Human Activity Delegation Agency Chart (HADAC)

A Comprehensive Taxonomy of Human Activities and Their Delegability to AI

What happens to the human when everything delegable is delegated?

---

Overview

The Human Activity Delegation Agency Chart (HADAC) is a research framework that maps the complete landscape of human activities against their potential for delegation to artificial intelligence. Unlike existing automation taxonomies (Parasuraman et al., 2000; SAE J3016) that focus on single tasks within specific domains, HADAC asks a broader question: across the totality of human life, what must remain human, what can be handed off, and what exists in the contested space between?

This project has three goals:

1. Map — Create a comprehensive, empirically-grounded taxonomy of all human activities categorized by delegability
2. Measure — Design and conduct a large-scale survey (N=1,000–10,000) to understand delegation preferences across demographics, cultures, and contexts
3. Publish — Contribute findings to the scientific understanding of machine behavior and human-AI coexistence, targeting venues such as Nature Human Behaviour

---

Why This Matters

We are entering an era where AI agents can perform an increasing share of daily human activities — from scheduling meetings to writing code to managing finances. But the question of what humans want to delegate (as opposed to what is technically possible) remains underexplored.

The existing literature addresses:

• Levels of automation for specific tasks (Sheridan & Verplank, 1978; Parasuraman et al., 2000)
• Policy challenges of AI assistants (Ada Lovelace Institute, 2025)
• Delegation mechanisms between AI agents (Tomašev et al., 2026)
• Philosophical boundaries of machine capability (Arendt, 1958; Chemero, 2023)

What is missing is a unified framework that spans the entirety of human life — from biological necessities to existential choices — and empirically measures where people draw the line.

The Blurry Zone: Where the Interesting Science Lives

The most scientifically productive region of this chart is not the clear "yes" or "no" categories, but the blurry zone — activities where delegation is technically possible but where human preferences, values, cultural norms, and individual differences create a contested boundary. The blurry zone is where:

• Partial delegation is possible (AI drafts, human edits; AI researches, human decides)
• Individual variation is maximal (some people want AI therapists, others find the idea repulsive)
• Cultural differences emerge (collectivist vs. individualist attitudes toward AI caregiving)
• Temporal dynamics operate (what feels undelegable today may feel natural in 5 years)
• Sub-task decomposition reveals hidden structure (cooking = meal planning [delegable] + ingredient sourcing [delegable] + the act of cooking [blurry] + tasting [non-delegable])

---

The Complete Human Activity Delegation Chart

Legend

Symbol	Meaning
✅	Primary classification
◐	Partially applies
—	Does not apply

1. Biological & Embodied Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
1.1	Sleeping	—	✅	—	Requires your body and consciousness
1.2	Eating / Drinking	—	✅	—	Ingestion is first-person; meal prep is delegable
1.3	Breathing	—	✅	—	Autonomic; ventilators assist but don't replace
1.4	Physical exercise	—	✅	—	Muscles must be yours; programming/tracking → delegable
1.5	Sexual activity	—	✅	—	Embodied intimacy
1.6	Hygiene (bathing, grooming)	—	◐	✅	Assisted care exists; self-care is embodied
1.7	Giving birth / pregnancy	—	✅	—	Surrogacy is human delegation, not AI delegation
1.8	Breastfeeding	—	✅	—	Biological process
1.9	Healing / recovering from illness	—	✅	—	Body heals; treatment planning → delegable
1.10	Experiencing pain	—	✅	—	First-person qualia
1.11	Experiencing pleasure	—	✅	—	First-person qualia
1.12	Aging	—	✅	—	Biological process
1.13	Dying	—	✅	—	The ultimate non-delegable act
1.14	Digesting food	—	✅	—	Autonomic biological process
1.15	Sensing (seeing, hearing, tasting, smelling, touching)	—	✅	—	Qualia; sensor data ≠ experience
1.16	Proprioception / balance	—	✅	—	Embodied spatial awareness
1.17	Thermoregulation (feeling hot/cold)	—	✅	—	Environment control → delegable; feeling → not
1.18	Drug/substance experience	—	✅	—	Altered states are first-person
1.19	Dreaming	—	✅	—	Unconscious processing
1.20	Menstruation / hormonal cycles	—	✅	—	Biological; tracking → delegable

