A comment on the pursuit to align AI: we do not need value-aligned AI, we need AI that is risk-averse
Date published
2024-01-31
Free to read from
2025-01-31
Authors
Supervisor/s
Journal Title
Journal ISSN
Volume Title
Publisher
Springer
Department
Type
Article
ISSN
0951-5666
Format
Citation
Raper R, (2024) A comment on the pursuit to align AI: we do not need value-aligned AI, we need AI that is risk-averse. AI & SOCIETY, Available online 31 January 2024
Abstract
AI Safety, AI Alignment, and the eventual demise of society due to superintelligent beings have recently become topics of public and even political interest. So much so that there were calls to halt all development on Artificial Intelligence until these issues had been addressed (see BBC News, May 2023), Elon Musk has established another organisation (‘xAI’) aimed at tackling such issues, and global leaders have organised global summits to prioritise research in the area. Needless to say; with the introduction of ChatGPT, there has been a growing interest in and precedence to solve issues to prevent the AI dystopian scenarios we might be familiar with from science fiction.