Skip to content
Longterm Wiki

Eliciting Latent Knowledge

Scalable Oversightactive

Extracting what an AI model 'actually believes' rather than what it says, addressing the distinction between model knowledge and model outputs.

Organizations
4
Key Papers
3
Grants
3
Total Funding
$181K
First Proposed: 2022 (Christiano et al., ARC)
Cluster: Scalable Oversight
Parent Area: Scalable Oversight

Tags

elkknowledge-elicitationalignment

Organizations1

OrganizationRole
Alignment Research Centerpioneer

Grants3

NameRecipientAmountFunderDate
Grant to "support a competition for work on Eliciting Latent Knowledge, an open problem in AI alignment, for talented high school and college students who are participating in Prometheus Science Bowl."Prometheus Science Bowl$100KFTX Future Fund2022-05
A research & networking retreat for winners of the Eliciting Latent Knowledge contest-$72KLong-Term Future Fund (LTFF)2022-10
3 months relocation from Chad to London to work on Eliciting Latent Knowledge with Jake Mendel from Apollo ResearchSienka Dounia$8.5KLong-Term Future Fund (LTFF)2024-01

Funding by Funder

FunderGrantsTotal Amount
FTX Future Fund1$100K
Long-Term Future Fund (LTFF)2$81K

Key Papers & Resources1

SEMINAL
Eliciting Latent Knowledge
Christiano et al. (ARC)2021