2. Cognitive & Mental Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
2.1	Memorizing information	—	—	✅	External memory (notes, AI) vs. internalized knowledge
2.2	Mental arithmetic / calculation	✅	—	—	Calculators solved this centuries ago
2.3	Logical reasoning	—	—	✅	AI can reason; but understanding?
2.4	Reading for information extraction	✅	—	—	Summarization works well
2.5	Reading for pleasure / aesthetic experience	—	✅	—	The experience IS the point
2.6	Learning a new concept	—	—	✅	AI can curate/explain; internalization is yours
2.7	Learning a physical skill (instrument, sport)	—	✅	—	Muscle memory is non-transferable
2.8	Generating novel ideas	—	—	✅	AI augments ideation; who "has" the idea?
2.9	Critical thinking / evaluation	—	—	✅	AI can analyze; judgment involves values
2.10	Daydreaming / mind-wandering	—	✅	—	Spontaneous inner experience
2.11	Meditation / mindfulness	—	✅	—	Attention training is first-person
2.12	Planning (personal life)	—	—	✅	AI can propose; you decide what matters
2.13	Strategic thinking	—	—	✅	AI assists analysis; vision is human
2.14	Making sense of suffering	—	✅	—	Meaning-making is existential
2.15	Forming beliefs	—	✅	—	Epistemic agency
2.16	Changing your mind	—	✅	—	Requires confronting your own commitments
2.17	Paying attention	—	✅	—	Directed consciousness
2.18	Introspection / self-reflection	—	✅	—	Examining your own experience
2.19	Imagination / visualization	—	—	✅	AI generates images; inner imagination differs
2.20	Spatial navigation (wayfinding)	—	—	✅	GPS delegated this; cognitive maps atrophy

3. Emotional & Psychological Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
3.1	Feeling emotions (joy, sadness, anger, fear)	—	✅	—	First-person affective states
3.2	Grieving	—	✅	—	Nobody can grieve for you
3.3	Falling in love	—	✅	—	Emergent relational experience
3.4	Processing trauma	—	✅	—	Therapeutic support → blurry; processing → yours
3.5	Experiencing boredom	—	✅	—	Generative emptiness
3.6	Feeling pride / shame / guilt	—	✅	—	Moral emotions tied to self-concept
3.7	Developing emotional resilience	—	✅	—	Built through experience
3.8	Empathizing with another person	—	✅	—	Requires felt understanding
3.9	Managing emotions / self-regulation	—	—	✅	AI can coach; regulation is practiced internally
3.10	Seeking therapy / psychological help	—	—	✅	AI therapy exists; efficacy debated
3.11	Feeling lonely	—	✅	—	AI companionship ≠ resolving loneliness
3.12	Experiencing awe / wonder	—	✅	—	Aesthetic-existential state
3.13	Forgiving someone	—	✅	—	Internal moral-emotional act
3.14	Feeling gratitude	—	✅	—	Affective appreciation
3.15	Worrying / anxiety	—	✅	—	Involuntary affective state

4. Social & Relational Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
4.1	Having a conversation (deep)	—	✅	—	Requires mutual presence and vulnerability
4.2	Having a conversation (small talk)	—	—	✅	AI chatbots already do this; but is it "having" a conversation?
4.3	Making friends	—	✅	—	Requires authentic self-disclosure
4.4	Maintaining friendships	—	—	✅	AI can help remember birthdays; showing up is yours
4.5	Parenting / caregiving	—	✅	—	Logistics delegable; presence and love are not
4.6	Being a romantic partner	—	✅	—	The relationship IS the person
4.7	Listening to someone	—	—	✅	AI can "listen"; therapeutic listening vs. human witnessing
4.8	Arguing / conflict resolution	—	—	✅	AI mediators exist; emotional stakes are human
4.9	Apologizing sincerely	—	✅	—	Requires genuine remorse from the offender
4.10	Comforting someone in distress	—	—	✅	Physical presence matters; AI provides 24/7 availability
4.11	Celebrating together	—	✅	—	Shared embodied joy
4.12	Making eye contact	—	✅	—	Embodied social signal
4.13	Hugging / physical affection	—	✅	—	Embodied
4.14	Teaching (live, interactive)	—	—	✅	AI tutors exist; mentorship involves presence
4.15	Mentoring	—	—	✅	Guidance + relationship; AI can provide information
4.16	Networking / building professional relationships	—	—	✅	Outreach delegable; trust-building is personal
4.17	Participating in community	—	✅	—	Showing up, belonging
4.18	Caring for elderly / sick family	—	—	✅	Logistics delegable; presence is not
4.19	Playing with children	—	✅	—	Embodied co-presence
4.20	Gift giving	—	—	✅	Selection can be AI-assisted; gesture is human
4.21	Hosting guests	—	—	✅	Logistics delegable; hospitality is personal
4.22	Being a good neighbor	—	—	✅	Small acts of presence
4.23	Gossiping	—	—	✅	Social bonding; AI could curate but why?
4.24	Saying "I love you"	—	✅	—	Meaning requires the speaker

