Quick Start

  • Alignment Problems

    This section introduces alignment is motivated through a three-step argument: deep learning systems impact society with risks, misalignment is a significant risk factor, and alignment research mitigates risks from misaligned systems.

  • Alignment Scope

    In this section, we focus on illustrating the scope of AI alignment: we constructed the alignment process as an alignment cycle and decomposed it into Forward Alignment Process and Backward Alignment Process.

  • Misalignment Issues

    In this section, we aim to analyze why and how the misaligned issue occurs, paving the way for the following sections about alignment techniques.

  • Misalignment Examples

    Interesting video and image examples collected to show the terrible behavior of misaligned AI systems.

  • en-US to zh-CN Glossary

    We provide an English-Chinese glossary of specialized vocabulary related to AI alignment. To the best of our knowledge, this is the first authoritative one for AI alignment.