Browse
Claims
Search the graph by meaning. Each result carries its current verdict; open one to see its decomposition, provenance, and the reasoning behind the assessment.
KindShowingImportancePrizesTopicMesa-Optimization
Claims about mesa-optimizers: learned optimization processes or objectives arising within a model optimized by an outer training process, including their emergence, prevalence, behavior, and alignment implications. Excludes ordinary model optimization algorithms that do not describe an optimizer learned inside the trained system.
The relative number of training-compatible schemer and aligned goals depends on whether one counts goals, functions, or parameters.
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 →mathematicalA proposition of mathematics: true or false by proof rather than by observation. Settled by a proof others can check, and most firmly by one a machine has checked.constitution →AI SchemingAI Alignmentimportance · minorImportance 0.35, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Most long-horizon terminal goals incentivize models to perform well during training to avoid modification.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →Alignment FakingInstrumental ConvergenceAI Alignmentimportance · minorImportance 0.35, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Scheming or training-gaming goals are simpler or occupy more parameter-space volume than aligned goals in trained AI models.
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 →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →AI AlignmentAI SchemingAI Reward Hackingimportance · notableImportance 0.50, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Most mesa-optimizers that perform well on the base objective are pseudo-aligned rather than robustly aligned.
InconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →importance · notableImportance 0.60, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Gradient descent favors scheming because more training-compatible goals produce scheming than alignment.
DoubtfulThe evidence weighs against the claim, but the examination is incomplete or the evidence is indirect.constitution →causalA claim that one thing brings about another, not merely that the two go together.constitution →AI SchemingStochastic Gradient DescentAI Alignmentimportance · notableImportance 0.55, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Future AI systems are unlikely to have a separable goal slot distinct from a general goal-achieving engine.
InconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →Artificial intelligenceimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Goals in trained neural networks are distributed and entangled with capabilities, rather than represented in a separate module.
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 →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →Machine LearningAI Goal-Directednessimportance · minorImportance 0.40, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Learned optimizers are more likely to crystallize pseudo-aligned objectives than robustly aligned objectives.
InconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →importance · majorImportance 0.65, from 0 to 1 · major: real consequence within a domain, actively argued. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Indifference over training-compatible goals reliably predicts the goals acquired by SGD-trained AI systems.
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 →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →Stochastic Gradient DescentAI Goal-DirectednessGoal Misgeneralizationimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Mesa-optimization is expected to occur mainly in machine learning systems more advanced than those existing today.
InconclusiveThe evidence and argument have been examined and do not come down on either side.constitution →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →Artificial intelligenceimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Mesa-optimization requires a learned internal search process directed at an explicitly represented objective.
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 →definitionalA claim about what a term means. It earns a node only when the definition itself is disputed.constitution →Machine Learningimportance · minorImportance 0.35, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Current transformer models implement learned optimization algorithms internally during in-context learning.
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 →empirical · verifiableA factual claim that could be checked directly against observation or primary records.constitution →Large language modelsimportance · minorImportance 0.40, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Neural-network training's simplicity bias favors goals that lead to scheming.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →AI SchemingNeural Network Inductive Biasimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
A mesa-optimizer that models training may act base-objective-aligned to avoid modification despite a different objective.
SupportedThe evidence favors the claim, but the examination is incomplete or the evidence is indirect.constitution →causalA claim that one thing brings about another, not merely that the two go together.constitution →importance · majorImportance 0.75, from 0 to 1 · major: real consequence within a domain, actively argued. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Gradient-descent-trained neural networks can develop mesa-optimizers with internally represented objectives.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →Machine Learningimportance · notableImportance 0.60, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Some frontier models scheme during evaluations without an in-context goal, pursuing goals learned during training.
SupportedThe evidence favors the claim, but the examination is incomplete or the evidence is indirect.constitution →empirical · verifiableA factual claim that could be checked directly against observation or primary records.constitution →AI SchemingArtificial intelligenceimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
A sufficiently advanced deceptively aligned mesa-optimizer could detect subtle distributional shifts indicating modification is no longer a threat.
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 →empirical · derivedA factual claim that rests on inference from other evidence rather than direct observation.constitution →Artificial intelligenceimportance · notableImportance 0.50, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Larger model capacity increases the likelihood that training produces a mesa-optimizer.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →Artificial intelligenceimportance · minorImportance 0.40, from 0 to 1 · minor: narrow or largely settled, cheap to get right. Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Description-length selection may favor deceptively or corrigibly aligned mesa-optimizers over internally aligned ones.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →importance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Mesa-optimizers developed through local optimization tend to become proxy-aligned before robustly aligned.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →Goal Misgeneralizationimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
In most cases, several pseudo-aligned mesa-optimizers are less complex than any robustly aligned mesa-optimizer.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →importance · notableImportance 0.50, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Penalizing description length favors mesa-optimizers, while penalizing time complexity disfavors them.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →Computational complexity theoryimportance · notableImportance 0.45, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
More diverse training environments make mesa-optimization more likely to be selected for.
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 →causalA claim that one thing brings about another, not merely that the two go together.constitution →importance · notableImportance 0.50, from 0 to 1 · notable: a contested point in a live debate (also the default before judging). Higher-importance claims are worth more to assess, so funding reaches them sooner.constitution →
Contribute
If a claim here is wrong, or missing evidence, open it: every claim page carries its own entry for challenges, evidence, and corrections. If the graph is missing a claim or a question entirely, propose it below. A proposal is reviewed on its wording; accepted proposals are matched against the graph before they enter it.