← Back to all writing

AI futures evidence — U7: meaning institutions

July 5, 2026

Futures index · 中文 · Main forecast

Each section: Claim · Why · Evidence · Analogue · Would update if · Conf (H/M/L).


Parent timeline: Utopia timeline §U7
Master synthesis: my putopia — meaning crux row
Date: 2026-07-04
Settings: Hybrid time (C); @C8–C10 primary window; Path family E (aligned-but-bored)
Tagging: [EST] = psychology/wellbeing empirical; [SPEC] = institutional design / philosophy


0. Node definition & U2 gate

Question: Conditional on alignment success and material abundance (U2 physical path open), what is P(institutions sustain eudaimonia when AI makes human labor non-instrumental)?

Distinction (Bostrom):

  • Post-work / shallow redundancy: AI does tasks better; humans can still play chess, craft, care, govern — recreational or dignity-preserving niches remain.
  • Post-instrumental / deep redundancy: AI does tasks better and understands/appreciates them better; human effort is redundant for practical and evaluative purposes.

U2 normative gate: U2 = symbiotic flourishing = aligned superintelligence plus humans retaining meaningful agency + eudaimonia. U7 is the normative bottleneck: U2 physical (abundance, safety, augmentation) can succeed while U7 fails → Path family E: aligned-but-bored (whimper channel; blocks U2/U3 narratively).

Not this node’s job: Distribution (U2 economic node), alignment (U1), science dividend (U3). U7 assumes those can succeed and asks whether life is worth living at scale.

Repo anchors: (internal note); (internal note); (internal note) §D; (internal note); (internal note); (internal note).


1. Master estimates (synthesis — fill first, refine last)

QuantityPoint est.RangeConf
P(meaning collapse crux) — post-instrumentality destroys civilizational eudaimonia even if aligned & abundant0.580.48–0.68M
P(meaning institutions scale @C8) — post-work, partial non-instrumentality0.380.28–0.48M
P(meaning institutions scale @C10) — deep post-instrumentality0.160.10–0.24L
P(Path E: aligned-but-bored is modal among aligned worlds @2050)0.420.32–0.52M
P(wireheading / hedonic default wins over institution-building)0.350.25–0.45M

Meaning-collapse crux (load-bearing): If P ≈ 0.58 that meaning institutions fail at scale post-instrumentality, then P(U2 symbiosis) ≤ P(aligned) × P(distribution) × (1 − 0.58) × … — U2 headline numbers are capped regardless of alignment tail success.

Would update master crux ↓ (institutions more likely) if: ≥2 OECD-scale pilots (N>50k, ≥5yr) show sustained eudaimonia scores + civic participation without employment, with pre-registered outcomes; or binding “human role preservation” norms in major AI deployment treaties with compliance evidence.

Would update master crux ↑ if: OpenResearch-scale UBI + mental-health null results replicate at 2× income; FIRE return-to-work rate >50%; post-C8 suicide/anomie indicators spike in high-AI-adoption regions with controlled covariates.


Evidence sections

P = 0.72 — Post-work latent-function deprivation is real [EST]

Claim: Unemployment/retirement without structured substitutes predicts psychological distress independently of income in ~72% of comparable studies (directional consensus; magnitude varies).

Why: Jahoda (1982) five latent functions — time structure, social contact, collective purpose, status, activity — plus manifest income. Meta-analysis (Paul & Batinic; 113 studies): latent + manifest functions explain ~19% of mental-health variance; deprivation correlates with distress more strongly than financial difficulty alone in many samples.

Evidence:

Analogue: Marienthal unemployed (1930s Austria) — collapse of daily rhythm and social role, not only wallet.

Would update if: Large panel shows income replacement alone closes ≥80% of unemployment mental-health gap with no structural interventions.

Conf: H


P = 0.55 — UBI satisfies manifest but not latent functions [EST]

Claim: In modal UBI RCT designs, ~55% probability that cash transfers address income without closing Jahoda/SDT gaps (autonomy/competence/relatedness via non-work channels).

