資料來源#
摘要#
Kropp、Bedard、Wiles、Hsu、Krayer 在 HBR 2026/03("When using AI leads to brain fry")提出的術語,指的是過度使用或監督 AI,超出認知能力所造成的精神疲勞。出現腦霧的工作者回報錯誤頻率顯著更高——輕微錯誤頻率高出 11%,重大錯誤頻率高出 39%——相較於沒有出現腦霧的同儕。2026 年 5 月 HBR 後續論文將其視為可能在 AI 員工框架下累積的認知機制。
機制#
當員工監督 AI 輸出時:
- 將 AI 視為工具的框架——審查的認知負擔仍落在人類身上。大量使用 → 腦霧 → 錯誤增加 11–39%。
- 將 AI 視為員工的框架——人類可能覺得不必完全投入審查負擔(「ALEX-3 已經做過了」)。這可能在短期內藉由導致審查不足來減少腦霧症狀,但也因此產生另一種失敗模式。實驗中觀察到的錯誤捕捉率下降 18%,與此一致。
因此,兩種框架都有成本曲面:工具框架讓審查者承擔負擔;員工框架則以投入不足取代這項負擔。
對 Human-AI Accountability Redesign 的意涵#
腦霧是「只要擴大管理幅度」不可行的認知負荷原因。在不重新設計審查方式的情況下,增加每位人類審查者負責的 AI 輸出量:
- 超過某個門檻後,腦霧開始出現 → 錯誤率上升。
- 即使尚未達到該門檻,邊際審查品質也會下降。
論文暗示的重新設計選項:
- 縮小審查範圍(以抽樣稽核取代逐一輸出審查)
- 將審查集中於高風險決策點(決策權閘門,參見 Claude Code Auto Mode)
- 將人類角色從逐一輸出審查轉為系統層級監督(編排品質、效能監控)
- 重新設定績效管理,獎勵編排能力,而非逐一捕捉錯誤
與程式設計工作流程研究的連結#
- Context Window Smart Zone——模型端的類比認知限制。模型超過約 100K tokens 後敏銳度下降;人類超過自身監督能力後也會失去敏銳度。兩者都有一個 smart zone,超過後效能下降速度會快於容量所暗示的程度。
- Harness Shrinkage as Models Improve——更好的模型降低每項任務所需的審查量,部分緩解腦霧;但更快產出更多輸出的代理又會重新帶來數量壓力。
- Agent Loop Pattern——迴圈是強力的輸出倍增器;腦霧則是它們在人類端碰到的限制。
相關連結#
- 將思考外包,而非理解——過度委派造成的理解變薄,是監督疲勞在認知負荷上的近親
- 自動化—樂觀連結——反向訊號:Anthropic 的 AEI 調查發現,重度委派者回報沒有學習缺口,且感受到更高的技能價值。不同的測量工具(自我回報對比測得錯誤)與機制(委派感受對比監督疲勞)形成了這個頁面的錯誤資料所凸顯的感受與測量之間的張力
- 生成式 AI 的實驗性學習影響——同樣將「客觀測量勝過自我回報」的做法應用於學習任務:Contractor & Reyes 隨機分配 AI 存取權,發現自動化模式使用者的進步在移除 AI 後消失——這是本頁認知成本故事中去技能化的一面,以因果方式測量,而非透過調查
- 驗證成為新的瓶頸——審查/驗證負擔正是監督疲勞累積之處
- 迴圈工程——Osmani 所說的「你實際能執行多少[迴圈],取決於你的審查頻寬,而不是工具」正是在迴圈層級命名這個上限:worktree 消除了機械性的衝突,但腦霧仍是人類端的限制
- 伴隨概念:AI 員工框架
- 重新設計目標:Human-AI Accountability Redesign
- 認知類比:Context Window Smart Zone(模型端)
- 輸出倍增器:Agent Loop Pattern
- 緩解方式:Claude Code Auto Mode(決策權)、系統層級編排
- 監督品質風險:Compute Allocator——「compute allocator」角色假設人類能做出良好決策;腦霧則是分配者改為不加思索地蓋章批准的失敗模式
- 單人創辦人放大器:Founder as Agent Orchestrator——同時執行許多平行代理工作階段,會比以人數為基礎的組織更快讓監督負擔超過腦霧門檻
- 加速反噬——Faros AI 對同一疲勞現象所做的組織規模遙測:每位開發者每日 PR 情境增加 67.4%、工作重啟增加 13.8%,且 31.3% 的 PR 在沒有審查的情況下合併——這是在 4,000 個團隊中測得的投入不足失敗模式
- 平行代理編排——並行處理下監督疲勞的上限:p99 OpenAI 使用者每天透過許多並行代理執行約 71 個代理小時,但代理執行時間總和並不等於人類注意力——每個代理的審查負荷在哪裡飽和,正是這個門檻
- 未知數作為代理瓶頸——對「不理解就批准」的反制措施:Thariq Shihipar 的測驗閘門(「我只有在完美通過測驗後才合併」)讓合併取決於審查者的理解,而非其簽名
- 審查作為控制點——同一疲勞機制,來源不是受控實驗,而是實務工作者的討論:更高的審查負荷會降低審查深度與動機,滑向不加思索地蓋章批准(「審查者也許能撐過一個 sprint,但他們會精疲力竭,或開始不加思索地蓋章批准」)——這正是 CMU 理論的 P2/P3
- 可設定的人類參與——參與成本的一面:HAS-Bench 的互動成本指標(回合數/人類步驟/tokens),以及「更多管道 ≠ 更好」的結果(在 6 種模式中的 5 種,最佳單一管道勝過所有管道;A4 過度介入會破壞原本已解決的任務),量化了過度詢問與時機不當的介入確實帶有成本——更多人類參與並非免費
推導#
- 編排與員工框架:調和創辦人實務指南與 HBR 的責任證據——將腦霧命名為該實務指南「精實的 10 人獨角獸」主張中未處理的成本曲面;提出以受限並行 + 抽樣審查 + 高風險集中,作為單人創辦人的緩解方案
資料來源#
- Research: Why You Shouldn’t Treat AI Agents Like Employees (May 2026, references brain fry)
- Original paper: Kropp et al., When using AI leads to brain fry, HBR 2026/03
Cited by 30
- Human-in-the-Loop Boundaries×5
Redesign the loop when the human is nominally accountable but cognitively overloaded; that is the…
- Does the Human-Facing Harness (HTML Artifacts) Hit Its Own Bloat Ceiling?×4
The binding constraint is "human attention and judgement, not generation cost" (Compute Allocator).…
- Opinions on Using AI Tools & the Future of the Software Engineering Role×3
"Brain fry" is real and measurable. Ai Brain Fry: mental fatigue from oversight beyond cognitive…
- The Automation–Optimism Link×3
Ai Brain Fry — the direct tension: measured oversight fatigue and error increases vs. self-reported…
- Is Human Review of AI-Authored Code Still a Real Control, or Already Rubber-Stamping?×3
But the failure mode is a threshold, not a destiny — the countermeasures are also in evidence. What…
- Orchestration vs Employee Framing: Reconciling the Founder's Playbook with HBR's Accountability Evidence×3
The error surface is real. Playbook flags it via Agentic Technical Debt, Zero Friction Scope Creep,…
- Parallel Agent Orchestration×3
Ai Brain Fry — the cognitive cost of overseeing many parallel streams; the oversight-fatigue limit…
