high-level behavioral patterns like uncertainty management and self-verification are fragile and can be suppressed by irrelevant context
「高级行为模式是脆弱的」——这句话揭示了推理模型的一个深层结构性弱点：自我验证不是一种稳健的、内置的能力，而是一种在特定条件下才会激活的脆弱涌现行为。这与人类认知科学的发现高度吻合：人在高负荷环境下，最先退化的是「元认知」能力（对自己思维的监控）。模型复现了这个人类弱点，却没有人类的生理疲劳触发机制——而是用「上下文长度」代替了「疲劳度」。
Sloman, S. A. (2021). How Do We Believe? Topics in Cognitive Science, 0(2021), 1–14. https://doi.org/10.1111/tops.12580
Lagnado, D. A. (2022). Explaining the evidence: How the mind investigates the world. Cambridge University Press.
As the power is unleashed, computers on the Semantic Web achieve at first the ability to describe, then to infer, and then to reason. The schema is a huge step, and one that will enable a vast amount of interoperability and extra functionality. However, it still only categorizes data. It says nothing about meaning or understanding.
The author presents an interesting progression for the Web to eventually learn to reason. The picture he paints of more accessible content on the internet hinges on the internet learning to reason, which is a human characteristic. It seems we need to apply human characteristics to all of our mechanics for them to progress in their usefulness.