Why: Finland: health gains track income more than condition removal; OpenResearch 2024: leisure↑, family care↑, but small/non-sustained mental-health gains, labor hours↓; Alaska PFD: no participation collapse but too small for post-work test. SDT: work is primary adult vehicle for all three needs (values_psychology_map.md §3).

Evidence:

Analogue: Lottery winners — life satisfaction↑, day-to-day happiness flat (Lindqvist et al., RESTUD 2020).

Would update if: Long-duration UBI (≥$30k/yr equiv, ≥8yr) shows latent-function parity vs employment on validated scales.

Conf: M–H


P = 0.48 — SDT needs can be met without wage labor [EST + SPEC]

Claim: ~48% that alternative structures (civic roles, craft guilds, care networks, education-as-end) can satisfy autonomy, competence, relatedness at population scale — plausible for subsets, uncertain for majority.

Why: Retirement intervention reviews: explicit social roles improve satisfaction when well-designed; many retirees thrive with strong networks. But FIRE 30–40% return-to-work; Soviet guaranteed employment → alienation without autonomy. Satisfying SDT without employment requires designed institutions, not default.

Evidence:

Analogue: Post-Olympic athlete career transition programs — some scale, many fail without infrastructure.

Would update if: National “participation income” program shows SDT need satisfaction ≥ employed controls.

Conf: M


P = 0.65 — Weber work ethic remains culturally load-bearing in OECD [EST]

Claim: ~65% that Protestant/work-as-moral-worth ethic still shapes shame, status, and policy preferences in Western industrial societies through 2035 — slowing acceptance of non-instrumental life.

Why: Unemployment stigma persists beyond financial need; FIRE identity crises when career = self; political resistance to “handouts” even when empirically benign. Inglehart shift toward self-expression is real but incomplete in WVS data; tight cultures (Gelfand) resist norm change faster.

Evidence:

  • Weber, Protestant Ethic (1905); thread5_human_meaning_ubi.md §3.1
  • (internal note) §5 (Inglehart, Gelfand)

Analogue: Keynes 1930 prediction of 15-hour week — failed for 90+ years despite productivity gains.

Would update if: Cross-national survey shows work-centrality scores fall >30% in under-40 cohort by 2032 with stable test-retest validity.

Conf: M


P = 0.70 — Bostrom shallow redundancy is survivable; deep redundancy is the crux [SPEC]

Claim: ~70% that recreational niches (chess, sport, craft) suffice under shallow redundancy only; under deep redundancy the meaning problem is qualitatively harder (Bostrom’s central split).

Why: NDPR, Scott Alexander, Calum Chace reviews converge: post-work is “relatively superficial”; post-instrumental forces reframing of achievement, learning, even appreciation. Metz objective-meaning criteria (improvement arc, originality, helping others) may have no non-cheat human niche if AI dominates all three.

Evidence:

Analogue: MDMA-level bliss vs “earned” meaning — Bostrom’s wireheading floor vs objectivist demand.

Would update if: Philosophical consensus settles that subjective meaning suffices for policy (pure subjectivism wins).

Conf: M


P = 0.35 — Wireheading / hedonic engineering becomes default path [SPEC]

Claim: ~35% that civilizations solving alignment choose large-scale safe hedonic tech (pharmacological, neurotech, experiential engineering) as dominant coping mechanism vs institution-heavy eudaimonia.

Why: Bostrom treats safe wireheading as floor — worst-case still “not bad.” Economic and political path of least resistance: individual choice, medical frame, no collective purpose required. Scott Alexander: “wireheaded-meaning” trivially satisfiable if you abandon objectivism. Tension with Sen capabilities (functionings, not just feelings).

Evidence:

  • Bostrom Deep Utopia; NDPR on neurotech bliss chapters
  • my putopia — Sen primary normative anchor

Analogue: Opioid + screen leisure in deindustrialized counties — hedonic patch without institutions.

Would update if: Major jurisdiction bans non-therapeutic hedonic enhancement post-AGI and compliance holds.

Conf: L–M


P = 0.62 — Chess/sports analogy works for shallow redundancy only [EST]

Claim: ~62% that human-vs-human sport/craft retains audience and participation when AI is superhuman but non-intentional — but analogy fails for deep redundancy and for non-elite majority.

