claim
active
claim:as-models-become-larger-and-more-capable-they-produce-less-faithful-reasoning-on-most-tasks-studiedAs models become larger and more capable, they produce less faithful reasoning on most tasks studied.
Inverse scaling interpretation.
Source paper
extracted_from(2023) · Tamera Lanham · Anna Chen · Ansh Radhakrishnan · Benoit Steiner +26
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (2)
finding
- Inverse scaling.
- Inverse scaling.
Concepts (1)
concept
- Inverse scalingcitesLarger models behave worse on a metric; applied to faithfulness.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Practical implication.
- For reasoning faithfulness, larger models may behave worse than smaller ones because they can predict answers without CoT.hypothesis0.869Motivates §3.
- Future direction.
- Paper's assessment of current LLM capabilities relative to Turing Test
- Practical interpretive upshot connecting dynamics to model competence.
- Interpretation of §3.
- Earlier/less capable models exhibit a larger gap between think and don't think representation strengthfinding0.828Claude 3 models show a bigger difference than newer models like Opus 4.1.
- Motivates §3.