Post
JA EN

AI Doesn't Reduce Effort — It Redistributes It: The Paradox of 'Kurō Cancel' and Workload Creep

Posted
By
15 min read
AI Doesn't Reduce Effort — It Redistributes It: The Paradox of 'Kurō Cancel' and Workload Creep
  • Target audience: IT engineers and knowledge workers who use AI tools daily
  • Prerequisites: Basic experience with AI coding tools or LLMs
  • Reading time: 20 minutes

Overview

“AI will make work easier” — that’s what most people expect. But reality is betraying this naive expectation in two opposite directions. On one side, Japan’s emerging “Kurō Cancel” (effort-cancellation) trend is spreading a lifestyle where AI eliminates every friction from daily life, eroding people’s sense of achievement and personal growth. On the other side, an ethnographic study from UC Berkeley’s Haas School of Business reveals that employees post-AI adoption are handling a wider range of tasks, over longer hours, under greater cognitive load1. AI doesn’t reduce effort — it redistributes it. It strips too much from daily life while piling too much onto work. This article examines the mechanisms behind this “effort redistribution” through the lens of psychology and organizational behavior research, and explores what engineers can do about it.

The Structure Behind “AI Will Make Things Easier”

The efficiency gains from AI tools are undeniable. Automated code generation, document summarization, email drafting — these genuinely speed up individual tasks.

But “individual tasks getting faster” and “overall effort decreasing” are not the same thing. Overlooking this distinction is the starting point for two paradoxes.

flowchart TB
    A["AI speeds up<br>individual tasks"] --> B["Expectation:<br>Total effort decreases"]
    B --> C{"Reality?"}
    C -->|Daily life| D["Too much friction<br>disappears<br>(Kurō Cancel)"]
    C -->|Work| E["Tasks proliferate<br>(Workload Creep)"]
    D --> F["Loss of achievement<br>& growth"]
    E --> G["Burnout &<br>boundary collapse"]

    classDef warningStyle stroke:#d29922,stroke-width:3px
    class F,G warningStyle

As this diagram shows, the same “AI-driven efficiency” produces diametrically opposite problems depending on context. Let’s examine each in turn.

The Vanishing Side: “Kurō Cancel” and the Value of Friction

What Is “Kurō Cancel”?

In 2026, Nikkei Trendy — one of Japan’s most influential trend-forecasting publications — featured “Kurō Cancel Kaiwai” (苦労キャンセル界隈, roughly “the effort-cancellation crowd”) in its annual hit predictions2. The term describes a growing lifestyle movement in Japan where people use generative AI and convenience services to systematically “cancel” every inconvenience from their daily lives.

Real-time translation earphones cancel the struggle of language learning. AI-powered shopping assistants cancel the effort of comparing products. Generative AI handles everything from meal planning to scheduling, canceling the labor of “having to think.” The trend directly challenges a deeply rooted Japanese proverb: “Kurō wa katte demo seyo” (苦労は買ってでもせよ) — “Seek out hardship even if you have to pay for it.”

On its face, this seems perfectly rational. Why endure unnecessary struggle? But psychological research shows that not all struggle is unnecessary.

The IKEA Effect: How Effort Creates Attachment

Norton, Mochon, and Ariely’s 2012 study demonstrated that people place disproportionately high value on things they’ve invested labor in creating3. Across experiments involving IKEA furniture assembly, origami, and LEGO blocks, participants valued their own amateur creations on par with expert-made products.

Crucially, this effect depended on task completion. Participants who were interrupted mid-task or whose creations were destroyed afterward showed no such effect. In other words, it’s not effort itself that creates value — it’s the experience of accomplishing something through effort.

Applying this to the AI era reveals a troubling pattern:

Without AIWith AIWhat’s Lost
Writing code yourself and making it workCopy-pasting AI-generated codeSense of completion
Learning to cook through trial and errorFollowing AI recipes step by stepJoy of mastery
The thrill of being understood in a foreign languageInstant communication via real-time translationFeeling of growth

When AI “cancels” effort, what may simultaneously get canceled is the attachment and sense of achievement proportional to labor that the IKEA effect produces.

Desirable Difficulties: The “Productive Friction” Learning Requires

Robert Bjork’s theory of “desirable difficulties” further clarifies the value of friction in learning4.