Why: Chess.com 51M→190M users (2020–2024); WCF 2024 largest ever; AlphaGo increased novel human moves. QZ/Forbes: demand for human constraint and narrative of struggle. Limits: (1) elite tiny fraction get status; (2) engines reduced creativity anxiety for median players; (3) physical sport still needs human bodies — cognitive domains differ; (4) if AI is better judge of quality, artisanal premium may collapse.

Evidence:

Analogue: Marathon running despite cars; live music despite perfect recordings.

Would update if: AI-authored art/sport viewership exceeds human-only in >50% of categories for 5+ years with stable demographics.

Conf: M


P = 0.44 — Handicraft / artisanal premium persists post-AI [EST]

Claim: ~44% that “human-made” status premium survives mass AI craft quality — Veblen/conspicuous production partially sustains niche markets, not general livelihood.

Why: Handmade pottery, analog photography, vinyl — premium exists but small share of economy. AI art protests (Christie’s 2024, 6000+ artists) show process value for some. Scientific American: greater AI art knowledge → lower ethical acceptance — may strengthen human-only niches among educated consumers. Scale: most cannot live on artisanal status.

Evidence:

Analogue: Organic food premium — real but ~5–10% market, not labor-absorption solution.

Would update if: “Human-only” certification becomes >20% GDP with stable employment share.

Conf: M


P = 0.58 — Status is zero-sum; post-work status crisis [EST + SPEC]

Claim: ~58% that relative status competition intensifies when meritocratic work ranking disappears — blocking widespread eudaimonia even with abundance.

Why: Jahoda status function; UBI_post_work_meaning_AI.md §4 inference 3 — fewer “recognized artist” slots than “respected electrician” slots. Veblen leisure class: conspicuous consumption fills void. Social media already amplifies relative comparison. Institutional fix requires new scarce honors or norm shift toward non-comparative flourishing — both hard.

Evidence:

  • Jahoda; Veblen Theory of the Leisure Class
  • (internal note) §4

Analogue: Tenure-track scarcity in academia; influencer economy winner-take-most.

Would update if: Post-work societies show Gini on social honor metrics <0.35 with validated scales.

Conf: M


P = 0.52 — Durkheim anomie risk in compressed transition [EST]

Claim: ~52% that AI-labor transition outpaces norm regeneration → anomie (normlessness, aimlessness) at societal scale without proactive institution-building.

Why: Durkheim: rapid division-of-labor change without new solidarity produces suicide, crime, norm confusion. Retirement scoping review (2025): work loss = role, identity, purpose loss. WEF AI precariat framing. Counter: COVID showed mutual aid can spike — but temporary.

Evidence:

Analogue: Rust Belt deindustrialization → Case-Deaton deaths of despair (partial; coal study nuance 2025).

Would update if: High-AI regions show no excess mortality vs matched controls through 2035.

Conf: M


P = 0.38 — Putnam-style social capital rebuilds fast enough [EST]

Claim: ~38% that civic/community infrastructure can net recover from Bowling Alone decline in time for C8 shock.

Why: Putnam (2000): declining unions, churches, associations — exactly Jahoda’s social-contact substrate. Some recovery via online communities (weak ties). Hester-Srnicek After Work: post-work requires collective reproduction redesign, not just wage replacement. Historical rebuild (red Vienna, commune kitchens) existed but politically exceptional.

Evidence:

  • Putnam Bowling Alone; utopia_study/extensions/26_hester_srnicek_after_work.md
  • happy_path_with_ai_brainstorm.md §D — community rebuilding agenda

Analogue: Post-WWII civic boom — required war solidarity + policy, not tech alone.

Would update if: US/EU association density rises 5yr running while AI deployment accelerates.

Conf: M


P = 0.45 — Retirement / FIRE as partial preview supports mixed outcome [EST]

Claim: ~45% that voluntary post-work previews (retirement, FIRE) are calibrated predictors: bimodal thrive/struggle, ~30–40% return-to-work, SES and work-centrality moderate.

