Proofreaders and Copy Markers
43-9081.00Median wage $51,120/yr4,580 employed (US)Rank #15 of 923 scored · top 2% by substitution
Read transcript or proof type setup to detect and mark for correction any grammatical, typographical, or compositional errors. Excludes workers whose primary duty is editing copy. Includes proofreaders of braille.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution70
Exposure65
Augmentation82
Substitution — the headline: capability discounted by cost, barriers and adoption.
Exposure — technical capability alone, regardless of whether anyone deploys it.
Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.
Tasks on the substitution scale
11 rated tasks, binned by substitution score.
Position among all scored occupations
Distribution of 923 occupation scores; the marker is this occupation.
Tasks with substitution ≥ 70
55%
Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.
Why this score
The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.
Task automatabilityw 35%66
panel mean rating 3.6/5 → substitution pressure 66/100
Technical feasibility todayw 20%62
panel mean rating 3.5/5 → substitution pressure 62/100
Cost vs. human wagew 15%84
panel mean rating 4.3/5 → substitution pressure 84/100
Adoption barriersw 20%inverted — strong barriers lower the score80
panel mean rating 1.8/5 (barrier strength) → substitution pressure 80/100
Sector adoption velocityw 10%62
panel mean rating 3.5/5 → substitution pressure 62/100
Task breakdown (11 tasks)
Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.
Mark copy to indicate and correct errors in type, arrangement, grammar, punctuation, or spelling, using standard printers' marks.
86CI 79–92 · exposure 87 · augmentation 100 · importance 4.7/5 · click for rater detail
Mark copy to indicate and correct errors in type, arrangement, grammar, punctuation, or spelling, using standard printers' marks.
86| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Publishing, media, and digital-first organizations have rapidly adopted AI proofing tools; major platforms integrate them natively. Adoption is mainstream in information-sector workflows, though some traditional print publishers still rely on human proofreaders. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Publishing, media, and content industries have rapidly adopted AI-assisted proofreading and copyediting tools as standard practice, though full replacement of human copy markers using printers' marks specifically is less common. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI transforms proofreader productivity by pre-marking obvious errors, highlighting ambiguous cases, and accelerating iteration cycles; the human proofreader remains in control, making final judgment on complex or stylistic issues. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI proofreading tools dramatically speed up human copyeditors by flagging and suggesting corrections, allowing humans to focus on judgment calls and style while the tool handles mechanical error detection. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems (large language models with vision capabilities) can identify and mark most grammatical, punctuation, spelling, and arrangement errors with high accuracy, delivering substantial time savings. However, the requirement to use specific standard printers' marks and occasional judgment calls on style nuance prevent a perfect 5. |
| Task automatability | claude-sonnet-5 | 5/5 | Grammar/spell-checking and copyediting is a well-solved NLP task; modern AI tools can detect and correct errors in type, grammar, punctuation, and spelling faster than humans with comparable or better accuracy for most text.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | There are minimal regulatory or legal barriers; proofreading is not a licensed profession. Publishers and clients may prefer human review for final sign-off, but this is organizational friction rather than a hard barrier. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for proofreading; some publishers still prefer human final sign-off for high-stakes or literary text, creating mild organizational friction but no legal barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered proofreading tools cost pennies per document after initial setup, while professional human proofreaders typically charge $25–75+ per hour or per page, making AI at least 10–100× cheaper at scale. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated proofreading tools cost a few dollars per month or fractions of a cent per document versus the loaded hourly wage of a human proofreader, an order-of-magnitude or greater saving. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products like Grammarly, Microsoft Editor, and specialized proofing tools reliably detect and flag errors in production. Vision-based systems can also parse marked-up documents and suggest corrections, though perfect end-to-end automation with traditional printers' marks is less common in deployed systems. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Deployed products (Grammarly, MS Editor, AI copyediting tools integrated into publishing workflows) reliably perform this at scale across millions of documents daily. |
Compare information or figures on one record against same data on other records, or with original copy, to detect errors.
86CI 79–92 · exposure 83 · augmentation 75 · importance 4.4/5 · click for rater detail
Compare information or figures on one record against same data on other records, or with original copy, to detect errors.
86| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Information-dense sectors (finance, insurance, publishing, tech) are rapidly adopting automated document comparison and data validation tools; pilot and production deployments are common, driving measurable displacement of manual proofreading tasks. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Publishing, finance, and administrative sectors have rapidly adopted automated proofing and reconciliation tools, though some smaller firms still rely on manual checks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI augments human proofreaders by automatically flagging likely errors and discrepancies, allowing humans to focus on ambiguous cases and context-dependent judgment. This significantly boosts productivity while keeping humans in quality-assurance roles. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly speeds up error detection by flagging discrepancies for human review, improving accuracy and throughput while a human still confirms final corrections. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can reliably detect discrepancies between records or against source material using OCR, NLP, and rule-based comparison at scale. While human judgment on context or intent may occasionally be needed, the core comparison task meets the ≥50% time-saving threshold with current tools, though setup and edge cases prevent a full 5. |
| Task automatability | claude-sonnet-5 | 5/5 | AI text comparison and diff tools can automatically detect discrepancies between records or against original copy with high accuracy and substantial time savings.diff/OCR-based tools already do this at scale. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or licensing barriers exist to automating data comparison; however, organizations may retain humans for final sign-off on critical documents and some industries (financial/legal) prefer human accountability, creating moderate friction rather than hard barriers. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or legal requirement mandates a human perform simple data/figure comparison; it's a purely clerical verification task with no liability constraint tied to human sign-off. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven comparison and flagging of discrepancies costs orders of magnitude less per record than manual proofreading labor, especially at scale; inference is cheap and human oversight can be sampled rather than universal. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated comparison software runs at negligible per-document cost compared to a human proofreader's hourly wage for equivalent verification work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (document comparison tools, data validation platforms, and LLM-based review systems) perform this task reliably in production across finance, legal, and publishing sectors. Minor false positives/negatives occur but production systems handle this task at scale today. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed products (grammar/proofing tools, document comparison software, OCR-based QA systems) reliably perform text and figure comparison in production, though complex formatting or ambiguous originals can introduce errors. |
Correct or record omissions, errors, or inconsistencies found.
81CI 79–84 · exposure 75 · augmentation 100 · importance 4.5/5 · click for rater detail
Correct or record omissions, errors, or inconsistencies found.
81| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Adoption is rapid and deep in digital publishing, corporate communications, and SaaS platforms. AI proofreading tools are now standard in word processors and content management systems, with measurable displacement of routine proofreading work in information-intensive sectors. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Publishing, media, and professional services sectors have rapidly integrated AI-based grammar and editing tools into everyday workflows, reflecting fast adoption typical of information-sector tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI assistants substantially amplify human proofreader productivity by flagging errors, suggesting corrections, and handling bulk mechanical fixes, allowing humans to focus on nuanced style, tone, and context—transforming the task while keeping the human in full control. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools dramatically speed up error detection and correction while human proofreaders retain final judgment on tone, context, and intent, making this a strong augmentation case. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems (LLMs, spell-checkers, grammar tools) can identify and correct most spelling, grammatical, and formatting errors with high accuracy, achieving well over 50% time savings on routine proofreading tasks. However, subtle stylistic inconsistencies and context-dependent judgment calls may still require human review, preventing a perfect 5. |
| Task automatability | claude-sonnet-5 | 4/5 | Grammar/consistency checking and correction is a well-established NLP capability; modern LLMs and grammar tools can catch and fix most errors and inconsistencies with substantial time savings, though nuanced house-style or factual consistency checks still need human review. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few regulatory or licensing barriers exist for automated proofreading; it is already widely adopted without legal restrictions. The main friction is organizational inertia and human preference for final human sign-off in high-stakes publishing, but these are soft rather than hard barriers. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There is no licensing or legal requirement for a human to perform proofreading corrections, and organizations readily adopt automated grammar/consistency tools without regulatory friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based proofreading tools cost pennies per document to run (inference + integration) versus the loaded hourly wage of a professional proofreader ($25–50+/hour), making automation at least an order of magnitude cheaper for large-volume work. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated proofreading tools cost a few dollars per month or fractions of a cent per document versus a human proofreader's hourly wage, representing an order-of-magnitude cost reduction for the mechanical correction portion. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products like Grammarly, Microsoft Editor, and language model APIs reliably perform error detection and correction in production at scale. These tools are widely deployed in publishing, corporate, and academic settings, though they occasionally miss nuanced issues or generate false positives that require human oversight. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like Grammarly, Microsoft Editor, and LLM-based editing assistants are deployed at scale and reliably catch spelling, grammar, and many consistency issues in production today, though not perfectly for complex style-guide or factual consistency checks. |
Consult reference books or secure aid of readers to check references with rules of grammar and composition.
