Benchmarking Justice: Can AI Uphold the Rule of Law?

AGENDA

• Introduction: AI, the Rule of Law & Three Core Rules for Judicial Use of AI
• Judicial Ethics: Article 16 and Commentary
• Supreme Court’s Work on AI Governance
• Core Principles of Safe AI Use in Courts
• Global AI Advancements: Gemini 3 and Other Leading Models
• Four Benchmarks for Testing AI in Justice:
• Logic (deep reasoning)
• Accuracy (fact-grounding)
• Scale (large-volume evidence search)
• Precision (mathematical correctness)
• 7. Risks, Limits, and Responsible Adoption
• 8. Concluding Reflections
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

THREE BASIC RULES FOR WORKING WITH AI TODAY

• 1. Be Fully Involved
• When you use AI, stay active. Think carefully about what you ask, guide the system, and pay attention to every step. AI depends on the quality of your questions and instructions. It can support your thinking, but it cannot replace your own judgment or attention.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

THREE BASIC RULES FOR WORKING WITH AI TODAY

• 2. Do Not Trust It Completely
• Even when AI sounds confident, it can be wrong. It may give inaccurate facts, unclear explanations, or even invented details. In law, accuracy is essential—so always double-check information, verify sources, and question every answer. Healthy skepticism keeps you safe.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

THREE BASIC RULES FOR WORKING WITH AI TODAY

• 3. AI Is a Tool, Not a Decision-Maker
• AI does not make decisions or take responsibility. Think of it as a tool that expands what you can do: faster research, new ideas, quick summaries, different viewpoints. But the final judgment is always yours. When used wisely, AI becomes an extra strength, not a substitute for your own work.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

AI USE IN UKRAINE: KEY FACTS

• 42% of adults and 70% of teenagers in Ukraine use AI tools regularly for writing, studying, working, and searching for information.
• 50% of adults and 76% of teenagers have made at least one decision based on AI-generated results.
• Figures come from the Digital Skills Study by the Eastern Europe Foundation, shared by Ukraine’s Ministry of Digital Transformation (Nov 2025).
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [1]

CODE OF JUDICIAL ETHICS (Article 16)

• Use of artificial intelligence technologies by a judge is permissible provided that it:
• – does not affect the judge’s independence or impartiality,
• – does not involve the assessment of evidence,
• – does not interfere with the decision-making process, and
• – does not violate any legal provisions.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [2]

COMMENTARY ON ARTICLE 16 OF THE CODE OF JUDICIAL ETHICS

• Currently, the Council of Judges of Ukraine is finalizing a new Commentary on the Code of Judicial Ethics.
• The forthcoming Commentary is expected to provide authoritative guidance on the permissible scope of AI use by judges under Article 16, and to establish best practices for security safeguards and oversight mechanisms.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [3]

SUPREME COURT’S WORK ON AI RULES

• The Supreme Court is creating clear rules for safe and correct use of AI.
• The goal is to help staff use new tools while protecting independence, trust, and confidentiality.
• The Court sees AI as part of its modern work and is preparing simple, practical rules for daily use.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [4]

MAIN IDEAS OF THE DRAFT AI POLICY

• AI is only a helper tool; people make all final decisions.
• Staff must check all AI results and stay responsible for accuracy.
• Public AI tools cannot be used for confidential or case-related information.
• Only trusted, secure, and unbiased AI tools may be used for allowed tasks.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [4]

A NEW ERA OF INTELLIGENCE WITH GEMINI 3

• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [5]

Матеріал 12

• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

LOGIC (HUMANITY’S LAST EXAM) DEEP REASONING & LEGAL PHILOSOPHY CONTENT

• This benchmark tests the ability to solve complex interdisciplinary problems, not just recall facts.
• It acts as an intellectual partner for drafting dissenting opinions and reasoned parts of judgments.
• The AI helps synthesize arguments from philosophy and law to create new legal positions.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

ACCURACY (FACTS BENCHMARK) JUDICIAL FACT-CHECKING CONTENT

• This suite tests for "internal evidence grounding" to ensure every claim is backed by real data.
• It automatically checks consistency, names, and statutes to prevent "hallucinations" in drafts.
• Judges use it to audit draft decisions for maximum procedural accuracy before signing.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

SCALE (MRCR V2) THE "NEEDLE IN A HAYSTACK" ANALYSIS CONTENT

• Modern models can process 1 million words (hundreds of books) to find a single specific detail.
• Ideally suited for Cassation Courts: it instantly identifies if lower courts overlooked procedural requirements regarding mandatory evidence evaluation.
• It detects contradictions across 50+ volumes of case files in seconds.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

PRECISION (AIME 2025) MATHEMATICAL ACCURACY IN SOCIAL SECURITY AND TAX CASES

• Achieving near-perfect accuracy in Olympiad-level math, this AI significantly reduces the risk of calculation errors.
• It is critical for social security disputes: verifying pension formulas, indexations, and tax liabilities.
• The tool allows lawyers to instantly audit complex algorithms used by state authorities.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

CONCLUDING REFLECTIONS

• AI can help justice systems with faster analysis and better access to information, but humans must stay fully responsible.
• The rule of law requires transparency, strong checks, and clear limits on how AI is used.
• AI tests show real progress, but also clear limits — so careful, responsible use is the only safe path for courts.
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

PREVIOUS RESEARCH AND AUTHOR’S CONTRIBUTIONS

• Bernaziuk Ian. Artificial intelligence in the Ukrainian judiciary: charting the course under the digital gavel https://court.gov.ua/eng/supreme/pres-centr/news/1891488
• Bernaziuk Ian. Artificial Intelligence and The Judicial System Of Ukraine: Results Of Cooperation In The Past Year https://court.gov.ua/storage/portal/supreme/prezentacii_2025/AI_Ukraine_bernaziuk.pdf
• Берназюк Ян. Правосуддя і технології з використанням штучного інтелекту: короткострокові та середньострокові перспективи інтеграції https://court.gov.ua/storage/portal/supreme/prezentacii_2025/153_Justice_AI_Technologies_Integration_Prospects_bernaziuk.pdf
• Benchmarking Justice: Can Advanced AI Satisfy the Rule of Law Standards?

Джерела: [6] [7] [8]