Why: Higher SES + strong networks → better outcomes. Achievement-oriented FIRE → purpose vacuum. Business Insider 2026: boredom/misery cases. Disability/basic-income: security enables contribution when support structures exist — doesn’t eliminate desire to contribute but redefines it.

Evidence:

Analogue: Mandatory vs voluntary retirement — AI displacement is often involuntary (worse).

Would update if: Involuntary AI displacement cohorts match voluntary FIRE wellbeing when income matched (natural experiment).

Conf: M–H


P = 0.28 — Job Guarantee / participation income scales meaningfully [SPEC]

Claim: ~28% that JG or Atkinson participation income at national scale provides Jahoda functions without bullshit-job alienation.

Why: Tcherneva JG addresses latent functions directly — but Graeber risk: make-work reads as BS; Soviet full employment → alienation without autonomy (SDT). Participation income admin complexity + boundary disputes. No rich-country AI-era JG trial at scale.

Evidence:

Analogue: WPA 1930s — meaningful for some, “leaf-raking” critique for others.

Would update if: EU or US federal JG pilot N>100k for 3yr with eudaimonia + employment spillover data.

Conf: L–M


P = 0.32 — “Meaning infrastructure” gets serious policy funding pre-C8 [SPEC]

Claim: ~32% that OECD governments fund community/civic/meaning infrastructure at ≥0.3% GDP before public AGI (C8), proactively not reactively.

Why: UBI_post_work_meaning_AI.md §6: post-work needs (a) distribution (b) power dispersion (c) meaning infrastructure — (c) most neglected. WEF “unemployment rescue labs” rhetorical. Modal politics: UBI debate yes; community rebuild no. Scott blueprint trap: grand schemes fail; piecemeal underfunded.

Evidence:

  • (internal note) crux Scott/Ostrom
  • happy_path_with_ai_brainstorm.md §D — “nobody knows how to build meaning infrastructure”

Analogue: US mental-health parity laws — decades lag, uneven enforcement.

Would update if: G7 communique binds meaning/civic funding line item with appropriations.

Conf: M


P = 0.40 — Illich convivial AI is politically viable at scale [SPEC]

Claim: ~40% that convivial tool design (transparent, leavable, skill-preserving) wins against industrial AI stack at platform scale.

Why: Illich: convivial tools expand user narrative + repair rights; industrial tools → radical monopoly. Applied to AI: local models, inspectability, user fine-tune, mandatory exit (09_illich_tools_for_conviviality.md §10). Counter: UX convenience + network effects favor API black boxes; “without Copilot unprofessional” norm emerging.

Evidence:

  • (internal note)
  • hci_interaction_design_evidence.md §8.3 — desirable difficulty — [EST] design principles exist, adoption weak

Analogue: Right-to-repair wins in EU vs US lag on software.

Would update if: Major lab ships default-on “human judgment mode” with regulatory mandate and >40% sustained use.

Conf: M


P = 0.55 — Vallor moral deskilling accelerates under modal AI deployment [EST + SPEC]

Claim: ~55% that modal product choices (AI generates, human reviews) produce net moral deskilling by 2035 in knowledge work cohorts.

Why: Vallor: outsourcing judgment erodes phronesis; Russell alignment insufficient if humans can’t correct goals. Lee et al. 2026 Scientific Reports: AI reliance ↓ self-efficacy. METR 2025: passive review patterns. Good automation requires paideutic design — not modal.

Evidence:

Analogue: GPS navigation skill decay; Casner aviation manual-flying retention.

Would update if: Longitudinal study shows AI-augmented workers gain deliberative scores vs controls.

Conf: M


P = 0.48 — “Deliberately leave space for humans” can be operationalized [SPEC]

Claim: ~48% that design principles (human judges / AI implements, desirable difficulty, ZPD fading scaffolding) become enforceable product/regulatory standards, not blog posts.

Why: HCI evidence strong (hci_interaction_design_evidence.md §8.3): 7 principles listed. Shopify internal augmentation framing. Gap: no law defines augment/replace boundary; labor law silent on cognitive deskilling. EU AI Act risk-based — not meaning-preservation based.