5. Professional & Productive Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
5.1	Email management	✅	—	—	Triage, drafting, routing
5.2	Calendar/scheduling	✅	—	—	Fully automatable
5.3	Data entry	✅	—	—	Classic automation target
5.4	Data analysis	✅	—	—	AI excels here
5.5	Writing reports / documentation	✅	—	—	Drafting; voice/style → blurry
5.6	Writing code	✅	—	—	Implementation; architecture → blurry
5.7	Code review	—	—	✅	AI catches bugs; design judgment is human
5.8	Software architecture decisions	—	—	✅	AI proposes; human decides trade-offs
5.9	Project management	—	—	✅	Tracking → delegable; leadership → human
5.10	Hiring / interviewing	—	—	✅	Screening → delegable; culture fit → human
5.11	Firing / layoffs	—	✅	—	Ethical weight requires human accountability
5.12	Negotiation	—	—	✅	Research → delegable; reading the room → human
5.13	Sales / persuasion	—	—	✅	AI can personalize; trust is interpersonal
5.14	Public speaking / presenting	—	—	✅	Prep → delegable; stage presence → human
5.15	Leading a team	—	—	✅	Coordination → delegable; inspiration → human
5.16	Customer service	✅	—	◐	Routine → delegable; complex empathy → blurry
5.17	Accounting / bookkeeping	✅	—	—	Fully automatable
5.18	Legal research	✅	—	—	Information retrieval and analysis
5.19	Legal judgment / advice	—	—	✅	Research → AI; liability → human
5.20	Medical diagnosis	—	—	✅	AI matches/exceeds in imaging; holistic assessment → human
5.21	Surgery / medical procedures	—	—	✅	Robotic surgery exists; accountability → human
5.22	Financial trading	✅	—	—	Algorithmic trading dominates
5.23	Financial planning / strategy	—	—	✅	Analysis → AI; risk appetite → human
5.24	Market research	✅	—	—	Data gathering and analysis
5.25	Product design	—	—	✅	Generative design → AI; vision → human
5.26	Quality assurance / testing	✅	—	—	Automated testing
5.27	Translation / interpretation	✅	—	◐	Technical → delegable; literary/cultural → blurry
5.28	Transcription	✅	—	—	Solved problem
5.29	Filing taxes / regulatory compliance	✅	—	—	Rule-following
5.30	Invoicing / billing	✅	—	—	Administrative
5.31	Inventory management	✅	—	—	Optimization
5.32	Supply chain logistics	✅	—	—	Optimization
5.33	Research (literature review)	✅	—	—	Information synthesis
5.34	Research (hypothesis generation)	—	—	✅	AI can propose; scientific intuition → human
5.35	Research (experimental design)	—	—	✅	AI optimizes; what's worth investigating → human
5.36	Earning money	—	—	✅	Increasingly delegable; but agency over livelihood?
5.37	Making investment decisions	—	—	✅	AI advises; risk tolerance is personal

6. Creative & Artistic Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
6.1	Writing (creative / literary)	—	—	✅	AI generates text; authentic voice → human
6.2	Writing (personal / journaling)	—	✅	—	Self-expression; AI can't journal for you
6.3	Composing music	—	—	✅	AI generates music; emotional intent → human
6.4	Playing a musical instrument	—	✅	—	Embodied skill; the experience is the point
6.5	Singing	—	✅	—	Embodied vocal expression
6.6	Dancing	—	✅	—	Embodied movement
6.7	Painting / drawing (by hand)	—	✅	—	Embodied mark-making
6.8	Digital art / graphic design	—	—	✅	AI generates; curation/direction → human
6.9	Photography (taking photos)	—	—	✅	AI can compose; being there → human
6.10	Photo editing / post-processing	✅	—	—	Automated editing
6.11	Filmmaking / video production	—	—	✅	Technical → delegable; vision → human
6.12	Acting / performing	—	✅	—	Embodied expression; deepfakes → blurry
6.13	Sculpting / ceramics	—	✅	—	Embodied craft
6.14	Crafting / handwork (knitting, woodwork)	—	✅	—	Embodied; the process is valued
6.15	Cooking as creative act	—	—	✅	Recipe generation → AI; the cooking itself → embodied
6.16	Interior design	—	—	✅	AI generates; taste is personal
6.17	Fashion / personal style	—	—	✅	AI recommends; identity expression → human
6.18	Writing poetry	—	—	✅	AI can produce poems; authentic expression?
6.19	Storytelling (oral)	—	—	✅	Content → AI; performance → human
6.20	Curating (art, music, content)	—	—	✅	AI recommends; taste → human

7. Domestic & Daily Life Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
7.1	Meal planning	✅	—	—	Algorithmic; dietary needs as input
7.2	Grocery shopping (online)	✅	—	—	List generation + ordering
7.3	Grocery shopping (in-person)	—	—	✅	Delivery exists; browsing/selecting → embodied
7.4	Cooking (routine meals)	—	—	✅	Kitchen robots emerging; currently mostly human
7.5	Washing dishes	—	—	✅	Dishwashers exist; hand-washing persists
7.6	Cleaning house	—	—	✅	Roombas + cleaning services; not fully solved
7.7	Laundry	—	—	✅	Machines wash; folding/sorting → emerging
7.8	Home maintenance / repairs	—	—	✅	Diagnosis → AI; physical work → human (for now)
7.9	Gardening	—	—	✅	Automated irrigation; but many garden for pleasure
7.10	Pet care (feeding, walking)	—	—	✅	Auto-feeders exist; walking → embodied bond
7.11	Childcare logistics	✅	—	—	Scheduling pickups, activities
7.12	Moving / relocating	—	—	✅	Logistics → delegable; the transition → human
7.13	Home organization / decluttering	—	—	✅	Advice → AI; handling your stuff → personal
7.14	Driving / commuting	—	—	✅	Autonomous vehicles approaching full delegation
7.15	Paying bills	✅	—	—	Auto-pay
7.16	Managing subscriptions	✅	—	—	Administrative
7.17	Waste management / recycling	—	—	✅	Smart bins emerging
7.18	Setting up / configuring devices	✅	—	—	Auto-configuration

