teleo-codex/inbox/archive/ai-alignment/2026-02-24-catalini-simple-economics-agi.md
m3taversal be8ff41bfe link: bidirectional source↔claim index — 414 claims + 252 sources connected
Wrote sourced_from: into 414 claim files pointing back to their origin source.
Backfilled claims_extracted: into 252 source files that were processed but
missing this field. Matching uses author+title overlap against claim source:
field, validated against 296 known-good pairs from existing claims_extracted.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-21 11:55:18 +01:00

41 lines
2.9 KiB
Markdown

---
type: source
title: "Some Simple Economics of AGI"
author: "Christian Catalini, Xiang Hui, Jane Wu"
url: https://arxiv.org/abs/2602.20946
date_published: 2026-02-24
date_archived: 2026-03-16
domain: ai-alignment
secondary_domains: [teleological-economics]
status: enrichment
processed_by: theseus
tags: [verification-bandwidth, economic-bottleneck, measurability-gap, hollow-economy]
sourced_via: "Alex Obadia (@ObadiaAlex) tweet, ARIA Research Scaling Trust programme"
twitter_id: "712705562191011841"
processed_by: theseus
processed_date: 2026-03-19
enrichments_applied: ["human verification bandwidth is the binding constraint on AGI economic impact not intelligence itself because the marginal cost of AI execution falls to zero while the capacity to validate audit and underwrite responsibility remains finite.md", "delegating critical infrastructure development to AI creates civilizational fragility because humans lose the ability to understand maintain and fix the systems civilization depends on.md"]
extraction_model: "anthropic/claude-sonnet-4.5"
claims_extracted:
- "human verification bandwidth is the binding constraint on AGI economic impact not intelligence itself because the marginal cost of AI execution falls to zero while the capacity to validate audit and underwrite responsibility remains finite"
---
# Some Simple Economics of AGI
Catalini et al. frame AGI economics around two competing cost curves. As AI decouples cognition from biology, the marginal cost of measurable execution falls to zero — but this creates a new bottleneck: human verification capacity.
Key framework:
- Verification bandwidth — the ability to validate, audit, and underwrite responsibility — is the binding constraint on AGI growth, not intelligence itself
- This generates a "Measurability Gap" between what systems can execute vs what humans can practically oversee
- Two destabilizing forces: "Missing Junior Loop" (collapse of apprenticeship) and "Codifier's Curse" (experts codifying their own obsolescence)
- These pressures incentivize "unverified deployment" as economically rational, driving toward a "Hollow Economy"
- Solution: scaling verification alongside agentic capabilities to enable an "Augmented Economy"
Directly relevant to [[scalable oversight degrades rapidly as capability gaps grow with debate achieving only 50 percent success at moderate gaps]] — Catalini provides the economic framing for WHY oversight degrades (verification bandwidth is finite while execution capability scales).
## Key Facts
- Catalini et al. paper published February 24, 2026 on arXiv
- Paper sourced via Alex Obadia tweet, connected to ARIA Research Scaling Trust programme
- Framework distinguishes between 'Hollow Economy' (unverified deployment) and 'Augmented Economy' (scaled verification) as competing trajectories
- Paper identifies verification bandwidth, not intelligence, as the binding constraint on AGI economic impact