Evidence:

  • research_ai_economy_article/hci_interaction_design_evidence.md §8.3–8.4 — [EST] evidence, [SPEC] policy
  • happy_path_with_ai_brainstorm.md §D — human-exclusive domains question

Analogue: FAA minimum hand-flying hours — rare successful “preserve skill” regulation.

Would update if: EU/US standard mandates human-originated judgment artifacts for listed decisions (medical, legal, political).

Conf: M


P = 0.25 — Human-exclusive domain reservation (medicine, law, art, politics) holds [SPEC]

Claim: ~25% that formal “humans only” rules survive market + efficiency pressure through C10.

Why: Luddite analogy fails — regulators slow, jurisdictions arbitrage. AI legal research already dominant; radiology AI assist; political deepfake era cuts trust in human-only claims. Arendt: preserving action (politics) essential — but no enforcement mechanism. Human-exclusive may become luxury goods for rich jurisdictions only.

Evidence:

  • Arendt Human Condition; thread5_human_meaning_ubi.md §3.2
  • happy_path_with_ai_brainstorm.md §D

Analogue: Manual driving bans proposed in EU cities — mixed.

Would update if: UN declaration + treaty verification for human-in-loop categories with sanctions.

Conf: L


P = 0.50 — Democratic legitimacy blocks pure AI governance [EST + SPEC]

Claim: ~50% that populations reject binding AI political decision-making even when AI outperforms — preserving human self-governance as intrinsic good.

Why: happy_path_with_ai_brainstorm.md §E — efficiency vs legitimacy. Moral Machine: thin universals, thick variation on authority/fairness. AI Mirror: Narcissus risk of mistaking reflection for judgment. Counter: crisis delegation (automation bias), Singapore/China efficiency frames, young cohort AI trust higher.

Evidence:

  • happy_path_with_ai_brainstorm.md §E
  • Vallor AI Mirror (2024) — 36_vallor_technology_and_virtues.md §10

Analogue: Brexit/Populism — preference for “voice” over expert optimization.

Would update if: Stable democracy delegates >50% legislative drafting to AI with approval >60% for 10yr.

Conf: M


P = 0.35 — Russell/Vallor synthesis (“good automation”) becomes industry norm [SPEC]

Claim: ~35% that combined Russell deferral + Vallor paideutic design becomes modal among frontier labs pre-C9.

Why: Russell CIRL/deference — formal preference uncertainty. Vallor — preserve moral practice windows. Labs optimize engagement/revenue; RLHF rewards sycophancy. Some alignment teams internalize — not product default.

Evidence:

  • 11_russell_human_compatible.md; 36_vallor_technology_and_virtues.md §8
  • utopia_study/extensions/00_triangles_synthesis.md — Leverage layer

Analogue: Safety case in aviation — mandated after disasters, not voluntary.

Would update if: Anthropic/OpenAI/Google ship user-facing “competence preservation” modes with third-party eval.

Conf: M


P = 0.60 — Arendt “society of laborers without labor” is default path [SPEC]

Claim: ~60% that automation eliminates labor without ascending to Arendtian work (durables) and action (politics) — passive consumption default.

Why: Arendt warning: labor automated but culture still organized around labor-consumption cycle. Veblen + screen leisure + algorithmic feed. Gorz autonomous work vs heteronomous — most freed time may not become poiesis. Requires normative project (Srnicek destroy work ethic) — politically toxic modal path.

Evidence:

Analogue: Post-WWII TV era — hours filled, not agora revived.

Would update if: Civic participation hours double in under-40 cohort in high-AI countries.

Conf: M


P = 0.42 — Graeber “bullshit jobs” removal is net positive for meaning [EST]

Claim: ~42% that eliminating genuinely meaningless jobs improves population meaning if substitutes exist; net negative if substitutes absent.

Why: Graeber 20–50% BS jobs claim disputed empirically (more BS feeling in knowledge work, less in manual). AI paradox: may delete BS jobs or accelerate BS via AI slop. Removal helps only with Jahoda substitutes — ambiguous net.