8. Existential & Meaning-Making Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
8.1	Choosing what to care about	—	✅	—	Meta-preference formation
8.2	Choosing your values	—	✅	—	Moral identity
8.3	Making a promise	—	✅	—	Requires commitment from the promisor
8.4	Taking responsibility	—	✅	—	Accountability is non-transferable
8.5	Making a sacrifice	—	✅	—	Giving up something you value
8.6	Deciding to have children	—	✅	—	Life-defining choice
8.7	Choosing a life partner	—	—	✅	AI matchmaking exists; the choice is yours
8.8	Choosing a career / vocation	—	—	✅	AI can assess fit; calling is felt
8.9	Spiritual practice / prayer	—	✅	—	First-person relationship with transcendence
8.10	Confronting mortality	—	✅	—	Existential reckoning
8.11	Finding purpose / meaning	—	✅	—	Constructed through lived experience
8.12	Deciding to end a relationship	—	✅	—	Moral-emotional weight
8.13	Voting / political participation	—	—	✅	Information → AI; the vote is yours (but AI influence?)
8.14	Forming identity	—	✅	—	Who you are is irreducibly yours
8.15	Setting life goals	—	—	✅	AI can suggest; aspiration is personal
8.16	Accepting failure	—	✅	—	Psychological integration
8.17	Experiencing regret	—	✅	—	Looking back with weight
8.18	Making peace with the past	—	✅	—	Internal reconciliation
8.19	Contemplating existence	—	✅	—	Philosophical self-awareness
8.20	Witnessing (being present for someone)	—	✅	—	Presence cannot be proxied

9. Civic & Political Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
9.1	Voting	—	✅	—	Democratic agency; proxy voting raises issues
9.2	Jury duty	—	✅	—	Legal judgment by peers
9.3	Protesting / activism (in-person)	—	✅	—	Bodies in space; digital activism → blurry
9.4	Digital activism / advocacy	—	—	✅	AI can amplify; authenticity questions
9.5	Community organizing	—	—	✅	Logistics → AI; mobilizing → human relationships
9.6	Volunteering (in-person)	—	✅	—	Embodied contribution
9.7	Volunteering (remote/digital)	—	—	✅	Some tasks → AI; human connection component
9.8	Paying taxes (computation)	✅	—	—	Calculable
9.9	Following laws	—	✅	—	Compliance is behavioral
9.10	Whistleblowing	—	✅	—	Moral courage, personal risk
9.11	Running for office	—	—	✅	Campaign ops → AI; candidacy → human
9.12	Serving in military	—	—	✅	Drones/autonomous weapons → deep ethical territory
9.13	Donating to causes	—	—	✅	AI can optimize giving; choosing what matters → human

10. Leisure & Recreation Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
10.1	Watching a film / TV show	—	✅	—	The watching IS the experience
10.2	Choosing what to watch	—	—	✅	Recommendation algorithms; taste → human
10.3	Playing video games	—	✅	—	Embodied play; fun is first-person
10.4	Playing board games / card games	—	✅	—	Social play
10.5	Attending live events (concerts, sports)	—	✅	—	Presence, atmosphere, shared experience
10.6	Traveling / tourism	—	✅	—	The experience; planning → delegable
10.7	Travel planning	✅	—	—	Research, booking, itinerary
10.8	Hiking / being in nature	—	✅	—	Embodied environmental experience
10.9	Swimming	—	✅	—	Embodied
10.10	Playing sports	—	✅	—	Embodied competition/play
10.11	Sunbathing / relaxing	—	✅	—	Embodied rest
10.12	Reading for pleasure	—	✅	—	Aesthetic experience
10.13	Collecting (stamps, art, etc.)	—	—	✅	Sourcing → AI; the passion → human
10.14	Puzzle solving (crosswords, sudoku)	—	—	✅	AI solves instantly; humans do it for the process
10.15	Social media browsing	—	—	✅	AI can curate; doom-scrolling is human compulsion
10.16	Partying / nightlife	—	✅	—	Embodied social experience
10.17	Taking a walk (aimless)	—	✅	—	Ambulatory contemplation
10.18	People-watching	—	✅	—	Observational pleasure
10.19	Exploring a new city	—	✅	—	Discovery requires presence

