Global Campus of Human Rights · New Technologies cluster

AI and Human Rights, from the inside

Four sessions: how language models work and fail, open tools built with AI on UN human rights sources, and two sessions building your own in Garage Research.

Scan, or open
lszoszk.github.io/teaching-materials/global-campus
Łukasz Szoszkiewicz · Adam Mickiewicz University, PoznańMonastery of San Nicolò, Lido di Venezia
/

Sessions

1 Day 1

Inside the Machine

How language models are made, where they fail, and what relying on them does to us.

Slides
2 Day 1

Tools Built with AI

Find it, cite it, check it in the UN Human Rights Database, then your first build in Garage Research.

Slides
3–4 Day 2

Garage Research

Describe a tool in plain words; a model writes it; it runs in your browser on UN data. The drafting history and the Council’s votes, then UN recommendations and a tool of your choice.

SlidesOpen Garage Research

Feedback, anonymous, five minutes each: Day 1 · Day 2

The tools

UN Human Rights DatabaseGeneral Comments, treaty body jurisprudence, Special Procedures, paragraph by paragraph
UN Travaux PréparatoiresThe drafting of the UDHR, the ICCPR and the ICESCR, 1946–1966
HRC VotingHow States vote at the Commission, the Council and the Third Committee
UHRI+Recommendations and observations addressed to States

From our guests

Shared after the online session of 11 November with HURIDOCS and Democracy Reporting International. Both are glad to stay in touch with students: write to me for an introduction.

HURIDOCS

Hyebin Bina Jeon, Programme Manager

Databases built with Uwazi

Uwazi and Tella

Machine learning

Democracy Reporting International

Jakub Jaraczewski, Senior Research Coordinator, Rule of Law

Policy briefs

Toolkit

To read after class

The tools are unofficial and not endorsed by OHCHR: in case of doubt, the official record prevails. Ask in the UN Human Rights Database, the Garage Research models and any AI assistant send what you type to the company that runs them: no confidential text there.