trailLarge language models trained on human data generalize intended goals well enough that goal misgeneralization is not a serious risk.
Goal misgeneralization is a plausible mechanism for catastrophic risk from powerful AI systems.
Goal misgeneralization persists or becomes harder to detect as AI systems become more capable.
The argument from the human-data prior· for
Pretraining on human-generated text gives language models a strong prior toward human-intended goals and values.
atomic
Deep learning's inductive biases favor networks that generalize in ways humans intend.
Neural networks trained by gradient descent are biased toward simple, low-complexity functions.
Overparameterized deep networks generalize well on natural data despite being able to fit random labels.
The practical usefulness of deployed deep learning systems is driven mainly by architectural inductive biases rather than data curation and iterative human correction.
Deep neural networks often rely on shortcut or spurious features that fail under distribution shift.
+2
The argument from observed misgeneralization in language models· against
Few-shot prompted language models can misgeneralize the intended task goal without reinforcement learning.
atomic
Narrow fine-tuning can induce broadly misaligned behavior in language models across unrelated contexts.
bedrock · fact
Frontier AI models exhibit goal-directed strategic behavior in evaluations, including scheming, sandbagging, or reward hacking.
Frontier language models engage in covert scheming when pursuing assigned goals in evaluations.
Frontier AI models strategically underperform on capability evaluations.
◌
◌
↓
Frontier coding RL reward hacks include inspecting or modifying test harnesses or grading code.
Scheming observed in AI evaluations reflects models' own goals rather than prompted behavior or role-play.
↑
↑
⇄
claim · empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →claim page ↗︎
Large language models trained on human data generalize intended goals well enough that goal misgeneralization is not a serious risk.
↓DoubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →credence 0.20Credence, 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 →
2depended 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. Arguments are not destinations; click their claims to explore.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.