Tax Preparers
13-2082.00Median wage $54,920/yr76,480 employed (US)Rank #40 of 923 scored · top 4% by substitution
Prepare tax returns for individuals or small businesses.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution59
Exposure58
Augmentation84
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
12 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
42%
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%59
panel mean rating 3.4/5 → substitution pressure 59/100
Technical feasibility todayw 20%57
panel mean rating 3.3/5 → substitution pressure 57/100
Cost vs. human wagew 15%68
panel mean rating 3.7/5 → substitution pressure 68/100
Adoption barriersw 20%inverted — strong barriers lower the score55
panel mean rating 2.8/5 (barrier strength) → substitution pressure 55/100
Sector adoption velocityw 10%62
panel mean rating 3.5/5 → substitution pressure 62/100
Task breakdown (12 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.
Schedule appointments with clients.
92CI 84–100 · exposure 92 · augmentation 75 · importance 3.8/5 · click for rater detail
Schedule appointments with clients.
92| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Tax and accounting firms have rapidly adopted automated scheduling tools; this is among the most widely implemented business processes in the professional services sector. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Scheduling automation is widely adopted across professional services and small businesses, including tax preparation firms, as a mature and common practice, though smaller solo practitioners may lag. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI scheduling systems augment staff by freeing them from manual calendar coordination, allowing receptionists and preparers to focus on higher-value client interactions and tax work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling assistants meaningfully reduce back-and-forth communication and administrative burden for tax preparers, freeing time for higher-value client-facing work. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Scheduling appointments is highly automatable via calendar APIs and email/phone integration; current AI systems can handle initial availability checks, client outreach, and confirmation with minimal human intervention, easily meeting the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 5/5 | Scheduling appointments is a well-bounded, structured task that off-the-shelf scheduling bots and calendar AI tools (e.g., Calendly with AI, virtual assistants) can fully handle with equal or better quality and major time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal barriers exist; scheduling is administrative and purely informational, requiring no licensing or regulatory sign-off, though firms may prefer some human touch for relationship-building. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing, liability, or regulatory requirement mandates a human tax preparer personally schedule appointments; this is pure administrative work with no legal barrier to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Scheduling automation costs pennies per appointment via SaaS tools ($10–50/month) versus a tax preparer's loaded hourly wage ($30–60+), making it at least an order of magnitude cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated scheduling tools cost a few dollars per month versus the loaded cost of staff time spent coordinating appointments, making AI dramatically cheaper per interaction. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Deployed scheduling products (Calendly, Acuity Scheduling, integration with Microsoft 365/Google Workspace) reliably perform this task in production at scale across thousands of businesses, including tax preparation firms. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Mature scheduling products (Calendly, Acuity, AI receptionists, chatbot-based booking systems) are deployed at scale across many service industries including tax and accounting firms today. |
Compute taxes owed or overpaid, using adding machines or personal computers, and complete entries on forms, following tax form instructions and tax tables.
91CI 87–95 · exposure 100 · augmentation 75 · importance 4.8/5 · click for rater detail
Compute taxes owed or overpaid, using adding machines or personal computers, and complete entries on forms, following tax form instructions and tax tables.
91| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Adoption of tax software has been rapid and deep in the information and professional services sector. E-file penetration exceeds 90% in the US, and DIY tax software (Turbo Tax, Credit Karma) captures roughly 40% of the individual-return market, showing strong displacement of manual tax-preparer work. |
| Sector adoption velocity | claude-sonnet-5 | 5/5 | Tax preparation software has achieved deep, fast adoption for decades, with the vast majority of returns filed using automated calculation tools rather than manual computation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | Tax software significantly augments human preparers by auto-populating forms, flagging deductions, cross-referencing tax tables, and detecting errors—allowing preparers to focus on complex scenarios, client advice, and verification rather than mechanical computation and data entry. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-enhanced tax software significantly boosts preparer productivity by automating calculations and flagging issues, though preparers still review, interpret edge cases, and ensure compliance. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Tax computation is a rule-based mathematical task where inputs (income, deductions, credits) map deterministically to outputs via published tax tables and algorithms. Modern tax software (TurboTax, Intuit, etc.) already automates the full workflow for standard returns, delivering >50% time savings compared to manual calculation and form completion. |
| Task automatability | claude-sonnet-5 | 5/5 | Computing taxes and completing forms based on tax tables and instructions is a rule-based, structured-data task that tax software has automated for decades, and modern AI/tax software combinations handle this end-to-end with substantial time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Tax preparation has modest barriers: the IRS permits electronic filing and software-assisted returns, and no license is strictly required to prepare taxes for oneself. However, liability concerns (errors triggering audit exposure), customer preference for human review on complex returns, and state-specific regulations create some friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | While a licensed preparer often must sign/certify the final return, the computation and form-entry itself carries no legal requirement that a human perform the arithmetic, only oversight of the outcome. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-based tax software costs pennies to dollars per return (or fixed SaaS subscription), while a tax preparer's loaded wage for a simple return is typically $150–$300+. AI is orders of magnitude cheaper per task equivalent. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Software-based tax calculation costs a small fraction of a preparer's hourly wage per return, especially at scale, making AI/software an order of magnitude cheaper for this specific computational subtask. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Mature, widely-deployed tax software products (IRS-approved e-file systems, commercial tax preparation platforms) perform this task reliably at scale in production. Millions of tax returns are filed annually through AI-assisted or fully automated systems. |
| Technical feasibility today | claude-sonnet-5 | 5/5 | Mature deployed products (TurboTax, H&R Block software, professional tax prep suites) reliably compute taxes and populate forms at massive scale in production today. |
Calculate form preparation fees according to return complexity and processing time required.
84CI 67–100 · exposure 83 · augmentation 75 · importance 4.0/5 · click for rater detail
Calculate form preparation fees according to return complexity and processing time required.
84| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 5/5 | Tax preparation is a high-digitization sector with rapid software adoption. Automated fee scheduling is already standard in accounting practice management platforms, and firms are actively deploying such tools in production workflows. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Tax preparation is a moderately digitized professional service with growing software adoption, but many small firms still use manual or simple spreadsheet-based fee schedules. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered fee calculators assist tax preparers by instantly computing fees based on dynamic inputs (complexity tier, hours billed, service tier), reducing manual arithmetic and enabling real-time quoting to clients. This meaningfully raises staff productivity while the preparer retains oversight. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven fee calculators and complexity-scoring tools can significantly speed up and standardize fee-setting while preparers retain final pricing discretion. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Calculating fees based on return complexity and processing time is a straightforward algorithmic task: given inputs (complexity level, hours), apply a formula to derive a fee. Current AI systems can extract these parameters and compute fees reliably, easily meeting the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 4/5 | Fee calculation based on rules like return complexity and time is a structured, formulaic task that current AI/software can execute reliably given inputs, though it requires integration with practice management data. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Fee calculation itself carries no regulatory or legal requirement for human sign-off; it is an internal business process. However, some organizational friction exists: firms may prefer a tax professional to review fee calculations for quality control and client relationships, creating modest adoption friction. |
| Adoption barriers | claude-sonnet-5 | 1/5 | No licensing or regulatory requirement mandates human calculation of preparation fees; it's an internal business/administrative decision. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI fee calculation requires minimal inference (seconds of computation, fractions of a cent in API cost) versus a human tax preparer's billable time (15–30 minutes, $50–150+ in loaded labor cost). The cost disparity is at least one order of magnitude. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated fee calculation via software rules engines costs a fraction of a preparer's billable time to manually compute fees for each client. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Multiple accounting and tax software platforms (QuickBooks, TurboTax, CPA practice management tools) already automate fee calculation based on return characteristics and time tracking. These are mature, deployed products used at scale in tax preparation firms. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Tax software and practice management tools already automate fee calculators based on forms/complexity, but many firms still rely on manual judgment or hybrid pricing models, so deployment is uneven. |
Review financial records, such as income statements and documentation of expenditures to determine forms needed to prepare tax returns.
77CI 75–79 · exposure 75 · augmentation 100 · importance 4.5/5 · click for rater detail
Review financial records, such as income statements and documentation of expenditures to determine forms needed to prepare tax returns.
77| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Tax preparation is a digitized, information-intensive sector with strong financial incentives and existing AI integration. Major platforms are already deploying these capabilities in production, and adoption is accelerating across small and mid-market preparers. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Tax preparation software and fintech have rapidly adopted AI-driven document processing and categorization, reflecting fast adoption patterns typical of financial services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI dramatically augments preparer productivity by automatically extracting, organizing, and classifying financial documents and recommending forms, allowing the human preparer to focus on judgment, strategy, and sign-off rather than manual data entry and form lookups. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially speeds up document review and form identification for preparers, letting them focus on judgment calls, exceptions, and client communication. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | AI systems can reliably extract and classify financial data from documents, determine applicable tax forms, and flag missing documentation with high accuracy. This would achieve well over 50% time savings for routine returns, though complex edge cases may still require human review. |
| Task automatability | claude-sonnet-5 | 4/5 | AI tools can extract data from financial documents, categorize income/expense items, and determine applicable tax forms with high reliability, saving substantial time versus manual review, though edge cases still need human verification. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While the final tax return must be signed by a licensed preparer, this particular task (reviewing records and determining forms) has no legal barrier preventing AI automation. Organizational friction and quality oversight remain, but are modest. |
| Adoption barriers | claude-sonnet-5 | 2/5 | Tax preparation has some regulatory oversight (e.g., PTIN requirements, signing preparer liability) but the document-review/form-determination step itself is not legally restricted to being performed by a human. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven document processing and form determination costs pennies per return via cloud APIs, while a tax preparer's loaded wage is $50–80+ per hour. The cost ratio is orders of magnitude in AI's favor for this specific subtask. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated document extraction and form-determination software costs a fraction of a preparer's hourly wage per return, though integration and oversight for complex cases add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products from major tax software vendors (Intuit, Thomson Reuters, etc.) already automate form selection and financial record analysis at scale in production. Error rates are low for standard returns, though some edge cases remain. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Products like TurboTax, tax software with OCR/AI document scanning, and AI-assisted bookkeeping tools already perform this reliably for standard returns at scale, though complex or unusual filings still require human review. |
Check data input or verify totals on forms prepared by others to detect errors in arithmetic, data entry, or procedures.
76CI 74–79 · exposure 75 · augmentation 100 · importance 4.4/5 · click for rater detail
Check data input or verify totals on forms prepared by others to detect errors in arithmetic, data entry, or procedures.
76| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Tax preparation is a high-digitization sector with strong financial incentives for efficiency; major tax software vendors (Intuit, Thomson Reuters, etc.) have embedded automated verification and validation for years, and adoption continues to deepen in corporate and mid-market tax operations. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Tax preparation and accounting software firms have rapidly integrated automated validation and error-detection into mainstream products used at scale during tax season. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI verification tools dramatically augment human preparers by catching errors before they leave the office, speeding review cycles, and allowing humans to focus on complex planning rather than tedious line-by-line checking. This is a mature augmentation pattern in tax practice. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI-driven validation tools significantly speed up and improve preparers' ability to catch errors, letting them focus on complex judgment calls rather than manual re-checking. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems (OCR, rule-based validation, and LLMs) can reliably verify arithmetic, cross-check data entry against source documents, and flag procedural errors in tax forms at scale. While human oversight may still be required for edge cases, the core verification work meets the ≥50% time-saving threshold with high accuracy. |
| Task automatability | claude-sonnet-5 | 4/5 | Checking arithmetic, data entry, and procedural consistency on tax forms is a structured, rules-based verification task well-suited to automated cross-checking and anomaly detection, though edge cases and ambiguous entries still need human judgment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Tax prep automation has no hard licensing barrier—algorithms and software can legally perform verification—but regulatory oversight (IRS compliance, quality standards) and organizational liability concerns (errors must be caught) create moderate friction that slows substitution. Many firms still require a qualified preparer to sign off. |
| Adoption barriers | claude-sonnet-5 | 2/5 | While final sign-off on tax returns has some regulatory requirements, internal error-checking of data entry itself is not a licensed or protected function, so barriers are modest. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated verification via deployed software costs pennies per form checked (post-setup), while a tax preparer's loaded hourly cost for the same verification task runs $25–50+. The cost differential is well above an order of magnitude. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated validation and reconciliation checks run at near-zero marginal cost compared to a preparer's hourly wage for manually re-verifying totals and entries. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed products (tax software with built-in verification, document automation platforms, and RPA solutions) routinely perform arithmetic and data-entry validation in production tax environments. Error rates are low for straightforward checks, though complex procedural validations may require human review. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Tax software (e.g., Intuit, H&R Block tools) and accounting audit features already perform automated error-checking, flagging arithmetic mismatches and missing fields reliably in production, though full procedural/contextual verification still often involves human review. |
Prepare or assist in preparing simple to complex tax returns for individuals or small businesses.
58CI 51–65 · exposure 58 · augmentation 100 · importance 4.5/5 · click for rater detail
Prepare or assist in preparing simple to complex tax returns for individuals or small businesses.
58| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Tax software adoption is already mature and widespread (TurboTax, H&R Block digital, ProConnect) with rapid uptake of AI assistance features. The sector is highly digitized, and AI tax tools are rapidly penetrating both consumer and professional segments. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Tax software and AI-assisted preparation have been adopted rapidly and broadly across individual and small business filers for years, representing one of the more mature examples of AI/automation penetration in professional services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI transforms tax preparer productivity by automating document parsing, suggesting deductions, flagging audit risks, and drafting return sections. The human remains in control of judgment and compliance, making this a strong augmentation case across simple and complex returns. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI dramatically speeds up data entry, error-checking, deduction identification, and scenario modeling for preparers, letting them handle more returns per hour while retaining final judgment and sign-off. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can automate data entry, basic tax form population, and straightforward calculations, offering modest time savings. However, complex scenarios—deductions, business structures, state/local interactions, and judgment calls—still require human expertise and discretion, preventing full end-to-end automation at equal quality. |
| Task automatability | claude-sonnet-5 | 4/5 | AI tools can already gather data, populate forms, and compute complex tax scenarios with high accuracy for a large share of returns, though edge cases and judgment calls on ambiguous deductions still need human review, meeting the 50% time-saving bar for most standard returns. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax preparation and filing involve regulatory oversight (IRS, state boards), authorized practice rules, and preparer liability that create legal and organizational friction. Licensed tax professionals or CPAs often must review and sign returns, and liability for errors constrains full automation substitution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Tax preparers face some licensing (e.g., EA, CPA) and liability requirements for signing returns and giving tax advice, but many jurisdictions allow self-preparation via software, so barriers are moderate rather than absolute. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-driven tax software is substantially cheaper than hiring a tax preparer per return once developed, though oversight, liability insurance, and integration costs apply. The cost advantage becomes pronounced at scale, particularly for simple returns. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Software-driven preparation costs a small fraction of a human preparer's hourly billing for comparable simple-to-moderate returns, though complex cases still require paid professional oversight, tempering the ratio slightly. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI-powered tax software and chatbots exist and perform reliably for simple returns (standard deductions, W-2 income). Material error rates and scope limitations emerge with complexity, and most deployed systems require human review and signing, stopping short of truly independent performance. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed tax software (TurboTax, H&R Block, professional platforms) already automates much of return preparation, but full end-to-end unsupervised handling of complex or unusual returns still requires a licensed preparer, so reliability is scope-limited. |
Explain federal and state tax laws to individuals and companies.
47CI 32–62 · exposure 50 · augmentation 88 · importance 4.2/5 · click for rater detail
Explain federal and state tax laws to individuals and companies.
47| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Tax prep firms have adopted AI-assisted drafting and law-research tools in pilots and some production settings, but widespread adoption of AI-led explanations remains limited by liability concerns, client expectations, and the fragmented, compliance-sensitive nature of small tax preparation practices. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Tax preparation is a professional services sector with growing AI tool adoption (AI-assisted tax software, chatbots), but many small firms and individual preparers still rely on traditional methods, placing adoption at a middling pace. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist tax preparers by quickly retrieving relevant law sections, generating preliminary explanations, and highlighting key rules—substantially raising a preparer's efficiency in researching and synthesizing guidance while they retain judgment and client interaction. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI substantially augments preparers by quickly drafting explanations, summarizing complex tax code changes, and answering client questions, freeing preparers to focus on judgment calls and personalized advice. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can retrieve and summarize tax law content, explaining complex tax laws to individuals and companies requires contextual judgment, personalization to specific circumstances, and clarification of ambiguous regulations—tasks where AI today produces incomplete and sometimes inaccurate explanations that typically require human verification. |
| Task automatability | claude-sonnet-5 | 4/5 | Explaining tax law provisions to clients is largely an information-retrieval and communication task that current LLMs can perform well, drafting clear explanations tailored to a client's situation with significant time savings.dinh However, nuanced edge cases and liability for advice still require human review. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax preparation and advice are regulated activities; while preparers need not be CPAs or attorneys in many cases, they must exercise professional judgment and bear liability for incorrect advice, and clients expect a human professional relationship, creating strong legal and reputational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement strictly bars AI from explaining tax law, but liability for incorrect advice and professional standards (e.g., IRS Circular 230 for enrolled agents) create moderate friction discouraging pure AI reliance for client-facing advice. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference for legal explanations is cheap, but the requirement for human tax preparer oversight and verification—especially for compliance and liability—keeps total cost per reliable explanation-delivery event comparable to or only modestly below human-only delivery. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Generating explanatory content via AI costs fractions of a cent per query versus a preparer's billable hourly rate, though integration and review oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI systems can generate explanations of tax code sections and rules, and some tax software includes AI-assisted guidance, but deployed products struggle with nuanced interpretation, edge cases, and providing legally sound advice tailored to complex financial situations without human review. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI chatbots and tax software already provide tax law explanations to consumers (e.g., TurboTax's AI assistant, general LLM chat), but accuracy on state-specific or complex nuances is inconsistent, so professional preparers still verify explanations. |
Consult tax law handbooks or bulletins to determine procedures for preparation of atypical returns.