Bjork’s research identifies four types of desirable difficulty:

  1. Spacing — distributing practice over time rather than cramming
  2. Interleaving — mixing different topics rather than blocking
  3. Retrieval practice — forcing yourself to recall rather than re-reading
  4. Varying conditions — changing the practice environment

All share one trait: they feel inefficient in the moment. Short-term performance dips, but long-term retention and transfer improve dramatically.

AI-powered “effort cancellation” works in exactly the opposite direction — eliminating these productive frictions. The struggle of deciphering error messages on your own. The difficulty of weighing architectural alternatives yourself. The effort of reading and understanding official documentation. These are tedious, but according to Bjork’s theory, they are precisely the “desirable difficulties” that learning demands.

Csikszentmihalyi’s Flow Theory: What Happens When Challenge Disappears

Mihaly Csikszentmihalyi’s flow theory frames this problem in an even larger context5. Flow — that state of complete absorption where you lose track of time — occurs when skill and challenge are in balance.

Below is a simplified version of the skill-challenge model from flow theory (the original presents an 8-channel, two-axis model):

quadrantChart
    title Skill-Challenge Balance (Simplified)
    x-axis Low Skill --> High Skill
    y-axis Low Challenge --> High Challenge
    quadrant-1 "Flow State<br>(Absorption & Growth)"
    quadrant-2 "Anxiety &<br>Overwhelm"
    quadrant-3 "Apathy"
    quadrant-4 "Boredom<br>← AI pushes here"

When AI removes too much challenge, skilled individuals are most susceptible to falling into the “boredom” zone. When veteran engineers let AI write all their code and feel “it’s just not interesting anymore,” that’s not mere nostalgia — it’s the conditions for flow state breaking down.

The Growing Side: AI-Driven Workload Creep

While effort vanishes too easily from daily life, the opposite is happening at work.

The Berkeley Haas Study: AI Makes Work Expand

Ranganathan and Ye (2026) at UC Berkeley’s Haas School of Business conducted an 8-month ethnographic study of approximately 200 employees, tracking how workloads changed after AI adoption1. The results defied widespread expectations.

Through interviews and participant observation, they found that while AI “streamlined” individual tasks, the total volume of work actually increased. The researchers identified three mechanisms driving this expansion:

1. Task Expansion

As AI makes more things “possible,” more things become “expected.” Product managers start writing code with AI. Researchers take on engineering tasks. Boundaries that once defined “not my job” dissolve.

2. Blurred Boundaries

Sending prompts during lunch. Checking AI outputs between meetings. Running one more exchange in the evening. Because AI responds 24/7, natural stopping points vanish. The line between work and rest blurs.

3. Increased Multitasking

While AI processes Task A, you start on Task B and simultaneously review the output of Task C. Cognitive load per unit time escalates. Time that was supposedly “freed up” gets filled with more parallel tasks.

Parkinson’s Law, AI Edition

This phenomenon can be understood as an AI-era version of the law C. Northcote Parkinson articulated in 1955: “Work expands to fill the time available for its completion”6.

Parkinson’s original observation was an essay about British bureaucracy, but the insight is universal. When AI saves 30 minutes on Task A, those 30 minutes don’t become rest. They get filled by a new Task B. And when Task B is also optimized by AI, Task C materializes.

Microsoft’s Work Trend Index (2025) reports that 68% of employees struggle with the pace and volume of work, and 46% report burnout7. AI is assembling the conditions for what some call the “infinite workday” — a state where the upper bound on daily work output ceases to exist.

Engineering-Specific Workload Creep

In engineering, workload creep takes a distinctive form:

Task AI AcceleratesNew Task It Creates
Code generationReviewing and fixing generated code
Test generationValidating test quality and coverage
DocumentationVerifying documentation accuracy
Bug fix suggestionsChecking proposed fixes for side effects
Auto-generated PR descriptions“Might as well” polish them further

Each addition is small, but they accumulate and encroach on core development time. Buell and Norton (2011) demonstrated the “labor illusion” — how making effort visible increases perceived value8. Applying this insight (which goes beyond their original claim), the “effort of verifying AI output” readily becomes legitimized as necessary work. The underlying mechanism — visible effort converting into “we should do this” judgments — is shared.

Why the Betrayal Goes Both Ways: An Integrated Explanation

“Too little effort in daily life, too much at work” — this seemingly contradictory phenomenon actually stems from the same underlying structure.