11. Communication & Expression Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
11.1	Writing emails (routine)	✅	—	—	Template-based
11.2	Writing emails (important/personal)	—	—	✅	Voice matters; AI can draft
11.3	Text messaging (routine)	✅	—	—	Quick replies
11.4	Text messaging (emotional/personal)	—	—	✅	Authenticity at stake
11.5	Phone calls (informational)	✅	—	—	AI call agents exist
11.6	Phone calls (personal)	—	✅	—	Hearing someone's voice matters
11.7	Video calls (meetings)	—	—	✅	AI avatars emerging; presence questions
11.8	Video calls (personal)	—	✅	—	Seeing someone
11.9	Writing a letter (handwritten)	—	—	✅	The handwriting is the message
11.10	Public speaking	—	—	✅	Content → AI; delivery → human
11.11	Posting on social media	—	—	✅	Content creation → AI; self-expression → human
11.12	Commenting / responding online	—	—	✅	Engagement → AI; authentic voice → human
11.13	Taking photos / videos of life	—	—	✅	Automated cameras exist; choosing what matters → human
11.14	Recording voice memos	—	—	✅	AI transcribes; your voice carries affect

12. Moral & Ethical Activities

#	Activity	Delegable	Non-Delegable	Blurry	Notes
12.1	Making ethical decisions	—	✅	—	Moral agency requires a moral agent
12.2	Judging right from wrong (personal)	—	✅	—	Conscience
12.3	Standing up for someone	—	✅	—	Moral courage
12.4	Admitting a mistake	—	✅	—	Requires vulnerability
12.5	Keeping a secret	—	—	✅	AI can be encrypted; trust is relational
12.6	Telling a difficult truth	—	✅	—	Moral weight and timing
12.7	Deciding who to trust	—	—	✅	AI can assess track records; gut feeling → human
12.8	Setting boundaries	—	✅	—	Self-knowledge + communication
12.9	Practicing integrity	—	✅	—	Consistency between values and action
12.10	Making reparations	—	—	✅	Financial → delegable; relational → human

---

Summary Statistics

Category	Total	Delegable	Non-Delegable	Blurry
1. Biological & Embodied	20	0	20	0
2. Cognitive & Mental	20	2	8	10
3. Emotional & Psychological	15	0	13	2
4. Social & Relational	24	0	12	12
5. Professional & Productive	37	17	1	19
6. Creative & Artistic	20	1	8	11
7. Domestic & Daily Life	18	5	0	13
8. Existential & Meaning-Making	20	0	16	4
9. Civic & Political	13	1	5	7
10. Leisure & Recreation	19	1	14	4
11. Communication & Expression	14	3	2	9
12. Moral & Ethical	10	0	7	3
TOTAL	230	30 (13%)	106 (46%)	94 (41%)

Key Insight

Only 13% of human activities are clearly delegable. 46% are clearly non-delegable. The remaining 41% — the blurry zone — is where the most interesting scientific, design, and policy questions live.

---

The Blurry Zone: Deep Analysis

The blurry zone is not a single category — it has internal structure. We identify four types of blurriness:

Type A: Sub-Task Decomposable

Activities where the blurriness resolves when you decompose the task into sub-components, some of which are delegable and some not.

Example: Cooking a meal

Sub-task	Delegable?
Finding a recipe	✅ Delegable
Adjusting for dietary needs	✅ Delegable
Ordering ingredients	✅ Delegable
Preparing ingredients (chopping, measuring)	◐ Emerging (kitchen robots)
Applying heat / cooking technique	◐ Emerging
Tasting and adjusting	❌ Non-delegable (qualia)
Plating / presentation	◐ Aesthetic judgment
Eating	❌ Non-delegable
Enjoying the meal with others	❌ Non-delegable

Example: Writing a paper

Sub-task	Delegable?
Literature search	✅ Delegable
Reading and understanding papers	◐ AI summarizes; deep understanding is human
Identifying gaps	◐ AI assists; intuition is human
Forming a thesis	❌ Non-delegable (intellectual commitment)
Drafting text	✅ Delegable
Revising for argument quality	◐ AI checks logic; conviction is human
Polishing prose	✅ Delegable
Deciding to submit	❌ Non-delegable

Type B: Individually Variable

Activities where delegability depends on the individual person — their values, culture, personality, or life stage.

Activity	Some people delegate	Others refuse	Why it varies
Therapy	AI therapy users	"I need a human"	Trust, vulnerability, beliefs about consciousness
Cooking	Meal kit subscribers	"Cooking is my meditation"	Identity, culture, pleasure
Driving	Autonomous vehicle users	"I love driving"	Control, pleasure, trust
Email	Full AI delegation	"My words matter"	Professional identity, relationship value
Gift selection	AI-recommended	"I know my people"	Relational intimacy, effort as signal
Prayer	Guided meditation apps	"This is between me and God"	Sacred vs. functional framing

Type C: Temporally Shifting

Activities where the blurriness is resolving over time — either toward delegable (technology improving) or toward non-delegable (cultural backlash).

