Continuous FunctionStart learning

Foundation Lab

Capability Elicitation & ELK

Safety evals must find worst-case, not average-case capability

Concept 93 of 100Scaling & AlignmentPhase 12
Inspect equationAll Foundations
#93ElicitationScaling & Alignment
key equation
gψ(h(x))≈zg_\psi(h(x)) \approx z
Reading map and next steps

Selected Foundation Object

Keep the equation fixed; move through the evidence.

Concept 93 of 100ElicitationScaling & Alignment / Phase 12: Advanced alignment & safety research
Current question

Scaffolding/prompting can dramatically change apparent capability

gψ(h(x))≈zg_\psi(h(x)) \approx z
PredictionCommit before tracing the equation.

Ask what should change under a concrete input, then trace that expectation through the equation.

EvidenceCompare the equation and source.

Use the key equation and canonical papers as the available witnesses, without implying that a runnable panel exists.

InvariantName what survives notation changes.

The useful learning product is the reusable mechanism you can carry into another model, paper, or engineering tradeoff.

Next moveContinue through the atlas.

Use prerequisites, dependents, and semantic links to repair the next gap without leaving the object behind.

Inspect equationBrowse atlasNext foundation

Why It Matters for Modern Models

  • Safety evals must find worst-case, not average-case capability
  • ELK: can we trust what model says when it could be deceptive?
  • Core theoretical obstacle to alignment

What Tutorials Skip

What is still poorly explained in textbooks and papers:

  • Scaffolding/prompting can dramatically change apparent capability
  • Model may "know" truth internally but output something else
  • Probes on activations might extract honest beliefs

Visualization Status

Core Math (Optional Deep Dive)

If you want intuition first, start with the key equation and cited sources. Come back here for the full walkthrough.

Key Equation
gψ(h(x))≈zg_\psi(h(x)) \approx z

Elicitation gap (capability as max over prompts):

Cap(M)=max⁡p∈PE[score(M,p)]\text{Cap}(M) = \max_{p \in \mathcal{P}} \mathbb{E}[\text{score}(M, p)]

ELK: extract truth from internals even when output unreliable:

gψ(h(x))≈zg_\psi(h(x)) \approx z

where h(x)h(x) = activations, zz = latent truth, even if output y≠zy \ne z.

Canonical Papers

ARC's First Technical Report: Eliciting Latent Knowledge

Christiano et al.2021Alignment Forum
Read paper →

Connections

Prerequisites

⚲Probing🔬Logit Lens

Next Moves

Choose the next question to carry this object forward.