining the rights to privacy and freedom of expression.
Social Impacts and the Risk of Discrimination
One of the greatest risks of unregulated AI is the embedding and amplification of social biases. Algorithms learn from historical data. If this data contains biases against specific social groups, the algorithm will replicate them with mathematical precision. In sectors such as justice or employment, this can lead to systemic discrimination against the most vulnerable members of society.
"AI in the public sector should not be a mechanism for evading responsibility, but a tool for enhancing justice. If we cannot explain the 'why' behind a decision, then we do not have administration; we have technocratic authoritarianism."
The need for 'human-in-the-loop' oversight is imperative. No decision that substantially affects a citizen's rights should be made exclusively by a machine. However, the understaffing of public services often leads employees to blindly trust system recommendations, turning human oversight into a formal, bureaucratic exercise devoid of substance.
The EU AI Act as a Lifeline
The recent passing of the European AI Act represents a significant step in the right direction. This framework categorizes AI uses based on the risk they pose, imposing strict obligations on 'high-risk' systems, which include most applications in the public sector.
For Greece, the challenge lies in the immediate and substantive alignment with this framework. Formal compliance is not enough; it requires the creation of an independent oversight body with the expertise to audit the code and logic of algorithms. Simultaneously, it is essential to train public officials and inform citizens, ensuring that technology serves democracy rather than the other way around. The digital transition is a journey that requires our values as a compass, not just the speed of our processors.