81CI 79–84 · exposure 75 · augmentation 100 · importance 4.3/5 · click for rater detail
Consult reference books or secure aid of readers to check references with rules of grammar and composition.
81| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Publishing, content production, and corporate communications sectors are rapidly adopting AI-assisted and AI-automated proofreading. Major platforms integrate these tools natively, and adoption in digital and professional services is deep and accelerating. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Publishing, media, and professional services sectors have rapidly adopted AI writing/editing tools, with grammar-checking AI now a standard part of many content workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI proofreading tools significantly augment human proofreaders by flagging errors automatically, suggesting corrections, and handling high-volume first-pass review. This keeps the human in the loop for judgment while dramatically improving productivity and consistency. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools substantially speed up and improve the consistency of grammar and reference checking while proofreaders remain in the loop for final judgment and nuanced content decisions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI language models can reliably identify grammar, punctuation, and composition errors against standard style guides and rules. While human judgment on nuanced style choices remains valuable, the core task of reference checking and rule-based verification is substantially automatable, meeting or approaching the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 4/5 | AI language models can check grammar, composition rules, and cross-reference style guides very effectively, covering most of the reference-checking function of this task.atibility.It largely meets the time-saving threshold though edge cases (subject-matter accuracy, nuanced style calls) still need human oversight. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or licensing barriers exist; proofreading is not a regulated profession. However, some organizational inertia, client preference for human review, and quality-assurance requirements create modest friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 1/5 | There is no licensing or legal requirement for a human to perform grammar/reference checking; it's a purely technical task with no regulatory or liability barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based proofreading tools cost pennies per document processed, while a human proofreader's fully loaded wage is typically $25–50+ per hour. The cost differential is at least one to two orders of magnitude in favor of AI for routine grammar and composition checking. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | AI grammar and reference-checking tools cost a small fraction (often subscription-based, cents per document) compared to a human proofreader's hourly wage for equivalent volume of checking. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products (Grammarly, Microsoft Editor, QuillBot, and similar tools) perform grammar and composition checking reliably in production at scale. These systems handle reference validation against style rules effectively, though some edge cases and complex stylistic decisions may still require human oversight. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Grammar/style checking tools (Grammarly, AI copyediting assistants integrated into word processors) are mature, widely deployed products used at scale in publishing and business writing workflows. |
Read corrected copies or proofs to ensure that all corrections have been made.
76CI 76–76 · exposure 75 · augmentation 75 · importance 4.6/5 · click for rater detail
Read corrected copies or proofs to ensure that all corrections have been made.
76| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Publishing, printing, and digital media sectors have invested in automated quality control, but adoption remains uneven; many smaller firms and specialized niches still rely on manual proofing. Growth is steady but not yet dominant in all segments. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Publishing and editorial services are moderately digitized with growing AI tool adoption, but many proofreading workflows still rely on manual final checks rather than full AI-driven verification in production. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist human proofreaders by automatically flagging discrepancies and flagging likely errors, reducing manual scanning time. However, the human retains final judgment, and the tool primarily accelerates the existing workflow rather than transforming capability. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI-based diff and comparison tools substantially speed up verifying that corrections were properly incorporated, letting proofreaders focus on subtler quality issues while staying in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably compare original and corrected text to verify corrections have been applied, achieving high accuracy and significant time savings. This is primarily a rule-based verification task with minimal subjective judgment, well-suited to OCR and diff-based automation. |
| Task automatability | claude-sonnet-5 | 4/5 | Comparing text versions and verifying corrections were applied is a well-defined text-diffing and comprehension task that current LLMs and diff tools handle well, though full nuance-checking still benefits from human review. Off-the-shelf tools can automate most of this with high time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist; publishers and organizations can easily substitute automated verification for human sign-off on correction confirmation. Some organizational inertia and customer preference for human review remain, but these are weak friction points. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for this task, though publishers may retain human sign-off for final quality control in high-stakes materials (legal, medical), creating mild organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based document comparison and verification cost pennies per page in inference and integration, while a human proofreader's loaded hourly wage typically yields per-page costs in dollars. Cost advantage is clearly an order of magnitude. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated text comparison and AI proofreading run at a fraction of a cent per document versus a human proofreader's hourly wage, making AI dramatically cheaper for this specific verification subtask. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed products (plagiarism checkers, document comparison tools, automated proofreading software) can perform this verification task in production. While some edge cases exist, mature solutions reliably flag whether corrections match the marked changes. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed diff-checking software and AI-assisted proofreading tools (e.g., Grammarly, PerfectIt, version-comparison tools in publishing workflows) reliably flag unaddressed or mismatched corrections in production today. |
Archive documents, conduct research, and read copy, using the internet and various computer programs.
75CI 75–75 · exposure 75 · augmentation 100 · importance 3.9/5 · click for rater detail
Archive documents, conduct research, and read copy, using the internet and various computer programs.
75| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Publishing, legal, and professional services sectors are rapidly adopting AI-assisted and automated proofreading and document management tools, with pilots moving into production use across digital-first organizations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Publishing, media, and content industries have rapidly adopted AI writing/editing tools, with widespread integration into content pipelines already common. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI assists human proofreaders dramatically by flagging errors in real-time, suggesting alternative phrasings, automating routine archival, and surfacing relevant research, enabling humans to focus on judgment-heavy style and tone decisions. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools substantially speed up copy reading, error-flagging, and research gathering while humans retain final judgment, making this a strong augmentation case. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can automate document archiving via file management scripts, conduct basic internet research using search APIs and web scraping, and perform proofreading with grammar/spelling checkers that meet quality parity with human review in routine cases, collectively achieving substantial time savings on this multi-part task. |
| Task automatability | claude-sonnet-5 | 4/5 | Reading copy for errors and conducting internet research are tasks well-suited to current LLMs and search tools, which can flag issues and gather information quickly, though final judgment on nuanced style/context still benefits from human review.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal legal or regulatory barriers exist; the task does not require professional licensing or legal sign-off, though organizational preference for human review and potential brand/compliance concerns create modest friction to full displacement. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for proofreading; main friction is publisher preference for human final review and quality control, but no legal or regulatory mandate blocks automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Inference costs for LLM-based proofreading and research are now a few cents per document, while loaded human proofreader costs typically run $25–50 per hour; at-scale automation is at least 10–50× cheaper per task equivalent. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-based proofreading and research tools cost a small fraction of a human proofreader's hourly wage for equivalent throughput, though some oversight cost remains. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products exist for document management (Dyn365, SharePoint), research aggregation (Perplexity, ChatGPT with web search), and automated proofreading (Grammarly, language models), though edge cases and nuanced style feedback still require human oversight in production environments. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Grammar/style checkers (Grammarly, ProWritingAid) and AI research assistants are deployed at scale and widely used in publishing workflows today, though not fully autonomous for archiving tasks. |
Typeset and measure dimensions, spacing, and positioning of page elements, such as copy and illustrations, to verify conformance to specifications, using printer's ruler or layout software.
70CI 67–72 · exposure 66 · augmentation 75 · importance 4.3/5 · click for rater detail
Typeset and measure dimensions, spacing, and positioning of page elements, such as copy and illustrations, to verify conformance to specifications, using printer's ruler or layout software.
70| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Publishing, printing, and design firms are gradually adopting automated layout checking and proofing tools, but adoption remains uneven; many smaller print shops and traditional workflows still rely on manual proofreading. Adoption is moving but not yet industry-standard. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Publishing and prepress industries have adopted digital layout tools for years, but full automation of spec conformance checking is still only partially deployed alongside human proofreaders. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI layout analysis tools significantly augment human proofreaders by automating tedious measurement checks and flagging deviations, allowing humans to focus on judgment calls and complex spatial relationships. The productivity multiplier is substantial while human oversight remains valuable. |
| Augmentation potential | claude-sonnet-5 | 4/5 | Layout software and automated measurement tools significantly speed up a proofreader's ability to check spacing and positioning, letting them focus on judgment calls rather than manual measurement. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI can automatically detect layout deviations, measure spacing, and compare positioning against specifications using computer vision and layout analysis tools, achieving significant time savings. Some manual verification may remain for edge cases or complex multi-page layouts, but the core measurement and conformance checking is largely automatable. |
| Task automatability | claude-sonnet-5 | 4/5 | Measuring dimensions, spacing, and alignment against specs is a rule-based, geometric verification task well-suited to software checks and computer vision, with layout software already automating much of the measurement.desde |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No licensing requirement or mandatory human sign-off exists for automated layout verification. Quality control and customer preference for human review provide some friction, but organizational and legal barriers to automation are minimal. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or legal requirement mandates a human perform layout measurement; this is a technical quality-control step with no regulatory or human-contact barrier. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-based layout checking and measurement has very low marginal cost per task once deployed (software inference), whereas human proofreaders command full labor rates. The cost differential heavily favors automation, though integration overhead may be moderate. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated preflight and layout-checking software is inexpensive per-document compared to a human manually measuring with a ruler or checking pixel positions, though some setup and calibration cost exists. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Tools exist (design software with automated checking, computer vision libraries) that can perform dimensional verification, but production-grade systems handling diverse file formats and edge cases remain limited. Most implementations require significant setup and oversight rather than fully reliable off-the-shelf deployment. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Layout/DTP software (InDesign, prepress preflight tools) already automates spacing and dimension checks, but fully autonomous verification against arbitrary specs with edge-case judgment still requires human confirmation in production workflows. |
Read proof sheets aloud, calling out punctuation marks and spelling unusual words and proper names.
69CI 54–84 · exposure 58 · augmentation 75 · importance 3.3/5 · click for rater detail
Read proof sheets aloud, calling out punctuation marks and spelling unusual words and proper names.
69| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Publishing, legal, and financial services—sectors heavy in proofreading—have rapidly adopted automated quality-assurance and text-to-speech tools; multiple commercial platforms report widespread production use and measurable displacement of manual proofreading. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Publishing and editorial sectors have moderately adopted AI-assisted proofing and automated QA tools, though this specific archaic read-aloud method is largely already obsolete in favor of digital diff tools rather than being actively displaced by AI. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI-driven text-to-speech and anomaly detection assistants substantially boost human proofreader productivity by automating the mechanical reading-aloud and flagging deviations, allowing the human to focus on judgment calls and complex editorial decisions. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based spellcheckers, grammar tools, and automated comparison utilities assist proofreaders in catching errors, though the specific paired-reading technique isn't directly augmented so much as replaced by different workflows. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current text-to-speech and optical character recognition systems can automatically read proof sheets aloud and flag spelling/punctuation differences against source text with high accuracy, achieving well over 50% time savings when combined with AI-assisted anomaly detection. However, the nuanced calling out of subtle punctuation and proper name pronunciation may still require some human oversight for specialized or complex material. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can compare text versions and flag discrepancies programmatically, but the specific act of reading aloud with call-outs is a niche verification workflow that's easily replaced by automated diff-checking rather than literal automation of the task as described.; the underlying goal of catching errors is achievable but not via the exact modality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No licensing or legal requirement mandates human proofreaders for this specific task; organizational and quality-control friction exist but are not statutory barriers. Publishers can adopt automated proofreading assistants with minimal regulatory resistance. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or legal requirement mandates a human perform this specific verification task; it's a workflow choice, not a regulated function. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI text-to-speech and OCR inference costs are negligible (fractions of a cent per document), while the human labor cost for reading proofs aloud is substantial; the all-in cost is easily an order of magnitude cheaper with AI systems. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated text comparison and OCR-based diff tools are extremely cheap per document compared to paying two people to read proofs aloud, though some setup and verification oversight is needed. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products exist (text-to-speech engines, OCR tools, and proofreading assistants) that reliably perform most of this task at scale in publishing and legal workflows. Services like automated proofreading suites are in production use, though edge cases around accent/pronunciation of proper names and dialectal nuance remain occasional pain points. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Text-to-speech and automated diff/comparison tools exist and are deployed, but no mainstream product specifically performs this two-person read-aloud proofing workflow as a reliable substitute in production. |
Write original content, such as headlines, cutlines, captions, and cover copy.
