Copyright

Carla Zoe Cremer; Jess Whittlestone

Published On

2024-09-03

Page Range

pp. 485–512

Language

  • English

Print Length

28 pages

17. Artificial Canaries

Early Warning Signs for Anticipatory and Democratic Governance of AI

Chapter of: An Anthology of Global Risk(pp. 485–512)
This chapter proposes the use of expert elicitation and collaborative causal graphs for identifying early warning signs of transformative progress in Artificial Intelligence (AI). Termed as “canaries”, these early warning signs can identify potential harmful impact on language models of political systems as well as progress towards high-level machine intelligence. The identification of these “canaries” can provide focal points for the anticipatory governance of AI, enabling the control and monitoring of AI development and the ability to steer AI in safe directions. The authors of this chapter offer the chance for a forward-looking, democratic assessment of rapidly developing, transformative technologies.

Contributors

Zoe Cremer

(author)
Doctoral Student at the Human Information Processing Lab at University of Oxford

Zoe Cremer is a Doctoral Student at the Human Information Processing Lab, University of Oxford.

Jess Whittlestone

(author)
https://orcid.org/0000-0002-1252-718X

Jess Whittlestone is Head of AI Policy at the Centre for Long-Term Resilience. She was previously a Senior Research Fellow and Deputy Director of the AI: Futures and Responsibility Programme at the Leverhulme Centre for the Future of Intelligence and the Centre for the Study of Existential Risk, both at the University of Cambridge. She holds a PhD in Behavioural Science from the University of Warwick and a master’s in Mathematics and Philosophy from University of Oxford.