AI能记住的内容暴涨，人专注力暴跌，形成恶性循环。
The Cognitive Divergence: AI Context Windows, Human Attention Decline, and the Delegation Feedback Loop
- 对比AI上下文窗口与人类有效注意力跨度的指数级变化
- 2026年AI记忆能力是人类的111倍，且差距持续扩大
- 提出'委托反馈环'模型，警示过度依赖AI会进一步削弱人类认知
本文揭示并理论化了两个可测量趋势间的自我强化动态：大语言模型（LLM）上下文窗口的指数扩张与人类持续注意力容量的长期收缩。我们称这一不对称现象为认知分歧。自2017年512个标记增长至2026年的2,000,000个标记（增长约3,906倍；拟合λ=0.59/年；翻倍周期约14个月）。同一时期，人类有效上下文跨度（ECS）——基于验证阅读速率元分析（Brysbaert, 2019）及经验性理解缩放因子推导的标记等价度量——从2004年约16,000标记下降至2026年估算的1,800标记（基于截至2020年的纵向行为数据外推，Mark, 2023；详见第9节不确定性讨论）。在考虑检索退化后（Liu et al., 2024; Chroma, 2025），AI与人类的比率从接近平衡的ChatGPT发布时（2022年11月）上升至556–1,111倍（原始）和56–111倍（质量调整后）。除记录该分歧外，论文提出委托反馈环假说：随着AI能力提升，人类委托给AI的认知门槛降低，延伸至需求极低的任务；由此导致的认知练习减少可能进一步削弱已观测到的衰退能力（Gerlich, 2025；Kim et al., 2026；Kosmyna et al., 2025）。两项趋势均不会自发逆转。论文从统计上刻画分歧，综述八项同行评审神经影像研究的神经生物学机制，提供支持委托阈值的实证证据，并提出以经验证的ECS心理测量工具和纵向研究为核心的研究议程。
原文摘要 · Abstract (English)
This paper documents and theorises a self-reinforcing dynamic between two measurable trends: the exponential expansion of large language model (LLM) context windows and the secular contraction of human sustained-attention capacity. We term the resulting asymmetry the Cognitive Divergence. AI context windows have grown from 512 tokens in 2017 to 2,000,000 tokens by 2026 (factor ~3,906; fitted lambda = 0.59/yr; doubling time ~14 months). Over the same period, human Effective Context Span (ECS) -- a token-equivalent measure derived from validated reading-rate meta-analysis (Brysbaert, 2019) and an empirically motivated Comprehension Scaling Factor -- has declined from approximately 16,000 tokens (2004 baseline) to an estimated 1,800 tokens (2026, extrapolated from longitudinal behavioural data ending 2020 (Mark, 2023); see Section 9 for uncertainty discussion). The AI-to-human ratio grew from near parity at the ChatGPT launch (November 2022) to 556--1,111x raw and 56--111x quality-adjusted, after accounting for retrieval degradation (Liu et al., 2024; Chroma, 2025). Beyond documenting this divergence, the paper introduces the Delegation Feedback Loop hypothesis: as AI capability grows, the cognitive threshold at which humans delegate to AI falls, extending to tasks of negligible demand; the resulting reduction in cognitive practice may further attenuate the capacities already documented as declining (Gerlich, 2025; Kim et al., 2026; Kosmyna et al., 2025). Neither trend reverses spontaneously. The paper characterises the divergence statistically, reviews neurobiological mechanisms across eight peer-reviewed neuroimaging studies, presents empirical evidence bearing on the delegation threshold, and proposes a research agenda centred on a validated ECS psychometric instrument and longitudinal study of AI-mediated cognitive change.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。