You would no longer have to wait months for a committee decision regarding a defective smartphone, a canceled flight, or an unfairly charged subscription fee. The system reads your uploaded invoice and complaint petition, scans hundreds of thousands of precedent Supreme Court and Consumer Court decisions in milliseconds, and presents an objective "settlement rate" to both parties (consumer and company) in seconds (e.g., 80% in favor of the consumer, 24,500 TL refund). If both parties click the "I Accept" button via e-Government, the file is legally closed in seconds without ever going to court. It seemed like a dream of immense efficiency; until the Union of Turkish Bar Associations (TBB) filed one of the most existential lawsuits in history at the Council of State (Danıştay).
1. The Anatomy of the System: How Does the Robot Mediator Work?
This model implemented by the Ministry of Justice is a version of Large Language Models (LLMs) trained as a legal "Narrow AI." The system's operation is entirely based on data and statistics:
- Data Input: The consumer writes their complaint in their own words and uploads documents via e-Government. Natural Language Processing (NLP) algorithms translate this text into legal terminology.
- Precedent Scanning: The system scans precedent decisions in the national judicial database (UYAP). It analyzes the outcomes of the past 5,000 cases similar to yours.
- Algorithmic Output (Proposal): If the consumer was found right in 92% of past cases, the system sends a settlement text to the company stating: "You will lose the case; accept a 92% refund to the consumer to avoid court costs."
The Ministry's argument was very strong: "Justice delayed is justice denied. It is against human dignity for a citizen to wait 8 months for a shoe refund. Artificial intelligence is a flawless mediator that does not tire, does not take bribes, does not take sides, and relies purely on mathematics."
2. The Rebellion of Law: TBB's Annulment Lawsuit at the Council of State
However, on the morning of September 1, the Union of Turkish Bar Associations (TBB) applied to the Council of State requesting a stay of execution and the complete annulment of the pilot program. The TBB's 40-page petition was a manifesto defending thousands of years of legal philosophy against technology's fetish for speed.
The Core Rationale of the Lawsuit: Lack of "Conscientious Conviction"
The TBB's strongest argument was the principle of "Conscientious Conviction" (Vicdani Kanaat), the cornerstone of Turkish procedural law. The petition stated: "Justice is not the process of pouring data into Excel spreadsheets and calculating statistical averages. Listening to a human being, weighing their special circumstances, the philosophical dimension of their grievance, their honesty, and 'equity' is only possible with a 'human will' possessing a conscience. Artificial intelligence cannot form a conscientious conviction; it can only execute codes and statistics."
The bar associations' objection was not merely philosophical. They also argued that the principles of the "Right to be Heard" and "Equality of Arms" were violated. While companies, thanks to their massive legal departments, quickly learned "prompt" tricks to manipulate the robot mediator; an ordinary citizen could be found unjustified by the system simply because they couldn't explain their issue to the robot in the right language.
3. Black Holes: The Legal Problems of AI
The main technical/legal "black holes" discussed in the hearings at the Council of State, which shook the legal legitimacy of the system, were:
3.1. Copying Past Mistakes into the Future (Algorithmic Bias)
The robot mediator uses past judicial data to predict the future. But what if past decisions were flawed, biased, or detrimental to a certain consumer group? A machine does not develop the law; it merely takes the average of the past. If Consumer Arbitration Committees in a specific city consistently ruled in favor of companies in the past, the machine learns this as the "correct law" and legitimizes this injustice by copying it eternally.
3.2. AI Hallucination
The biggest chronic problem of artificial intelligence models is "Hallucination"; that is, fabricating highly convincing and real-seeming lies (non-existent laws or fake Supreme Court precedents) about a subject they do not know. If the robot mediator fabricates a non-existent "Consumer Law Article 85/C" to persuade the citizen to settle, will the e-Signature the citizen signs based on this lie be valid?
3.3. Privatization of Justice and the Black Box
The company that developed this system (likely a private tech firm) will keep the algorithm's codes secret citing "Trade Secrets." For a state to delegate its authority to dispense justice to its citizens to a non-transparent "Black Box" whose algorithm cannot be audited is contrary to the Constitutional principle of the "Rule of Law."
The Spectrum of AI Use in the Legal System
4. The Global Context: E-Justice or Digital Divide?
The robot mediator debate is not unique to Turkey. "Internet Courts" in China are already resolving small disputes at a rate of millions per day using AI judges. Estonia and Canada are conducting similar "robot resolver" pilot programs for small claims cases.
However, the European Union, with the EU AI Act issued in 2024, categorized artificial intelligence used in the justice system as "High-Risk" and introduced very strict human oversight requirements. The September 1 pilot application in Turkey is a local reflection of this global conflict: Saving a delayed justice system with a practical technological revolution, or sacrificing the soul of justice to an algorithmic efficiency monster?
5. Conclusion: The Codes of Justice Must Remain with Humans
The decision the Council of State will make in this case will draw the boundary between technology and law in Turkey forever. Certainly, the burden on consumer arbitration committees and courts, which have to read hundreds of files a day, must be lightened. However, the solution is not to remove the human from the process. A machine can be an excellent "legal assistant"; it can read and summarize hundreds of pages in seconds, and detect missing documents. But the authority to make the final offer, evaluate "equity," and reconcile the parties by looking them in the eye (mediation) must belong to a flesh-and-blood human being with a conscience.
Expert Opinion: Burhan Doğuş Ayparlar
The 'Robot Mediator' application launched on September 1, followed by the TBB's annulment lawsuit, represents an existential trauma our legal system is experiencing in the face of technology. Pragmatically, I understand and value the Ministry of Justice's reflex to solve bureaucratic sluggishness and backlog with technology. However, 'Efficiency' and 'Justice' are not the same thing. If our goal was merely to close files quickly, we could do it by flipping a coin. The purpose of the law is not just to end a dispute, but to create the feeling in the conscience of the parties that justice has been served (trust in justice).
Where is the Fundamental Flaw? An AI algorithm is a sophisticated statistical engine fed by past data. Statistics reflect the tendency of the majority. Yet justice is often about resisting statistics; it is about delivering the right of a single victim (the exception) where the majority is wrong. An algorithm can flawlessly read the numbers on an invoice; but it cannot calculate the human weight in the outcry of an elderly consumer who was defrauded and victimized for months. 'Equity' in law is the art of stretching the rigid boundaries of written rules with human conscience; a machine does not stretch, it only calculates.
So, what should be done? It is a requirement of the rule of law for the Council of State to annul this application in its current form (with the black box and unmanned approval model). However, completely banishing AI from the justice system is also anachronistic. The correct model is the 'Decision Support System' (Human in the Loop) model. The AI should read the complaint from the citizen, find the relevant legal articles, pour the past 10 years of precedent Supreme Court decisions in seconds, and prepare a file for the Human Mediator (or Arbitration Committee) saying: 'Dear Mediator, here is the draft summary and legal basis of this file.' Following this massive time savings, the human mediator should engage with the parties (via online meeting if necessary) and present the settlement offer based on their own conscientious conviction. We can let algorithms carry the burden of justice, but we can never hand over the scales of justice to a machine.
