Foundation Lab

Iterated Amplification

Concrete proposal for scalable oversight when AI exceeds human capability

Concept 87 of 100Scaling & AlignmentPhase 12

#87IDAScaling & Alignment

key equationA' = \arg\min_\pi \mathrm{KL}(\text{Amp}(H,A) \| \pi)

Selected Foundation Object

Keep the equation fixed; move through the evidence.

Concept 87 of 100IDAScaling & Alignment / Phase 12: Advanced alignment & safety research

Current question

Like teaching: break hard problems into pieces students can help with

A' = \arg\min_\pi \mathrm{KL}(\text{Amp}(H,A) \| \pi)

PredictionCommit before the demo.

Ask what should change when the equation is manipulated, then let the visualization test that expectation.

EvidenceCompare local witness and source.

Use the runnable panel, the key equation, and canonical papers as separate forms of evidence for the same object.

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.

What is still poorly explained in textbooks and papers:

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

Key Equation

A' = \arg\min_\pi \mathrm{KL}(\text{Amp}(H,A) \| \pi)

Amplify human $H$ with assistants $A$ , then distill:

A' \approx \arg\min_\pi \mathbb{E}_x\left[\mathrm{KL}\big(\text{Amp}(H,A)(\cdot|x) \| \pi(\cdot|x)\big)\right]

Then iterate: $A \leftarrow A'$ .

Recursion: as $A$ improves, $\text{Amp}(H,A)$ becomes more capable.

Christiano et al.2018arXiv

Choose the next question to carry this object forward.