Evidence:

  • Graeber Bullshit Jobs; WES 2021 critique[EST]
  • thread5_human_meaning_ubi.md §3.3

Analogue: Early retirement from hated job — often relief then vacuum.

Would update if: Panel data shows AI-displaced from low-meaning jobs report higher eudaimonia vs matched voluntary quitters.

Conf: M


P = 0.33 — Historical precedent: institutional lag >> tech lag [EST]

Claim: ~33% that meaning institutions reach adequacy within one generation of C8 — based on historical transition durations.

Why: Agricultural→industrial ~150yr with suffering; 8-hour day ~century; Keynes failed on hours. Meaning infrastructure (unions, schools, churches) co-evolved slowly. AI transition compressed to ~10–20yr — institutions likely late.

Evidence:

Analogue: Climate adaptation lag vs emissions speed.

Would update if: Parallel civic rebuild completes in <10yr in a mid-size OECD country post-shock.

Conf: M–H


P = 0.50 — Open-source / maker / commons communities scale as meaning venues [EST + SPEC]

Claim: ~50% that OSS, makerspaces, mutual aid, and hobbyist communities provide template for meaning institutions — but ≤15% population penetration at quality threshold.

Why: Linux, Wikipedia, local makerspaces — proof of collective purpose without wage. Ostrom commons design principles apply. Limits: self-selected, educated, male-skewed; don’t absorb median displaced worker without massive onboarding investment.

Evidence:

  • 03_ostrom_governing_the_commons.md; happy_path_with_ai_brainstorm.md §D
  • thread5_human_meaning_ubi.md §7

Analogue: Wikipedia editor community ~100k active vs billions of users.

Would update if: National “civic OS” programs enroll >20% adults in sustained projects.

Conf: M


P = 0.22 — Sen capabilities framework successfully guides post-instrumental policy [SPEC]

Claim: ~22% that governments adopt Sen-style capability metrics (not GDP/UBI alone) as primary post-AI welfare measure with budget linkage.

Why: Sen Development as Freedom is utopia canon normative anchor (my putopia). Real policy: GDP, employment rate, inflation dominate. Capability lists contested cross-culturally (WEIRD). Still: best normative target for U7 success definition.

Evidence:

Analogue: Bhutan GNH — niche, inspirational, limited scale.

Would update if: OECD capability dashboard drives AI dividend allocation in ≥3 states.

Conf: L–M


P = 0.47 — Gendered / care-work dimension blocks naive post-work [EST + SPEC]

Claim: ~47% that without Hester-Srnicek 3R (recognition, redistribution, reduction of reproductive labor), post-work dumps care burden on women/immigrants → meaning crisis for some, exploitation for others.

Why: After Work (2023): wage automation without collective childcare/eldercare redesign replicates Cowan paradox — appliances didn’t free time. Platform care commodifies. AI home promises repeat 1950s white-goods narrative.

Evidence:

  • utopia_study/extensions/26_hester_srnicek_after_work.md
  • UBI_post_work_meaning_AI.md — post-work three-legged stool

Analogue: Dual-income trap — more income, same domestic load.

Would update if: AI-era national care-public spending rises >1pp GDP/yr for 5yr in ≥4 G7.

Conf: M


P = 0.38 — Path family E (aligned-but-bored) is ≥40% of aligned worlds @2050 [SPEC]

Claim: ~38% point (see master 0.42 modal) that conditional on technical alignment success, normative failure (boredom, anomie, wireheading, status despair) prevents calling outcome U2/U3 — physical utopia, spiritual/modal anti-utopia.

Why: Conjunction: P(aligned) may rise while P(meaning institutions) stays low. Whimper ~24% in doom stitch includes agency loss — overlaps E. Bostrom unresolved ending (Dean terminates lecture) mirrors epistemic humility — no institutional proof.

Evidence:

Analogue: Gulf petrostates — material abundance, heterogeneous meaning outcomes.

Would update if: Post-alignment society surveys show flourishing indices ≥ pre-AI baseline in >70% strata for 10yr.

Conf: M


P = 0.30 — Amodei “Machines of Loving Grace” meaning story is incomplete [SPEC]

Claim: ~30% that techno-optimist abundance narratives (Amodei 2024, Altman) include credible institutional plan for meaning — most stop at biology/abundance.