- Acceleration Whiplash×2
Ai Brain Fry — the cognitive-load channel: context-switching and under-review are the human-side…
- AI Employee Framing×2
Brain-fry-adjacent disengagement. When output is "from an employee," reviewers may feel less need…
- Experimental Learning Impact of Generative AI×2
Ai Brain Fry — both put an objective, measured number on AI's cognitive effect (there, oversight…
- Founder as Agent Orchestrator×2
How does the orchestration role change the founder's decision burden? Fewer hands-on tasks but more…
- Loop Engineering×2
A fourth thread runs through the skills primitive: without skills the loop re-derives your whole…
- Outsource Your Thinking, Not Your Understanding×2
The framing Thawar gives it is physiological rather than economic — "the brain is a muscle; if you…
- Risk-Tiered Auto-Approval×2
A refusal doesn't dump the PR back into a queue; it routes to a subject-matter expert, selected by…
- Security Debt of Agent-Generated Code×2
Ai Brain Fry — the security register of oversight fatigue: humans committed 67.6% of the genuine…
- Unknowns as the Agentic Bottleneck×2
That is a direct, testable answer to a problem stated three ways across the wiki and solved in none…
- Agent Loop Pattern
Ai Brain Fry — the human-side limit on output multipliers: more loop output → more review → more…
- Claude Code
> Reading this as evidence — interpretation, flagged. Taken together the caps, the workflow-size…
- Claude Code Auto Mode
Ai Brain Fry — concentrating human review on high-stakes decision points rather than every action…
- Compute Allocator
Does treating humans as "compute allocators" risk the oversight-fatigue / accountability failure…
- Configurable Human Participation
Ai Brain Fry — the interaction-cost metrics (turns / human steps / tokens) and the "more channels ≠…
- Context Window Smart Zone
Ai Brain Fry — human-side analog of the smart zone: oversight has its own degradation curve past…
- Harness Shrinkage as Models Improve
Ai Brain Fry — partially mitigated by harness shrinkage (less to oversee), reintroduced by output…
- Human-AI Accountability Redesign
HBR five-pillar prescription: span-of-control redesign, role redesign, performance management reset, decision-rights/es…
- AI Economics & Labor
Ai Brain Fry — Kropp et al. 2026/03: mental fatigue from excessive AI oversight increases minor…
- The Orchestrator's Real Workload: Decision Burden, Framing Discipline, and Whether Taste Scales
The load that arrives is the error-prone kind. Oversight fatigue raises minor errors +11% and major…
- Output Length Calibration
Ai Brain Fry — narration volume is oversight load: more per-message output across more parallel…
- Review as the Control Point
Ai Brain Fry — review load → fatigue → rubber-stamping (P2/P3) is the oversight-fatigue mechanism,…
- The Tragedy of the Cognitive Commons
Ai Brain Fry — the other cost of oversight: brain fry measures the fatigue of validating, this…
- Verification as the New Bottleneck
Ai Brain Fry — the risk if verification stays manual: oversight fatigue increases errors as volume…
Related articles
- Harness Shrinkage as Models Improve
Prompt scaffolding shrinks each model release; Cat Wu's pruning discipline; Boris Cherny "100 lines of code a year from…
- Human-AI Accountability Redesign
HBR five-pillar prescription: span-of-control redesign, role redesign, performance management reset, decision-rights/es…
- Verification as the New Bottleneck
Fiona Fung: coding is no longer the bottleneck — verification, review, maintenance are; shift-left; TDD loses its tax;…
- Claude Code
Anthropic's agentic coding product; created by Boris Cherny late 2024; TypeScript/React on Bun (itself Claude-rewritten…
- Agentic Technical Debt
Debt that *compounds* (not just accumulates) because each agentic-coding session re-derives architectural decisions wit…
