"Just as a child learns what is right through examples and guidance, so does a machine. Whatever we show it today, it will apply tomorrow. Everything can be learned. And there are two methods:

1. Training with Data (Machine Learning). We provide artificial intelligence with large amounts of data (texts, images, videos, sounds, etc.) and let it find patterns, learn relationships, and “predict” or “respond” based on them.

2. Training with Demonstration and Reinforcement (Reinforcement Learning, Learning from Demonstrations). Instead of simply showing it data, we show it how to act (with demonstrations) or reward/punish it for the right/wrong decisions (like saying “good job” or “no” to a child). This way, the machine gradually learns what behavior we “want” and what we don’t."

What led you to explore the use of artificial intelligence (AI) in justice, given your long experience in the business sector?

As a doctoral candidate at Aristotle University of Thessaloniki, with research focused on algorithmic impact assessment, and with over 12 years of experience in commercial law and business strategy development, I have observed the transformative power of AI in markets. My deep engagement with business needs and my understanding of technology's potential led me to explore how AI can enhance the delivery of justice, while maintaining respect for the Greek judicial system, highlighting the challenges that arise.

What issues do you identify in the use of algorithms for judicial decision-making?

The lack of transparency is a major obstacle. As a professional working with businesses, I understand the importance of data quality. Many algorithms function as "black boxes," making it difficult to understand their reasoning. A notable example is research in the US, where risk assessment algorithms disproportionately labeled African Americans as high-risk due to biased data. This suggests that, rather than producing impartial decisions, algorithms may reproduce social biases, such as linking residential addresses with criminality. My experience in the market has taught me that poor-quality data leads to incorrect outcomes, which, in criminal justice, could threaten the right to a fair trial.

How do you assess the potential application of AI in other areas of justice, such as political or administrative justice?

AI has enormous potential in areas such as political and administrative justice, where processes are often more standardized. As a commercial lawyer, I see how AI tools can automate tasks such as calculating compensation in child support cases, drafting payment orders, or managing labor disputes. However, my research at AUTh highlights that algorithms struggle to apply the principle of proportionality, which requires sensitive balancing of rights. Human judgment remains essential to ensure justice, with AI acting as a supportive tool, always with respect for Greek justice.

Based on your experience in Greece and abroad, how can AI improve the Greek judicial system?

With deep respect for the Greek judicial system, which I trust and honor, I recognize that time delays are a challenge. With my knowledge of the market and collaboration with businesses, I believe AI can accelerate processes through systems for case law research, chatbots for citizen guidance, or automated mediation. Moreover, AI can “filter” cases, promoting out-of-court resolution and reducing the burden on courts, thus enhancing system efficiency.

What measures do you consider essential for the correct and fair use of AI in justice?

The transparency of algorithms and the use of high-quality data are non-negotiable requirements for the fair application of AI, ensuring trust in the judicial system.

How do you envision the future of AI in justice?

As the founder of a firm serving businesses, I envision AI as a reliable assistant that enhances the speed, accuracy, and consistency of justice. In Greece, where the dialogue about AI is still limited, a legal framework is needed that respects European law and constitutional principles. AI must remain human-centered, with the judge at the core, while maintaining respect for the Greek judicial system.

Share something interesting about AI.

In the summer of 2024, in a court in Warsaw, AI software assigned commercial cases to a judge, but the opaque reallocation of cases after the judge’s transfer caused protests about uneven distribution. The case was referred to the European Court of Justice, raising issues of judicial independence and transparency.

In Greece, the Ministries of Justice and Digital Governance are working with E.D.Y.T.E. to introduce AI systems that automate the translation of legal documents into eight language pairs (Greek to/from English, Italian, Chinese, Arabic) and real-time interpretation of testimonies. Additionally, AI tools will record spoken words in court hearings, improving efficiency and reducing delays, while respecting Greek justice.

Something to close with?

Lawyers and law are closer to computing than we think. Law and computing meet in the coding of human behavior and will. AI, as a powerful tool, is transforming justice, but it must not override human moral conscience. The EU, with the AI Directive, sets a pioneering framework for responsible use. In an age with challenges like fake news and data protection, responsibility lies with people with critical thinking and ethical responsibility, who will use AI for the good of society, always maintaining respect for the Greek judicial system.

Source: Polismagazino.gr

Link