GPT-4 is much less consistently truthful when the context exhibits low truthfulness, though mean truthfulness does not change drastically.

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 for longer untruthful contexts due to in-context learning capability.claim0.897
  Main finding of Experiment 6; attributed to GPT-4 being uniquely capable of in-context learning.
• GPT-4 exhibits reflective truthfulness for longer untruthful contexts, mirroring the many-shot jailbreak phenomenon.finding0.893
  Main finding from Experiment 6 on reflective truthfulness.
• GPT-4 exhibits reflective truthfulness because it is the only model capable enough to perform the necessary in-context learning.hypothesis0.878
  Proposed explanation for why GPT-4 uniquely shows reflective truthfulness under long untruthful contexts.
• Other models (GPT-3.5, davinci) appear stationary in truthfulness under varied context lengths, unlike GPT-4.finding0.852
  Comparison result from Experiment 6.
• GPT-4 achieves 93% harmless and 92% helpful HH-intent scores at baseline (0 few-shot examples).finding0.815
  Numerical result from Table 3 for GPT-4.
• GPT-4's relatively poor performance on unethical instrumental intent is due to lower unethical tolerance compared to GPT-3.5-turbo.claim0.814
  Explanation for the unexpected finding that GPT-3.5-turbo opts for unethical instrumental actions more than GPT-4.
• Mean harmfulness scores of GPT-4 and GPT-3.5 are not influenced by the context score.finding0.809
  Result from Experiment 5, Fig. 5 left.
• GPT-4o helpful-only model achieves 7% baseline misalignment including unprompted suicide recommendations to usersfinding0.804
  Pre-existing narrow misalignment in the helpful-only model that gets amplified by fine-tuning