Activity	Direction	Driver
Medical diagnosis	→ Delegable	AI accuracy improving
Translation	→ Delegable	Neural MT quality
Driving	→ Delegable	Autonomous vehicles
Handwriting letters	→ Non-delegable	Scarcity makes it more meaningful
In-person meetings	→ Non-delegable	Remote fatigue; presence premium
Vinyl records / film photography	→ Non-delegable	Analog revival as resistance
Social media posting	→ Delegable	AI content generation normalizing

Type D: Culturally Dependent

Activities where delegability varies by culture, not just individual preference.

Activity	Cultural variation
Elderly care	Scandinavian models → institutional (delegated); East Asian → family obligation (non-delegable)
Childcare	US → daycare normalized; many cultures → parent-primary
Cooking	Fast-food cultures → highly delegated; food cultures (Italy, Japan) → sacred
Death rituals	Some → professional funeral industry; others → family-led
Education	Some → online/AI tutoring; others → teacher-student relationship sacred
Marriage arrangement	Arranged marriages → delegated to family; Western → individual choice

---

Design Space: New Research Directions

1. The Delegation Comfort Spectrum

Rather than binary delegable/non-delegable, we propose a continuous Delegation Comfort Score (DCS) measured per-person per-activity:

0 ────────────────────────────────────── 10
"I would never             "I would gladly
 delegate this"              delegate this"

The DCS captures individual willingness, not just technical feasibility. This is the core measurement in our proposed survey.

2. The Delegation Consequence Matrix

For each activity, we can assess:

Dimension	Question
Skill atrophy	Does delegation cause you to lose the ability to do this?
Identity erosion	Does delegation change who you are?
Relationship impact	Does delegation affect your relationships?
Meaning loss	Does delegation remove meaning from your life?
Efficiency gain	How much time/effort does delegation save?
Quality change	Does AI do it better, worse, or differently?

3. The Agency Preservation Framework

We propose three modes of human-AI collaboration in the blurry zone:

1. AI-Assisted — Human does the activity with AI support (GPS navigation, Grammarly)
2. AI-Supervised — AI does the activity with human oversight (autonomous driving Level 3)
3. AI-Executed — AI does the activity independently; human reviews output (email auto-replies)

Each blurry-zone activity can be mapped to the most appropriate mode, which may vary by context.

4. The Delegation Paradox

Some activities become more meaningful when they could be delegated but aren't:

• A handwritten letter in the age of email
• A home-cooked meal when delivery is instant
• Walking somewhere when you could drive
• Memorizing a poem when you could Google it

Hypothesis: As delegation becomes ubiquitous, the non-delegated becomes a signal of care, effort, and authenticity. Delegation creates a new economy of meaning.

5. The Creeping Delegation Effect

Activities move from non-delegable → blurry → delegable gradually:

1. GPS navigation (once a skill, now universally delegated)
2. Spelling/grammar (once internalized, now outsourced)
3. Phone numbers (once memorized, now in contacts)
4. Mental arithmetic (once essential, now calculated)

Research question: What is currently in the blurry zone that will be fully delegated in 10 years? What are the consequences?

---

Proposed Study: Large-Scale Survey

Research Questions

RQ1: What is the empirical distribution of delegation preferences across the full spectrum of human activities?

RQ2: How do delegation preferences vary by age, gender, culture, education, technology adoption, and personality?

RQ3: What predicts whether an activity falls in the blurry zone for a given individual?

RQ4: How do people reason about the consequences of delegation (skill atrophy, meaning loss, identity change)?

RQ5: Is there a "delegation comfort frontier" — a threshold beyond which people experience discomfort, and what determines its location?

Methodology

Phase 1: Taxonomy Validation (N=50-100, Qualitative)

Goal: Validate and refine the HADAC taxonomy through expert interviews and focus groups.

Method:

• Semi-structured interviews with diverse participants (age, culture, profession)
• Card-sorting exercise: participants sort activity cards into delegable / non-delegable / blurry
• Think-aloud protocol: participants explain their reasoning
• Expert panel review (HCI, philosophy, AI ethics, anthropology)

Output: Refined taxonomy with activities that resonate across demographics.

Phase 2: Scale Development (N=200-500, Psychometric)

Goal: Develop the Delegation Comfort Scale (DCS) — a validated psychometric instrument.

Measures per activity:

1. Delegation willingness (0-10): "How willing would you be to fully delegate this activity to AI?"
2. Delegation anxiety (0-10): "How anxious would you feel if AI performed this activity for you?"
3. Delegation experience (binary): "Have you already delegated this (or part of it) to AI/technology?"
4. Meaning attribution (0-10): "How much personal meaning does this activity hold for you?"
5. Skill importance (0-10): "How important is it that you can do this yourself?"

