In a pivotal moment for the digital transformation of Greek public administration, the Independent Authority for Public Revenue (AADE) is moving forward with the integration of Artificial Intelligence (AI) tools to manage the massive volume of objections filed by citizens regarding vehicle-related fines. This move, while presented as a leap toward efficiency, raises critical questions about transparency, legal validity, and the role of human agency in administrative decision-making.
The Auditor's Digital Co-pilot
The new system is not intended to fully replace human auditors but to act as a sophisticated "advisor." According to reports, the AI will be tasked with "reading" the appeals, cross-referencing data with AADE's databases, the Ministry of Transport, and insurance companies, and subsequently proposing a decision—rejection or acceptance—to the relevant official. The final signature will remain human; however, the speed at which AI can process thousands of cases is expected to drastically reduce waiting times.
The need for such an intervention is urgent. Every year, thousands of vehicle owners face fines for uninsured vehicles, unpaid road taxes, or missing technical inspections (KTEO), many of which are due to database errors or bureaucratic bottlenecks. The sheer volume of objections often paralyzes services, leading to delays of months or even years.
The Challenge of the 'Black Box' and Good Administration
Despite the obvious benefits in speed, the use of algorithms in administrative procedures is not without risks. The primary concern revolves around transparency. How can a citizen be certain that the algorithm considered all the parameters of their specific case? Artificial Intelligence, especially machine learning models, often functions as a "black box," where the decision-making process is not easily explainable.
- Ensuring the "explainability" of decisions.
- Avoiding automated errors that could lead to mass injustices.
- Compliance with the European AI Act.
AADE emphasizes that the AI's role will be "advisory." This serves as a legal safeguard, as the law requires administrative acts to bear the signature of a natural person who carries the relevant responsibility. However, experience from other jurisdictions suggests that employees tend to trust system recommendations blindly, effectively turning an "advice" into a de facto decision.
Towards an Automated Tax Administration
This application is just the tip of the iceberg. AADE is planning broader use of AI to detect tax evasion through the analysis of market behaviors and social networks. In the case of vehicles, the interconnection of systems (MyCar, insurance, KTEO) creates a digital net that will be difficult to escape. The challenge for the state is to use this power not just for collecting fines, but also for protecting citizens from erroneous charges.
"Technology must be the tool for justice, not its replacement," say IT market circles, pointing out that the success of the project will depend on the quality of the data feeding the algorithm.
In conclusion, the entry of AI into vehicle fine objections is a bold experiment. If successful, it will serve as a model for the entire public administration. If it fails due to a lack of transparency or technical errors, it will further erode citizen trust in institutions. The fine line between digital modernization and algorithmic authoritarianism is now the new field of political and social debate.