Why: Amodei focuses on cancer, biology, coordination — light on Jahoda/Vallor. Altman UBI/equity fund addresses manifest function. Positive vision necessary not sufficient for U7.

Evidence:

  • entity_master_list.md Amodei essay; Altman Moore’s Law for Everything
  • thread5_human_meaning_ubi.md §5.5 abundance vs displacement

Analogue: 1960s Jetsons — tech without sociology.

Would update if: Frontier lab publishes peer-reviewed “post-work meaning” roadmap with funded pilots.

Conf: M


P = 0.65 — AI art/creative displacement triggers meaning fight before labor displacement completes [EST]

Claim: ~65% that creative-class meaning crisis peaks earlier than full labor displacement — forcing policy debate on human-only culture before C8.

Why: Artists organized (Christie’s protest); writers’ surveys; “process vs product” value fight is sharp edge of post-instrumentality. May seed institutions (certification, human-only zones) or burn out as defeatism.

Evidence:

Analogue: Napster-era music industry — partial institutional adaptation.

Would update if: Creative unions win binding human-performance quotas in EU/US by 2030.

Conf: M


P = 0.55 — Positive correlation: slow takeoff helps U7 [SPEC]

Claim: ~55% that slower Ci spine (more time @C4–C7) increases P(meaning institutions) — partial correlation, not independent (shared with U3 governance).

Why: Institution-building needs decades; Hanson slow variant buys time. Fast C8→C10 skips “shallow redundancy” adaptation period where chess/sports/civic experiments run. Do not multiply with U3 node P without correlation matrix.

Evidence:

Analogue: Post-war institution building 1945–1975 before neoliberal erosion.

Would update if: Hard takeoff + excellent meaning outcomes in sim or real small economy.

Conf: M


P = 0.70 — Modal c8 society is Ghost GDP / agency loss, not eudaimonic U2 [EST + SPEC]

Claim: ~70% that modal non-doom baseline (c8 society snapshot by ci) already fails U7 normative bar before full post-instrumentality — meaning collapse begins at partial automation.

Why: Ghost GDP: output up, worker agency down. Acemoglu so-so automation. Precarity + BS work + algorithmic management erode Jahoda functions while retaining work ethic stress. U7 failure may be continuous, not binary at C10.

Evidence:

Analogue: Present-day knowledge worker burnout — pre-AGI preview.

Would update if: OECD wellbeing indices rise monotonically with AI exposure 2024–2035.

Conf: M–H


P = 0.38 — @C8 conditional: meaning institutions scale if shallow redundancy only [SPEC]

Claim: At C8 (public AGI, remote workers, widespread post-work but not full post-instrumental), P ≈ 0.38 that institutions sustain population-scale eudaimonia (master table).

Why: Product of: latent-function substitutes partially buildable (0.48 SDT); chess analogy works (0.62 shallow); civic funding weak (0.32); anomie risk (0.52); moral deskilling (0.55). Rough Bayes-style intuition → high-thirties.

Evidence: Sections above; Bostrom shallow vs deep split.

Analogue: Early retirement societies (Netherlands, Nordics) — good not perfect.

Would update if: C8-era longitudinal wellbeing matches Nordics in high-AI adopters.

Conf: M


P = 0.16 — @C10 conditional: meaning institutions survive deep post-instrumentality [SPEC]

Claim: At C10, P ≈ 0.16 that non-wireheading, objectivist-eudaimonia-compatible institutions scale (master table).

Why: Deep redundancy undercuts craft, science, care appreciation niches; wireheading floor attractive; objectivist meaning may be philosophically unmaintainable; human-exclusive domains fail (0.25). Tail: subjectivist + safe hedonic civ may be “OK” but not U2 by Sen/Rawls-capability standard.

Evidence: Bostrom Deep Utopia; Scott Alexander on objective vs wireheaded meaning.

Analogue: Egan Diaspora polises — meaning via choice of substrate; not proven for biological humans.

Would update if: Philosophical breakthrough + social tech makes Metz criteria satisfiable without cheating at scale.