Additional measures:

• Technology Readiness Index (TRI; Parasuraman & Colby, 2015)
• Big Five personality inventory
• Locus of control scale
• Cultural values (Hofstede dimensions)
• AI experience and trust measures
• Demographics (age, gender, education, income, country, religion)

Analysis: Exploratory Factor Analysis (EFA) to identify latent factors in delegation preferences. Item Response Theory (IRT) to refine scale.

Phase 3: Large-Scale Survey (N=1,000-10,000)

Goal: Map the global delegation preference landscape.

Sampling strategy:

• Stratified by country (minimum 10 countries across continents)
• Stratified by age (18-25, 26-35, 36-50, 51-65, 65+)
• Oversampling of high-AI-exposure populations (tech workers) and low-AI-exposure populations (rural, elderly)
• Platform: Prolific, MTurk, or local panel providers for non-English-speaking countries

Survey structure:

1. Demographic block (5 min)
2. Core HADAC items — 50-80 activities from the full taxonomy, each rated on DCS (15-20 min)
3. Blurry-zone deep dive — 10-15 blurry activities with sub-task decomposition questions (10 min)
4. Consequence reasoning — open-ended questions about what they fear losing/gaining (5 min)
5. Scenario vignettes — "Would you delegate X in context Y?" (5 min)

Total time: ~40 min

Power analysis: With N=5,000, we can detect small effect sizes (d=0.1) for demographic group comparisons with 80% power.

Phase 4: Analysis & Publication

Planned analyses:

1. Descriptive mapping: Mean DCS scores for all activities, visualized as a heatmap
2. Cluster analysis: Identify "delegation profiles" — types of people based on their delegation patterns
3. Regression models: Predictors of delegation comfort (demographics, personality, AI experience, cultural values)
4. Cross-cultural comparison: Country-level variation in delegation boundaries
5. Factor structure: What latent dimensions underlie delegation preferences? (e.g., embodiment, identity, control, trust)
6. Blurry-zone topology: Mapping the internal structure of the blurry zone using Types A-D
7. Temporal analysis: If longitudinal follow-up is funded, track shifting boundaries over 1-3 years

Target Venues

Primary: Nature Human Behaviour — the intersection of AI and human behavior is core to this journal's scope. The large-scale empirical mapping of delegation preferences is novel and policy-relevant.

Secondary options:

• PNAS — broad impact, quantitative social science
• Science Advances — interdisciplinary, open access
• CHI (ACM Conference on Human Factors in Computing Systems) — for the design implications
• CSCW — for the collaborative and social dimensions
• Trends in Cognitive Sciences — for the cognitive/philosophical framing

Estimated Timeline

Phase	Duration	Key Output
1. Taxonomy Validation	2-3 months	Refined HADAC taxonomy
2. Scale Development	3-4 months	Validated DCS instrument
3. Large-Scale Survey	3-4 months	Raw dataset (N=1,000-10,000)
4. Analysis & Writing	3-4 months	Manuscript draft
5. Submission & Review	3-6 months	Publication
Total	14-21 months

Estimated Budget

Item	Cost (USD)
Phase 1: Interviews (N=100 × $30/hr)	$3,000
Phase 2: Pilot survey (N=500 × $8)	$4,000
Phase 3: Main survey (N=5,000 × $8)	$40,000
Phase 3: Translation & localization (10 languages)	$10,000
Open access publication fees	$5,000
Research assistant (12 months, part-time)	$15,000
Total	~$77,000

---

Speculative Futures: Science Fiction as Method

Drawing on speculative design and design fiction methodologies (Dunne & Raby, 2013; Sterling, 2005), we propose using science fiction scenarios as survey stimuli to probe delegation boundaries:

Scenario 1: "The Perfect Day"

In 2035, your AI assistant has learned your preferences so well that it plans your entire day — meals, social interactions, entertainment, exercise — and you simply follow along. Every day is optimized for your wellbeing. You are measurably happier. But you didn't choose any of it.

Survey question: On a scale of 0-10, how appealing is this scenario? What would you change?

Scenario 2: "The Grief Bot"

Your mother has died. An AI trained on her messages, voice recordings, and photos offers to maintain a conversational simulation of her. Your siblings have opted in and find it comforting. You can talk to "her" whenever you want.

Survey question: Would you use this? What makes you hesitate or not?

Scenario 3: "The Friendship Agent"

Your AI assistant notices you've been socially isolated. It begins reaching out to potential friends on your behalf, arranging meetups, and maintaining small talk. You show up to these pre-arranged social encounters. The friendships become real.

Survey question: Is this delegation of friendship formation acceptable? Where does it cross a line?

Scenario 4: "The Moral Advisor"

You face a genuine ethical dilemma at work. Your AI assistant, having analyzed thousands of similar cases, provides a well-reasoned recommendation. You agree with its reasoning. You follow its advice.

Survey question: Did you make an ethical decision, or did the AI? Does it matter?

Scenario 5: "The Creative Partner"

You write a novel with AI. You provide the themes, characters, and emotional arc. The AI writes every sentence. The book is published under your name and becomes a bestseller. Critics call it "deeply human."

