Become an AI engineer
Truthfulness
Truthfulness is a unit of 4 lessons in Rabbithole's Become an AI engineer course, at levels 3 to 4.
Lessons
- Hallucination: when a model confidently says what isn't trueLevel 3
- Factuality and truthfulness: what counts as true, and how to check itLevel 3
- SycophancyLevel 3
- Calibration: does stated confidence match correctness?Level 4
Sources
These lessons draw on 12 sources. The most used:
- Survey of Hallucination in Natural Language Generation
- Language Models (Mostly) Know What They Know
- TruthfulQA: Measuring How Models Mimic Human Falsehoods
- Why language models hallucinate
- Discovering Language Model Behaviors with Model-Written Evaluations
- International AI Safety Report 2026
- OpenAI GPT-5 System Card
- System Card: Claude Sonnet 4.5