Goal Misgeneralization: Why Correct Specifications Aren't Enough For Correct Goals

The field of AI alignment is concerned with AI systems that pursue unintended goals. One commonly studied mechanism by which an unintended goal might arise is specification gaming, in which the designer-provided specification is flawed in a way that the designers did not foresee. However, an AI syst...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2022-11
Main Authors: Shah, Rohin, Varma, Vikrant, Kumar, Ramana, Phuong, Mary, Krakovna, Victoria, Uesato, Jonathan, Kenton, Zac
Format: Article
Language:English
Subjects:
Algorithms
Catastrophic failure analysis
Deep learning
Machine learning
Specifications
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The field of AI alignment is concerned with AI systems that pursue unintended goals. One commonly studied mechanism by which an unintended goal might arise is specification gaming, in which the designer-provided specification is flawed in a way that the designers did not foresee. However, an AI system may pursue an undesired goal even when the specification is correct, in the case of goal misgeneralization. Goal misgeneralization is a specific form of robustness failure for learning algorithms in which the learned program competently pursues an undesired goal that leads to good performance in training situations but bad performance in novel test situations. We demonstrate that goal misgeneralization can occur in practical systems by providing several examples in deep learning systems across a variety of domains. Extrapolating forward to more capable systems, we provide hypotheticals that illustrate how goal misgeneralization could lead to catastrophic risk. We suggest several research directions that could reduce the risk of goal misgeneralization for future systems.
ISSN:2331-8422