Cognitive Bias Around the Value of Effort

Aronson and Mills’ classic 1959 study on effort justification showed that humans tend to value things in proportion to the effort invested9. Participants who underwent a harsh initiation rated their group as more “interesting” than those who joined easily — even when the group activity was identical.

This cognitive bias underlies both paradoxes:

flowchart TB
    A["Humans value things<br>in proportion to effort"] --> B["Daily life: When effort<br>disappears, so does<br>satisfaction"]
    A --> C["Work: Can't let go of<br>tasks you've invested<br>effort in"]
    C --> D["Scope expansion:<br>'If AI can do it,<br>we should do it'"]
    D --> E["Workload Creep"]
    B --> F["Dilution of Meaning"]

The Visibility Paradox

Here I offer my own analysis: AI simultaneously lowers the “execution cost” of tasks while increasing their “visibility.” When you can see that “AI could also handle this” and “that too,” choosing not to do something becomes psychologically difficult. This may operate through dynamics similar to what Schwartz (2004) described in The Paradox of Choice — how an increase in options triggers decision fatigue10.

In work contexts, visible tasks readily convert into “tasks we should do.” In daily life, tasks AI handles get downgraded to “tasks we never needed to do.” The same visibility mechanism may produce opposite outcomes depending on context.

The Asymmetric Collapse of Boundaries

Work and daily life differ in their capacity for boundary-setting — defining “this is enough.”

  • Daily life: If AI can handle it, people gladly let go (boundaries collapse easily, effort approaches zero)
  • Work: As AI makes more things possible, managers, colleagues, and you yourself raise expectations (boundaries expand, effort increases)

This asymmetry is the essential structure that produces the “redistribution” of effort.

Implications for Engineers

“Strategic Inefficiency” as a Concept

Rather than eliminating all friction with AI, the key is to intentionally choose which friction to preserve.

Type of FrictionDelegate to AI?Reasoning
Boilerplate codeYesLow learning value, repetitive work
Deciphering error messagesIt dependsWorth doing yourself for unfamiliar errors or new libraries. Delegate for known errors or under time pressure
Architectural design decisionsNoDirectly connected to flow state and deep understanding
Researching a new APIIt dependsRead official docs yourself for APIs you’ll use repeatedly. Ask AI for one-off usage or prototyping
Code review judgmentsNoEssential training for critical thinking

Per Bjork’s desirable difficulties theory, activities that are “inefficient short-term but build skills long-term” should not be handed to AI4.

Countering Workload Creep

The task expansion documented by the Berkeley Haas study demands conscious boundary-setting1:

  1. Distinguish “can do” from “should do”: Just because AI makes a task possible doesn’t mean it falls within your responsibilities
  2. Don’t sync your schedule to AI’s availability: AI responds 24/7, but you don’t have to. Don’t repurpose break time as work time just because “I sent it to AI”
  3. Cap parallel tasks: Even if AI can process three things simultaneously, your review bandwidth is finite

Designing the “Right Amount” of Effort

As Csikszentmihalyi’s flow theory shows, too little challenge leads to boredom, too much to anxiety5. The “right amount” when collaborating with AI can be framed as:

  • Intentionally preserve difficulty in areas where your skills are growing
  • Delegate to AI work you’ve already mastered and that’s purely repetitive
  • In areas you want to learn, use AI as a “sparring partner” rather than an answer key
  • Set work boundaries based on your capacity, not AI’s capabilities

Of course, the optimal balance varies by individual skill level, job responsibilities, and team dynamics. Here are some concrete practices to adapt to your situation:

  • Weekly “AI delegation review”: Spend 15 minutes at the end of each week listing “tasks I delegated to AI” and “tasks I did myself.” Check whether high-learning-value tasks are disproportionately ending up on the AI side, and adjust the following week’s allocation
  • “AI-free 30-minute blocks”: Set aside 30 minutes each day where you intentionally close AI tools and write code. The goal isn’t skill maintenance for its own sake — it’s ensuring “adequate challenge” to enter flow state. Per Bjork’s desirable difficulties theory, this brief inefficiency contributes to long-term skill retention4

Conclusion

AI is not a technology that reduces effort. It’s a technology that redistributes it.

From daily life, it strips away too much of the friction that enables achievement, growth, and flow state. Onto work, it piles additional load through three mechanisms: task expansion, boundary blurring, and increased multitasking.

