# ALMOST RIGHT — Final Illustration Program

Ten analytical/historical figures remain in the final art program. Five weaker or redundant schematics were cut: Figures 4.3, 5.1, 10.1, 19.1, and 23.1. The documentary-photo program is maintained separately.

Editorial principle: every retained visual must either document a major factual event, carry quantitative evidence that is easier to understand visually, or express a central idea that recurs through the book. Generic process diagrams and illustrative-but-not-data-driven charts have been removed.

## Figure 1.1 — The Dartmouth Proposal (1955)
- Chapter: 1
- Type: Public-domain facsimile of title page
- Purpose: Ground the origin story in the document that named the field.
- Source/Credit: John McCarthy historical archive; source file via Wikimedia Commons.
- Rights: CLEAR — U.S. PUBLIC DOMAIN — Commons marks source work public domain in the U.S. as a 1931–1977 U.S. publication without copyright notice; facsimile prepared for this edition. International rights can differ.
- Alt text: Title page of the August 31, 1955 Dartmouth artificial intelligence proposal naming McCarthy, Minsky, Rochester, and Shannon.

## Figure 4.1 — The Verification Gap
- Chapter: 4
- Type: Original conceptual diagram
- Purpose: Visual statement of the book's core thesis.
- Source/Credit: Author diagram.
- Rights: CLEAR — Original artwork.
- Alt text: Flow from AI generation through verification to consequential reliance.

## Figure 4.2 — Developers Use AI — and Still Don't Trust It
- Chapter: 4
- Type: Original bar chart
- Purpose: Connect the title to the 2025 Stack Overflow survey.
- Source/Credit: Stack Overflow 2025 Developer Survey.
- Rights: CLEAR — Recreated from cited data; do not reproduce source design.
- Alt text: Bars show 66% almost-right frustration, 45% debugging frustration, 33% trust, 46% distrust.

## Figure 11.1 — Where an AI-Built App Can Fail
- Chapter: 11
- Type: Original systems diagram
- Purpose: Separate UI, auth, server, database and production failure surfaces.
- Source/Credit: Author synthesis from Chapter 11 cases.
- Rights: CLEAR — Original artwork.
- Alt text: Five layers where an application can fail independently.

## Figure 13.1 — The Productivity Perception Gap
- Chapter: 13
- Type: Original bar chart
- Purpose: Show expected/perceived speedup against measured slowdown.
- Source/Credit: METR, July 10, 2025.
- Rights: CLEAR — Recreated from study values.
- Alt text: Expected +24%, perceived +20%, measured -19%.

## Figure 14.1 — Unassisted Adenoma Detection Before and After AI Exposure
- Chapter: 14
- Type: Original bar chart
- Purpose: Make the observational medical finding concrete while retaining caveat.
- Source/Credit: Budzyń et al., Lancet Gastroenterology & Hepatology, 2025.
- Rights: CLEAR — Recreated from study values; observational caveat mandatory.
- Alt text: 28.4% before versus 22.4% after AI exposure.

## Figure 15.1 — The Young-Worker Employment Gap
- Chapter: 15
- Type: Original bar chart
- Purpose: Anchor the apprenticeship/labor argument.
- Source/Credit: Stanford Digital Economy Lab, revised Aug. 12, 2026.
- Rights: CLEAR — Recreated from revised finding.
- Alt text: Young workers in highly AI-exposed occupations show a 19% relative employment gap; experienced workers show no comparable gap.

## Figure 17.1 — The Expertise Ladder
- Chapter: 17
- Type: Original conceptual diagram
- Purpose: Show how repetitive work can build later judgment.
- Source/Credit: Author diagram.
- Rights: CLEAR — Original artwork.
- Alt text: Repetition leads to pattern recognition, diagnosis, judgment and verification.

## Figure 20.1 — The Verification Economy
- Chapter: 20
- Type: Original conceptual diagram
- Purpose: Show the economic consequence of abundant generation.
- Source/Credit: Author diagram.
- Rights: CLEAR — Original artwork.
- Alt text: Cheap generation creates abundant output, a verification bottleneck and a trust premium.

## Figure 19.1 — Who Verifies the Verifier?
- Chapter: 19
- Type: Original conceptual diagram
- Purpose: Present the book's proposed layered verification architecture.
- Source/Credit: Author diagram.
- Rights: CLEAR — Original artwork.
- Alt text: AI output passes through tests, domain expert, independent review and escalation.

## Figures removed in final trim

- Figure 4.3 — legal-case count callout: informative but visually thin; the prose carries the number adequately.
- Figure 5.1 — adoption-speed schematic: not to scale, so it added design without enough evidentiary value.
- Figure 10.1 — prompt-to-production flow: generic process diagram; Chapter 10 reads better without it.
- Old Figure 19.1 — liability-gap flow: concept already explained clearly in prose and overlaps the verification architecture. (Distinct from the current Figure 19.1, renumbered from 25.1.)
- Figure 23.1 — jagged-frontier chart: illustrative rather than empirical values; removing it avoids a graphic that could be mistaken for measured data.