OPEN KNOWLEDGE LIBRARY

Practical guides for using, teaching, and verifying AI.

Sixteen bilingual resources with primary sources, examples, FAQs, and a related free course.

Four pillar guides

Professionals, managers, and enablement teams

A practical framework for using generative AI in real work without outsourcing judgment, privacy, or accountability.

Schools, universities, teachers, and instructional leaders

A people-first approach to AI literacy, privacy, assessment, and teacher-controlled classroom use.

Teams using AI in research, operations, education, or decisions

A practical control system for checking AI output, protecting data, and assigning human accountability.

Developers, technical leads, and product teams

A reviewable workflow for planning, editing, testing, and shipping repository changes with Codex.

A five-check method for validating claims, calculations, sources, omissions, and suitability before using AI output.

Use AI for account research, outreach, discovery, and follow-up without fabricated personalization or careless data handling.

Measure whether AI training changes a real workflow, improves quality, and saves time without adding unacceptable risk.

Use AI to support career and scholarship exploration while minimizing student data and verifying every opportunity.

Design assessment around authentic performance, visible reasoning, source use, feedback, and transparent AI assistance.

A practical sequence for understanding generative AI, checking claims, protecting privacy, and disclosing use.

A practical data-minimization checklist for personal, confidential, regulated, and security-sensitive information.

Place qualified review where it can catch consequential errors instead of adding a ceremonial approval step.

Create a small evaluation set with clear criteria, representative cases, edge cases, and documented reviewer decisions.

Turn a bug report or behavior contract into focused tests, a minimal fix, and evidence from the real runtime.

Refactor in behavior-preserving checkpoints with characterization tests, small diffs, and rollback points.

Build a release handoff that proves what shipped, where it runs, how it was tested, and how to roll it back.