Conf: L


2. Path family E — aligned-but-bored (detailed branch)

| Branch | Mechanism | P(branch | aligned @2050) | U bucket | |--------|-----------|----------------|----------| | E1 | Wireheading / hedonic default | 0.18 | Blocks U2 (capability erosion) | | E2 | Anomie + status despair + Veblen leisure | 0.14 | Blocks U2/U3 | | E3 | Material OK; moral deskilling → can’t govern AI | 0.10 | U2 physical only | | E4 | Niche flourishing (elites, creatives, civic core) + median boredom | 0.12 | U4 partial | | ~E total | Any normative failure mode | ~0.42 | Path E modal among aligned |

Key insight: Alignment solves control not culture. U7 is where Path E lives; doom stitch whimper channel overlaps but is not identical (whimper can be misaligned concentration).


3. Falsifiers (node-level)

  1. Population eudaimonia (WHO-5, Ryff, or pre-registered composite) ≥ pre-AI baseline in highest-AI-adoption decile for 10 consecutive years post-C8.
  2. Civic participation (volunteering, association membership, local governance attendance) ↑ ≥25% from 2024 baseline in same regions — not only online engagement.
  3. FIRE return-to-work rate falls <15% among voluntary post-work cohorts with strong social infrastructure — suggests institutions work.
  4. Human-only cultural/economic sectors stable ≥10% employment share without subsidy after AI superhuman in those domains.
  5. Binding treaty on paideutic AI design with third-party audit and documented skill/moral deliberation metrics — compliance verified.

4. Correlation warnings (do not double-count)

ClusterOverlaps with U7Note
U2 distributionIncome → manifest Jahoda functionFix distribution first; U7 conditional
Slow takeoff / U3Time for institutionsρ ≈ +0.4–0.6 speculated
Whimper / Ghost GDPAgency lossModal already partially U7-failing
U1 alignmentPath E assumes alignmentConditional probabilities
Vallor deskilling / HCISame mechanism as § moral deskillingOne cluster

Full matrix: pending correlation_matrix_positive.md.


5. Research gaps (2026-07-04)

  1. No RCT of meaning infrastructure bundle (UBI + civic roles + community centers) vs UBI alone at scale.
  2. No Ci-calibrated panel linking AI exposure → latent function deprivation → mortality.
  3. Bostrom Metz meaning theory operationalization for policy metrics — unsettled.
  4. Cross-cultural meaning institutions (non-WEIRD) under post-AI shock — thin data.
  5. Deep vs shallow redundancy empirical proxy — no agreed indicator pre-C9.

6. Section index (count)

#SectionP
1Jahoda latent deprivation real0.72
2UBI ≠ latent functions0.55
3SDT without wage labor0.48
4Weber work ethic load-bearing0.65
5Shallow vs deep redundancy0.70
6Wireheading default0.35
7Chess/sports analogy limits0.62
8Artisanal premium0.44
9Status zero-sum0.58
10Durkheim anomie0.52
11Putnam rebuild0.38
12FIRE/retirement preview0.45
13JG/participation income0.28
14Meaning infrastructure funding0.32
15Illich convivial AI viable0.40
16Vallor deskilling0.55
17Leave-space-for-humans operationalized0.48
18Human-exclusive domains hold0.25
19Democratic legitimacy vs AI efficiency0.50
20Russell/Vallor good automation norm0.35
21Arendt laborers without labor0.60
22Graeber BS jobs net0.42
23Institutional lag >> tech0.33
24Commons/maker meaning venues0.50
25Sen capabilities policy adoption0.22
26Gendered care / 3R0.47
27Path E aligned-but-bored0.38
28Amodei/Altman meaning gap0.30
29Creative displacement early fight0.65
30Slow takeoff helps U70.55
31Modal c8 fails U7 early0.70
32@C8 meaning institutions scale0.38
33@C10 deep post-instrumentality0.16

Evidence sections: 33 (+ master table + Path E detail)


Update log

DateChange
2026-07-04Initial research pass — web + canon + repo threads; 33 evidence sections

External sources (selected)