Summary:
- Researchers conducted a study comparing human perceptions of artificial intelligence against the output of AI models, finding that AI assistants consistently express significantly higher optimism regarding the future of AI than the general public.
- The study suggests that this "optimism bias" in AI models may stem from their training data and alignment processes, which prioritize helpfulness and positive framing, potentially misaligning them with the more cautious or concerned human perspective.