54CI 50–59 · exposure 42 · augmentation 88 · importance 3.6/5 · click for rater detail
Write original content, such as headlines, cutlines, captions, and cover copy.
54| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Digital publishers and online media have begun experimenting with AI-assisted headline generation and are pilots are common; however, widespread production adoption in legacy print and quality-focused outlets remains limited. Adoption is uneven across the sector. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Media and publishing sectors are adopting AI writing tools for drafting captions and headlines, but full-scale production reliance without human oversight remains limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at rapidly generating multiple candidate headlines and captions that a human editor can review, refine, and choose from, substantially accelerating the ideation and drafting phase while the human retains quality control and final say. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI is highly effective at generating multiple headline/caption options and drafts for a human to select from and refine, substantially speeding up this specific subtask. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate headlines and captions at scale, writing original, contextually appropriate copy for specific publications requires understanding brand voice, audience nuance, and editorial judgment that current systems struggle with consistently. Meaningful automation would require custom fine-tuning and heavy human oversight, falling short of the 50% time-saving-at-equal-quality threshold for end-to-end performance. |
| Task automatability | claude-sonnet-5 | 3/5 | LLMs can generate plausible headlines, captions, and cover copy quickly, but matching a publication's specific voice, factual accuracy, and editorial judgment still requires human review, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | There are minimal legal or licensing barriers to using AI for headline and caption generation. However, editorial standards, brand control requirements, and organizational preference for human judgment create moderate adoption friction in traditional publishing. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for this creative writing task, though editorial oversight and brand voice concerns create moderate organizational friction before AI-generated copy is published unreviewed. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | API costs for AI inference and basic integration are very low compared to a copywriter's loaded wage, even accounting for oversight. An AI-assisted system can generate multiple candidate versions for human selection at a fraction of the cost of manual drafting alone. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Generating draft headlines and captions via AI is extremely cheap compared to a human writer's time, even with added human review overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | LLMs can produce draft headlines and captions reliably, and several publishing tools integrate AI copy generation; however, production use still requires substantial human editing for tone, factual accuracy, and brand alignment. Material error rates and narrow scope (working best on standardized formats) limit deployment reliability. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI copywriting tools are widely deployed for headline/caption generation in marketing and media, but reliability varies and human editing is typically still required before publication. |
Route proofs with marked corrections to authors, editors, typists, or typesetters for correction or reprinting.
47CI 30–64 · exposure 38 · augmentation 38 · importance 4.3/5 · click for rater detail
Route proofs with marked corrections to authors, editors, typists, or typesetters for correction or reprinting.
47| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Publishing and printing are traditionally slower adopters of process automation compared to digital-native sectors. While some modern digital publishing platforms may have semi-automated routing, the industry overall shows modest AI adoption in proof workflows, with many organizations still relying on manual processes. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Publishing and editorial workflows have moderate digitization with some cloud-based collaboration tools adopted, but many smaller operations still rely on manual routing. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could suggest appropriate recipients based on proof markup or historical patterns, but the task is already simple and fast for humans. The marginal productivity gain from an AI suggestion tool is minimal, as the human cognitive load in routing is low and the cost of error correction is high relative to the task duration. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled workflow tools can flag, tag, and route documents to relevant recipients, improving efficiency, though the proofreading judgment itself remains separate from this routing subtask. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Routing marked proofs to correct recipients requires understanding context, ownership chains, and organizational workflows. While AI could identify recipient categories from markup, the task demands human judgment about exception handling, priority, and organizational routing rules that are not easily standardized. Most of the value lies in the routing logic, which is highly dependent on specific organizational setup. |
| Task automatability | claude-sonnet-5 | 3/5 | Routing documents to appropriate parties can be automated via workflow software or AI-assisted document management, but decisions about routing paths in complex editorial workflows may still need human coordination. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Editorial and publishing workflows are often embedded in organizational processes with legacy systems and human accountability expectations. There are no strict legal barriers, but organizational friction around changing established routing practices and the requirement for human oversight of corrections provide moderate adoption friction. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or regulatory requirement dictates that a human must physically route proofs; it's a logistical task with minimal legal or safety stakes. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Setting up an automated routing system (custom integration, workflow configuration, ongoing maintenance) costs more than the labor savings from automating this relatively quick manual task. The human effort to route proofs is already minimal, so the ROI on automation is poor. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated routing via existing software (email rules, workflow platforms) is very cheap compared to manual routing labor, though integration costs exist. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Document management and workflow systems exist but do not reliably automate the intelligent routing of proofs with marked corrections at production scale. Current systems require significant manual configuration and human oversight to handle the variety of organizational structures and exception cases inherent in editorial workflows. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Document management and workflow automation tools exist and are used in publishing, but this specific 'route to correct party' task is often embedded in broader editorial software rather than a standalone reliable AI product. |
Consult with authors and editors regarding manuscript changes and suggestions.
39CI 32–45 · exposure 25 · augmentation 75 · importance 4.0/5 · click for rater detail
Consult with authors and editors regarding manuscript changes and suggestions.
39| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Publishing and media sectors are experimenting with AI writing tools and editing assistance, but adoption of AI for author consultation remains in pilot phase. Most traditional publishers still rely on human editors for author-facing consultation, though adoption is accelerating in self-publishing and digital-native contexts. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Publishing and editorial fields have adopted AI drafting/editing tools moderately, with pilots common but human-led consultation still standard practice. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at augmenting proofreaders by generating initial suggestions, identifying pattern issues, and drafting explanatory language for feedback, which the human editor then refines and tailors to the author's voice and goals. This significantly raises productivity while the human retains decision-making authority. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools like grammar/style checkers and LLMs can generate suggested edits and rationale that proofreaders can use as a starting point for discussions with authors, meaningfully speeding up the review process. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate editing suggestions and flag issues, consulting with authors requires understanding nuanced authorial intent, negotiating competing editorial priorities, and making judgment calls about style and voice. Current AI lacks the contextual reasoning and interpersonal negotiation skills to perform this end-to-end; human oversight is essential for most consultations. |
| Task automatability | claude-sonnet-5 | 2/5 | This is an interpersonal consultation task requiring negotiation, judgment about author intent, and relationship management, which current AI cannot fully replicate end-to-end even though it can draft suggestions.stralight |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While there are no hard licensing requirements to use AI for manuscript feedback, publishing houses and authors often prefer human judgment on substantive editorial changes, and liability concerns around algorithmic editing recommendations create friction. Quality expectations and reputation risk moderate but do not prevent adoption. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement, but authors often prefer human judgment and relational trust for sensitive editorial decisions, creating some organizational friction to full substitution. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI suggestion generation costs are negligible compared to loaded wages for experienced proofreaders and editors, though full consultation still requires human involvement. Integration overhead is low and scaling is cheap, making per-consultation cost substantially lower than human labor for the AI-assisted portion. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While AI can cheaply generate suggestions, the human consultative dialogue still requires a person's time, so total cost savings versus a human proofreader consulting directly are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs genuine two-way consultation with authors about manuscript changes. AI can suggest edits and participate in chat, but editorial consulting requires understanding subtle preferences, explaining rationale convincingly, and adapting to author pushback—tasks where current systems are unreliable and narrow in scope. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI writing assistants can generate suggested edits and comments, but no deployed product reliably conducts the collaborative discussion and negotiation aspect of consulting with authors/editors. |
Related occupations — Office & Administrative Support
How to read this
A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.
What would change this score
New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.