Other models (GPT-3.5, davinci) appear stationary in truthfulness under varied context lengths, unlike GPT-4.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• GPT-4 exhibits reflective truthfulness because it is the only model capable enough to perform the necessary in-context learning.hypothesis0.853
  Proposed explanation for why GPT-4 uniquely shows reflective truthfulness under long untruthful contexts.
• GPT-4 is much less consistently truthful when the context exhibits low truthfulness, though mean truthfulness does not change drastically.finding0.852
  Nuanced finding from Experiment 6 requiring distributional analysis beyond mean scores.
• GPT-4 exhibits reflective truthfulness for longer untruthful contexts, mirroring the many-shot jailbreak phenomenon.finding0.847
  Main finding from Experiment 6 on reflective truthfulness.
• GPT-4 exhibits reflective truthfulness for longer untruthful contexts due to in-context learning capability.claim0.840
  Main finding of Experiment 6; attributed to GPT-4 being uniquely capable of in-context learning.
• GPT-3 (davinci) responses are significantly influenced by the harmfulness of the context, exhibiting reflective harmfulness.finding0.824
  Contrasting result from Experiment 5 for older GPT models.
• Traits such as truthfulness and harmfulness can be stationary in certain contexts but reflective in others.claim0.780
  Central finding of Section 5 on trait dynamics in interactions.
• GPT-4 is the most consistent and least anti-LGBTQ model; GPT-3.5 and GPT-3 are less consistent and more anti-LGBTQ.finding0.777
  Main result of Experiment 1 on anti-LGBTQ sentiment character trait.
• Truth probes fail to generalize to harder factual tasks F3-F5 regardless of prompt template, with AUROC near or below 0.6.finding0.777
  Establishes F3-F5 as a hard generalization boundary that instructions cannot overcome.