teaching

Materials for my courses and workshops.

🎯 Responsible Vibe-Coding: Programming with Natural Language for Human Rights

Teaching non-programmers to build working applications through AI-assisted coding. After two days, human rights lawyers in Venice built legal document search tools from scratch—with zero prior coding experience.

"Programming apps doesn't look hard anymore."
— Workshop participant, Venice 2025

Why Human Rights Experts Need NLP & Vibe-Coding

Human rights work is fundamentally about text: treaties, reports, legal documents, testimonies. Even basic NLP techniques can be gamechangers for human rights work:

  • Transform collection of documents into searchable databases
  • Find patterns invisible to human reading (trends across decades of reports)
  • Quantify arguments with data (e.g., "Internet access" mentioned 27 times in 2024 reports - see example)
  • Build tools like General Comments Database
  • Democratize access to legal information

Read about promises and pitfalls of NLP and AI in judiciary in the recent report of the SR on Independence of Judges and Lawyers (A/80/169).

Six Lessons from Teaching Vibe-Coding

  1. 🔓 The AI Exposure Gap is Massive: Only 6% of participants had used paid LLMs. Most judged AI's capabilities based on free tiers—creating misleading impressions of what AI can do.
  2. 🎯 Purpose Must Precede Process: Top complaint: "I didn't understand WHY we were building this." Start with the problem (why this beats Ctrl+F), not the solution (how to build a search engine).
  3. ⚡ The "Seeing It Work" Moment is Critical: For non-technical learners: working demos before explanations. Abstract concepts mean nothing until they see concrete output.
  4. 📊 Bimodal Learning Outcomes: 38% "got it" immediately, 25% remained lost—little middle ground. AI tools don't reduce learning curves; they change their shape entirely.
  5. 🔄 Application Overwhelm > Concept Overwhelm: Participants weren't confused by programming—they were overwhelmed by tool-switching (Cursor → Colab → Terminal). Friction reduction > technical purity.
  6. 💬 "Natural Language" Isn't Natural to Everyone: Prompting in English ≠ accessible. Programming is structured thinking. Some struggle with computational logic, whether in Python or prompts.

Workshop Materials

Teaching AI literacy to human rights professionals isn't about making them programmers—it's about showing what's possible and giving them confidence to try.

⚖️ Human Rights and Technology in the Digital Age – An Interdisciplinary Perspective

Digital technologies – such as Artificial Intelligence, biometric recognition systems, or credit scoring systems – are gradually entering our daily lives, without us even noticing them. This course aims at providing students with solid knowledge on the impact of these technologies on the protection of human rights. To this end, this course highlights the importance of an ethical, human-centric, and accessible tech-infused future.

The course is divided into two modules, namely: 1) introduction to data science; and 2) international human rights law. The first module allows the students to better understand the concepts of programming, statistical modeling, and algorithms and gain some basic practical skills in these areas. In this context, the students will be provided with an opportunity to experience first-hand the dilemmas and challenges of designing an application (e.g. through running a code on their own). Within the second module, the students will learn about the existing case-law, regulations, and soft law initiatives related to the intersection of international human rights law and digital technologies. Completion of this module improves skills in analyzing legal texts and case law.

Hands-on Laboratory Classes

  • Analyzing Legal Text as Data | Google Colab: EN version | PL version
  • Statistical Modeling - Predicting ECtHR Outcomes | Google Colab: EN version | PL version
  • Detecting Bias in Data & Simpson's Paradox | Google Colab
  • AI Transparency and Human Rights | Google Colab: PL version
    Real-life examples: SyRI (Netherlands) and SLPS (Poland)

More information

🤝 Summer Institute in Computational Social Science – SICSS-AMU/Law

The Summer Institutes in Computational Social Science (SICSS) are a global network of two-week summer schools that train early-career researchers to use data, code, and modern AI tools to study society – from social media and politics to law and human rights. They combine lectures, coding labs, and group projects, and all teaching materials are openly available online.

SICSS-AMU/Law is the Poznań node of this network, hosted at the Faculty of Law and Administration, Adam Mickiewicz University. It is one of the few SICSS locations worldwide focused specifically on legal applications of machine learning, natural language processing, and statistics, with materials and projects maintained openly on GitHub.

What Happens at SICSS-AMU/Law?

Over two intensive weeks, participants learn how to turn legal and policy questions into empirical research using code and data. Classes mix short expert talks, hands-on coding in Python or R, and collaborative projects guided by teaching assistants. The school follows the core SICSS format – lectures, exercises, and participant-led research – but adapts it to law and human rights.

  • Tools: Python and R, Jupyter/Google Colab, large language models (LLMs) from OpenAI and Perplexity, collaborative platforms like Slack, Zoom, and Miro.
  • Teaching format: expert lectures, guided coding notebooks, and group research projects developed in the second week of the school.

Topics and Example Projects

The school is designed for people who work with legal texts and data but may never have written a line of code. Typical exercises and projects include:

  • Natural Language Processing (NLP) for law: analyzing large collections of judgments, statutes, or reports as data.
  • Programming with LLMs: using systems like ChatGPT to prototype legal tools via prompting and APIs.
  • Predictive modeling: exploring how machine learning can forecast court decisions and what its ethical limits are.
  • Sentencing and trends: using statistics to reveal patterns in the Criminal Code and case law.
  • Human rights investigations: OSINT techniques and AI tools to help document and measure human rights violations.
  • Legal chatbots & RAG: building AI-powered assistants that answer questions about laws and regulations (e.g. EU AI Act) using retrieval-augmented generation.

Who Is It For?

SICSS-AMU/Law is aimed at advanced students, PhD candidates, post-docs, and early-career researchers, as well as practitioners interested in empirical and data-driven approaches to law. Past editions have brought together participants from universities and institutions across Europe, Asia, and North America, making it a genuinely international and interdisciplinary environment.

Participation is tuition-free thanks to grants and support from Duke University and the Faculty of Law and Administration, lowering the barrier for researchers who want to experiment with computational methods in legal research.

Materials & Further Information

SICSS-AMU/Law is where legal scholars, human rights practitioners, and data scientists meet to ask a simple question: what happens when we treat law not only as a system of rules, but also as a rich source of data?