This article discusses the fluctuations in predictions regarding artificial general intelligence (AGI) in 2025, particularly after the release of OpenAI's reasoning models. It explores the initial excitement over these models, followed by a shift back to longer timelines due to limitations in their generalization capabilities and the challenges of scaling reasoning tasks.

Reasoning models, which utilize extended chain-of-thought (CoT) reasoning, demonstrate enhanced performance in both problem-solving and accurately expressing confidence compared to non-reasoning models. This study benchmarks six reasoning models across various datasets, revealing that their slow thinking behaviors facilitate better confidence calibration. The findings indicate that even non-reasoning models can improve calibration when guided towards slow thinking techniques.