关于 AI 与家庭教育，流行的说法有两种：一种说 AI 把家长从辅导中解放出来，另一种说 AI 制造了新的数字鸿沟。两种说法都停在「任务」这一层——谁做了讲题、纠错、检索、规划。这四篇论文换一个层次发问：当一项知识任务被 AI 接手，原本借由这项任务发生的亲子接触，是否仍然存在？家庭的压力、身份与责任，又如何在外部信号松动后继续自我维持？四篇各attack一个环节，共用一条方法纪律——不把换名当创新，不把类比当机制，每一个新概念都必须与最邻近的既有构念划出可裁决的边界，每一篇都给出允许竞争理论获胜的研究设计与明确的失败条件。
Two stories dominate talk of AI and family education: one says AI frees parents from tutoring, the other says it opens a new digital divide. Both stop at the level of the task — who explains, corrects, retrieves, plans. These four papers ask one level down: when a knowledge task is taken over by AI, does the parent–child contact that used to happen through that task still occur? And how do a family's pressure, identity and responsibility keep sustaining themselves after external signals loosen? Each paper attacks one link, under one shared discipline: renaming is not innovation, analogy is not mechanism; every new concept must draw a decidable boundary against its nearest existing construct, and every paper offers a design that lets a rival theory win, with explicit conditions of failure.
从一个孩子的书桌，到一个家庭的身份，再到整套教育责任如何被制度悄悄配置给母亲——四篇由近及远，把「AI 时代的家庭教育」拆成四个可测的机制。
From a child's desk, to a family's identity, to how the whole burden of education is quietly assigned to mothers by institutions — four papers move from near to far, taking "family education in the age of AI" apart into four measurable mechanisms.
认知接触替代与控制补偿锁定Cognitive-Contact Substitution and Compensatory-Control Lock-In
AI 把讲题、纠错、检索、规划从亲子之间搬到人机界面。真正被改变的不是「谁完成了任务」，而是「亲子之间原本借由任务发生的接触是否还在」。当关系更新的隐性输入被切断，家长控制往往作为断供后的短期代偿而抬升——本文给出二维判别框架、可观测变量与角色分离型 AI 随机干预。
AI moves explanation, correction, retrieval and planning off the space between parent and child. What changes is not who finishes the task but whether the contact that used to happen through it survives. When relational updating is cut off, parental control tends to rise as a short-term compensation.
读全文 →Read → 之二 · 跨层机制模型Paper II · Cross-Level Mechanism双源驱动冗余与干预反弹Dual-Source Motivational Redundancy and Intervention Rebound
减负、评价改革、技术替代都以削弱压力源为目标，家庭反应却分裂为松弛、转隐与增压三种。本文提出「外部合法性供给—内部身份供给」双源模型：二者可代偿，故单端干预可能触发反弹。附 2×2 因子干预与自然政策纵向研究，以及明确的证伪条件。
Policies, evaluation reforms and technologies all aim to weaken sources of pressure, yet families split three ways: they relax, go covert, or intensify. A dual-source model — external legitimacy supply and internal identity supply — explains why single-sided intervention can rebound.
读全文 →Read → 之三 · 家庭身份脆弱性的理论模型Paper III · A Theory of Identity Vulnerability竞速依附型家庭与身份意义自持Competition-Dependent Families and Identity-Meaning Autonomy
外部竞速信号松动，许多家庭并未退出，而是把竞争迁往素质、项目、「会不会用 AI」。本文区分「关系凝聚」与「身份意义自持」，追问一个尖锐问题：一个家庭的意义，能否在没有外部成就证明时继续运作？附三源独立测量与目标冲击恢复指标。
As external competitive signals loosen, many families do not exit — they migrate competition to "quality", projects, "can you use AI". Separating relational cohesion from identity-meaning autonomy, the paper asks: can a family's meaning keep running with no external proof of achievement?
读全文 →Read → 之四 · 责任配置的多机制模型Paper IV · Multi-Mechanism Model制度性家庭化与教育劳动性别化Institutional Familialization and the Gendering of Educational Labor
母职耗竭与父职低参与，常被归咎于观念落后。本文给出「教育责任外部化—性别吸收链」四机制，并严格定义「制度性默许」——把立法沉默、违法加班、学校责任转嫁、福利覆盖不足分辨为不同的可测环节，附家庭教育制度负荷指数与 AI 平台审计设计。
Maternal exhaustion and low paternal involvement are usually blamed on outdated beliefs. This paper offers a four-mechanism "externalization–gender-absorption chain" and a strict definition of institutional acquiescence, separating legislative silence, illegal overtime, school offloading and welfare gaps into distinct measurable links.
读全文 →Read →四篇可以单读，也可以连读。连读的线索是「距离」：之一停在一个孩子的书桌前，问的是最贴身的接触；之二退到一个家庭内部，问它为什么在压力被拿走后仍不松手；之三退到家庭与家庭之间，问竞速松动后意义靠什么撑住；之四退到家庭之外，问是谁、用哪几道制度，把这一切的成本悄悄配置给了母亲。四篇都不报告已完成的经验检验——它们是理论建构与研究设计，把每个命题放到可以被数据推翻的位置上。150 分以上的最终认证，仍需独立的经验数据。
The four can be read alone or in sequence. The thread is distance: Paper I stops at a child's desk and the most intimate contact; Paper II steps back inside one family, asking why it won't let go after the pressure is removed; Paper III steps back between families, asking what holds meaning up once competition loosens; Paper IV steps outside the family, asking who — through which institutions — quietly assigns the cost to mothers. None reports completed empirical tests: they are theory-building and research design, each proposition placed where data could overturn it. Certification above 150 still requires independent empirical data.