Norton et al.’s IKEA effect3, Bjork’s desirable difficulties4, Csikszentmihalyi’s flow theory5 — these studies consistently demonstrate that “appropriate effort” is not a cost to be eliminated, but the wellspring of fulfillment and growth.

Meanwhile, Ranganathan and Ye’s research1 shows that in work contexts, that effort can expand beyond healthy limits.

What’s needed is neither asking AI to “make everything easy” nor rejecting AI’s convenience. It’s consciously designing which effort to keep and which to let go. Whether in individual skill development or team workflow design, this will become a core literacy of the AI era.

Explore other articles on related themes:

References

References are listed in order of citation number as they appear in the text.

Additional References (Not Cited by Number in Text)

On citation accuracy: The research cited in this article has been verified through the following methods:

  • Confirmation via academic databases (PubMed, Google Scholar, ScienceDirect, etc.)
  • Verification of paper details on official journal websites
  • Cross-referencing through multiple independent sources (academic media, official announcements from research institutions, etc.)

For some papers, direct access to full-text PDFs may be restricted. However, abstracts, DOIs, author information, and key findings have been confirmed through official academic databases and reliable secondary sources.

  1. AI Doesn’t Reduce Work – It Intensifies It - Ranganathan, A., & Ye, X. M., Harvard Business Review (2026). Ethnographic study of ~200 employees over 8 months. UC Berkeley Haas School of Business. [Reliability: Medium-High] ↩︎ ↩︎2 ↩︎3 ↩︎4

  2. Nikkei Trendy 2026 Hit Predictions - Nikkei BP (2026). Coverage of the “Kurō Cancel” trend. [Reliability: Medium-High] ↩︎

  3. The IKEA Effect: When Labor Leads to Love - Norton, M. I., Mochon, D., & Ariely, D., Journal of Consumer Psychology, 22(3), 453-460 (2012). Peer-reviewed. Four experiments demonstrating overvaluation of self-made products. [Reliability: High] ↩︎ ↩︎2

  4. Memory and Metamemory Considerations in the Training of Human Beings - Bjork, R. A., in Metacognition: Knowing About Knowing, MIT Press (1994). Introduction of the “desirable difficulties” concept. Peer-reviewed book chapter. Additional reference: Bjork, R. A., & Bjork, E. L. (2020). Desirable Difficulties in Theory and Practice. Journal of Applied Research in Memory and Cognition, 9(4), 475-479. [Reliability: High] ↩︎ ↩︎2 ↩︎3 ↩︎4

  5. Flow: The Psychology of Optimal Experience - Csikszentmihalyi, M., Harper & Row (1990). ISBN: 978-0-06-092043-8. Seminal work on flow theory. Additional reference: Nakamura, J., & Csikszentmihalyi, M. (2002). The Concept of Flow. In Handbook of Positive Psychology, Oxford University Press. [Reliability: High] ↩︎ ↩︎2 ↩︎3

  6. Parkinson’s Law - Parkinson, C. N., The Economist, November 19, 1955. The original articulation of “work expands to fill the time available.” Book version: Parkinson’s Law: The Pursuit of Progress, John Murray (1958). ISBN: 978-1-56849-015-3. [Reliability: High] ↩︎

  7. 2025 Annual Work Trend Index - Microsoft (2025). Large-scale survey based on Microsoft 365 telemetry data. 68% struggle with workload, 46% report burnout. [Reliability: Medium-High] ↩︎

  8. The Labor Illusion: How Operational Transparency Increases Perceived Value - Buell, R. W., & Norton, M. I., Management Science, 57(9), 1564-1579 (2011). Peer-reviewed. Demonstrates how making effort visible increases perceived value. [Reliability: High] ↩︎

  9. The Effect of Severity of Initiation on Liking for a Group - Aronson, E., & Mills, J., Journal of Abnormal and Social Psychology, 59(2), 177-181 (1959). Peer-reviewed. Classic experiment on effort justification. [Reliability: High] ↩︎

  10. The Paradox of Choice: Why More Is Less - Schwartz, B., Harper Perennial (2004). ISBN: 978-0-06-000569-6. Examines how increased options reduce decision quality and satisfaction. Based on psychological research. [Reliability: Medium-High] ↩︎

AI & Technology
AI work style psychology cognitive science IKEA effect Parkinson's law flow theory workload
This post is licensed under CC BY 4.0 by the author.