Attention-Commitment Learning / 注意力承诺式学习

Definition

Attention-commitment learning is the view that durable understanding requires sustained time, active questioning, friction, association, and personal reconstruction even when AI can supply explanations or summaries instantly.

Current Synthesis

AI can change the route through difficult material without abolishing the cognitive investment. A reader might replace linear reading with repeated chapter-level dialogue, thematic comparison, or immediate questions about intellectual lineage. If that process persists across evenings and demands clarification, it can still constitute a serious attention commitment; a brief summary consumed and forgotten usually does not.

Writing has a parallel function. Generating polished prose can remove the struggle through which an author discovers and remembers a position. Drafting personally and using AI for later feedback preserves more of that epistemic work. Efficiency therefore has to be evaluated against the purpose of the activity: if the friction builds judgment, memory, voice, or unexpected association, removing it may remove part of the learning.

Key Claims

• Explanation speed and learning speed are not the same because memory and understanding require engagement over time.
• AI dialogue can support serious reading when it sustains questions, clarification, comparison, and return.
• Difficult, redundant, or unexpected material can create productive associations absent from goal-only summaries.
• Self-written drafts preserve thought formation and authorial memory that generation may bypass.
• Reading and writing formats may change while their attention requirement remains.
• The value of friction depends on purpose; not every slow or difficult step is educational.

Evidence

• Fast-explanation limit - E257 对话少楠：明明 AI 有问必答，我怎么还总是信息焦虑？ reports that quickly understood AI explanations were quickly forgotten.
• Dialogic depth - E257 对话少楠：明明 AI 有问必答，我怎么还总是信息焦虑？ describes a multi-evening AI-guided inquiry into Being and Time that required repeated questions and left a stronger impression.
• Productive non-target material - E257 对话少楠：明明 AI 有问必答，我怎么还总是信息焦虑？ argues that long or difficult passages can interrupt the reader’s plan and generate new connections.
• Writing as thinking - E257 对话少楠：明明 AI 有问必答，我怎么还总是信息焦虑？ reports that AI-generated prose was hard for 少楠 to remember or own, leading him to write first and request feedback later.

Counterevidence & Qualifications

This framework rests on one speaker’s experience and interpretive discussion, not controlled learning research. AI summaries can improve accessibility, orientation, translation, and efficient review; difficult prose can also waste attention or exclude readers without adding insight. Personal drafting is not required for every utilitarian document, and engineering or backstage contexts may benefit from AI’s consistency. The relevant question is which cognitive work the task is meant to preserve.

What Changed

• Created the concept by joining reading persistence, dialogic inquiry, writing-as-thinking, and productive friction.
• Distinguished purposeful attention from an automatic preference for slowness.

Related Concepts

• AI-Assisted Reading - tool-supported reading practice that can either deepen or bypass engagement.
• Non-Instrumental Literary Reading - reading whose value includes experience beyond extracted conclusions.
• Reading As Frame Training - use of reading to develop the reader’s own interpretive structure.
• AI Writing Pedagogy - educational design question about which parts of composition learners should retain.
• AI Use Pacing - boundary against generating more material than a person can review and integrate.
• Human Judgment Under AI - capability strengthened when learners preserve evaluative practice.