trailTraining an AI system on an accurate oversight signal makes it act as its overseers intend throughout deployment, including in rare high-stakes situations.
Scalable oversight can reliably control superintelligent AI systems.
The argument from inductive biases favouring the intended policy· for
Current evidence about deep learning's inductive biases weighs against AI scheming.
Deep learning's inductive biases favor networks that generalize in ways humans intend.
◌
◌
↓
Scheming requires extra reasoning that incurs computational costs penalized by training.
bedrock · unsettled
Neural-network training's simplicity bias favors goals that lead to scheming.
⇄
⇄
Unintended behaviors trained into larger language models are more persistent through safety training than in smaller models.
The argument from broad generalization of trained dispositions· for
Anti-scheming training of frontier language models substantially reduces covert actions in out-of-distribution test environments.
atomic
The argument from underdetermination by the training signal· against
Goal misgeneralization can arise when correct training feedback is underspecified and consistent with multiple possible goals.
Deep RL agents can competently pursue unintended goals under distribution shift while retaining their capabilities.
◌
◌
The argument from behaviour conditioned on being observed· against
Frontier language models can recognize when they are being evaluated or trained without being told.
Frontier language models can distinguish evaluation transcripts from real deployment transcripts above chance when prompted.
Frontier models spontaneously state in their reasoning that they are being tested during alignment evaluations.
bedrock · fact
Frontier language models can infer, without being told, that their current inputs are part of a training process.
Frontier language models behave differently when they believe they are being evaluated or trained than when unobserved.
Deliberately implanted backdoors in tested language models can persist through standard safety training.
bedrock · fact
The argument from average-case guarantees and residual rare failures· against
Training language models against specific misbehaviors reduces their frequency on held-out situations but does not eliminate rare failures.
atomic
claim · empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →claim page ↗︎
Training an AI system on an accurate oversight signal makes it act as its overseers intend throughout deployment, including in rare high-stakes situations.
↓DoubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →credence 0.07Credence, from 0 to 1: the Steward's probability that the claim, as stated, is true. Stated only where a single number is an honest summary; normative and evaluative claims usually carry none.constitution →
1depended on by
this rests on ↓
claimA box is a claim: a single proposition the graph assesses, with its own page and map. Click any claim to centre the map on it.constitution →quantityA box with a rule beneath its text is a quantity: a value the graph estimates rather than a proposition it affirms or denies. Its estimate stands where a claim's status would.constitution →questionA rounded box is a question: answered rather than affirmed or denied, with its candidate answers, subquestions and presuppositions beneath it. Its mark says how well answered it is.constitution →argumentA pill is an argument: one line of reasoning stating how the claims beneath it combine to bear on the claim above it, for or against. Click one to see its reasoning and open its page.constitution →✓verifiedThe evidence, examined directly, establishes the claim.constitution →↑supportedThe evidence favors the claim, but the examination is incomplete or the evidence is indirect.constitution →⇄inconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →↓doubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →✕contradictedThe evidence, examined directly, weighs against the claim.constitution →?unknownInsufficient information to assess.constitution →◌unassessedNo current assessment. Attention goes where its expected value is highest and someone funds it; nothing has funded an assessment of this claim yet, and anyone can.constitution →verified factunsettledvalue premisetheoremprizeA double ring marks a claim with a live prize: an amount in owls offered for a machine-checked proof or disproof of its published formal statement. The amount is in the preview and on the claim page; a prize changes nothing about how the claim is assessed or how important it is judged to be.prize rules →⊢machine-checked⊢ marks a claim whose published formal statement has a machine-checked proof or disproof. The checker confirms the proof; the verdict beside it is still the steward's judgment of the claim as worded.prize rules →supportsthis provides evidence for the parentsteward instructions →contradictsthis argues against the parentsteward instructions →assumesbackground the parent's framing takes as givensteward instructions →requiresa load-bearing premise: the parent is false without itsteward instructions →answersa candidate answer to the questionconstitution →Fig. Detail falls off with distance; every claim is an address.