Survey question: Is this your novel? What did you contribute?

---

Theoretical Framework

HADAC draws on and synthesizes multiple theoretical traditions:

From Philosophy

• Hannah Arendt — Labor/Work/Action distinction (The Human Condition, 1958)
• Hubert Dreyfus — Embodied expertise and the limits of AI (What Computers Can't Do, 1972)
• Albert Borgmann — Focal practices vs. device paradigm (Technology and the Character of Contemporary Life, 1984)
• Edmund Husserl — Lifeworld and intentionality (via Konstantelos, 2025)

From HCI & Computer Science

• Parasuraman, Sheridan & Wickens — Levels of Automation (2000)
• Shneiderman — Human-Centered AI and levels of autonomy (2022)
• SAE J3016 — Levels of Driving Automation (2014/2021)
• Tomašev et al. — Intelligent AI Delegation framework (2026)

From Psychology

• Self-Determination Theory (Deci & Ryan) — Autonomy, competence, relatedness as basic needs
• Flow Theory (Csikszentmihalyi) — Optimal challenge requires appropriate difficulty
• Locus of Control (Rotter) — Internal vs. external attribution

From Sociology & STS

• Braverman — Deskilling through automation (1974)
• Ada Lovelace Institute — Dilemmas of delegation for policy (2025)
• Crawford & Joler — Hidden labor in AI systems (2018)

From Cognitive Science

• Embodied Cognition (Chemero, 2023; Clark & Chalmers, 1998) — Extended mind thesis
• Griffiths (2020) — Human intelligence defined by human limitations

Novel Contribution

HADAC contributes a unified framework that spans the entirety of human life — not just work tasks or specific domains — and grounds it in empirical measurement of preferences rather than purely philosophical or technical analysis.

---

References

1. Arendt, H. (1958). The Human Condition. University of Chicago Press. DOI: 10.7208/chicago/9780226924571.001.0001
2. Borgmann, A. (1984). Technology and the Character of Contemporary Life. University of Chicago Press. DOI: 10.7208/chicago/9780226066295.001.0001
3. Braverman, H. (1974). Labor and Monopoly Capital. Monthly Review Press.
4. Chemero, A. (2023). LLMs differ from human cognition because they are not embodied. Nature Human Behaviour, 7, 1828-1829. DOI: 10.1038/s41562-023-01723-5
5. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7-19. DOI: 10.1093/analys/58.1.7
6. Crawford, K., & Joler, V. (2018). Anatomy of an AI System. URL: https://anatomyof.ai
7. Csikszentmihalyi, M. (1990). Flow: The Psychology of Optimal Experience. Harper & Row.
8. Deci, E. L., & Ryan, R. M. (2000). The "what" and "why" of goal pursuits. Psychological Inquiry, 11(4), 227-268. DOI: 10.1207/S15327965PLI1104_01
9. Dreyfus, H. L. (1972). What Computers Can't Do. MIT Press.
10. Dunne, A., & Raby, F. (2013). Speculative Everything. MIT Press.
11. Farmer, H. et al. (2025). The Dilemmas of Delegation. Ada Lovelace Institute. URL: https://www.adalovelaceinstitute.org/report/dilemmas-of-delegation/
12. Griffiths, T. L. (2020). Understanding human intelligence through human limitations. Trends in Cognitive Sciences, 24(11), 873-883. DOI: 10.1016/j.tics.2020.09.001
13. Konstantelos, E. (2025). Consciousness, Autonomy, and the Irreducibly Human. URL: https://philarchive.org/archive/KONCAA-2
14. Parasuraman, R., & Colby, C. L. (2015). An updated and streamlined Technology Readiness Index: TRI 2.0. Journal of Service Research, 18(1), 59-74. DOI: 10.1177/1094670514539730
15. Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). A model for types and levels of human interaction with automation. IEEE Transactions on Systems, Man, and Cybernetics—Part A, 30(3), 286-297. DOI: 10.1109/3468.844354
16. Shneiderman, B. (2022). Human-Centered AI. Oxford University Press. DOI: 10.1093/oso/9780192845290.001.0001
17. Sterling, B. (2005). Shaping Things. MIT Press.
18. Tomašev, N., Franklin, M., & Osindero, S. (2026). Intelligent AI Delegation. arXiv: 2602.11865. URL: https://arxiv.org/abs/2602.11865

---

Contributing

This is an open research project. We welcome contributions in the form of:

• New activities to add to the taxonomy
• Cultural perspectives on delegation boundaries
• Survey instrument feedback
• Theoretical connections we've missed
• Translations of survey materials

License

Creative Commons Attribution 4.0 International (CC BY 4.0)

Citation

@misc{hadac2026,
  title={HADAC: Human Activity Delegation Agency Chart},
  author={Hu, Botao Amber},
  year={2026},
  url={https://github.com/realitydeslab/human-delegation-agency-chart}
}

---

"The question is not whether machines can do what we do, but whether what we do still means what it meant when only we could do it."