44CI 25–62 · exposure 45 · augmentation 88 · importance 4.2/5 · click for rater detail
Consult tax law handbooks or bulletins to determine procedures for preparation of atypical returns.
44| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Tax preparation firms have adopted AI for routine tasks (data entry, document classification) but remain cautious on atypical case handling. Liability, compliance complexity, and client-facing professional standards limit deep automation; adoption is mainly assistive rather than replacement-oriented. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Tax and accounting is a professional services sector with growing AI tool adoption, but full production reliance for atypical return research is still emerging rather than widespread. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered research tools (legal retrieval, precedent matching, bulletin summarization) substantially assist tax preparers in locating and organizing relevant guidance for atypical returns. LLMs can draft reference notes and flag similar cases, materially reducing manual handbook research time while the preparer retains judgment and sign-off responsibility. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI dramatically speeds up locating and summarizing relevant tax law provisions, letting preparers focus judgment on applying rules to atypical facts. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can retrieve and summarize tax law information from handbooks or bulletins, determining the correct procedures for atypical returns requires applying contextual judgment to edge cases where precedent is ambiguous. Current systems cannot reliably synthesize complex, fact-specific tax scenarios without significant human review, falling short of the 50% time-saving threshold for full automation. |
| Task automatability | claude-sonnet-5 | 4/5 | AI language models can search, interpret, and synthesize tax law handbooks and bulletins quickly, and can identify likely procedures for atypical returns with substantial time savings, though edge cases still need expert verification.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax preparation is heavily regulated; CPAs and tax professionals must sign off on returns and bear liability for errors. IRS rules and state licensing requirements mean responsibility for correctness cannot be fully delegated to AI, creating a hard requirement for human professional judgment and attestation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement blocks AI from assisting in research, but professional liability for filing accuracy and preparer sign-off requirements create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted research tools (LLMs, legal databases) reduce lookup time modestly, but the human tax preparer must validate findings and apply judgment. The cost of incorrect output in tax compliance is high, requiring oversight that limits savings to perhaps 20–30% of task time, not sufficient to achieve cost advantage. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-driven legal/tax research is dramatically cheaper per query than billable preparer research time, even accounting for subscription and oversight costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | General-purpose LLMs can search and retrieve tax information but lack production-grade verification for accuracy and liability. No deployed tax-preparation product reliably substitutes for human judgment on atypical returns; existing tools assist research rather than execute independent determination. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like tax-focused LLM assistants and research tools (e.g., Thomson Reuters, CCH AI tools) exist and are used in practice, but reliability on truly atypical/novel returns remains imperfect and requires human review. |
Use all appropriate adjustments, deductions, and credits to keep clients' taxes to a minimum.
43CI 25–61 · exposure 38 · augmentation 88 · importance 4.7/5 · click for rater detail
Use all appropriate adjustments, deductions, and credits to keep clients' taxes to a minimum.
43| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Tax preparation remains dominated by human preparers despite decades of available tax software. Adoption has plateaued at the assisted-preparation level (e.g., TurboTax for self-filers), not replacement. Accountancy firms and tax services have shown minimal velocity toward autonomous AI optimization, reflecting both regulatory barriers and client expectations for human expertise. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Tax preparation software and embedded AI features have been rapidly and widely adopted across both DIY consumer and professional markets over the past decade. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly augments tax preparers today through deduction-finding, credit-matching, scenario modeling, and return drafting—raising their throughput and reducing errors. Tools like AI-powered tax research and automated data extraction help preparers work faster while maintaining their strategic role in optimization and sign-off. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI-powered tax software dramatically speeds up identification of applicable deductions and credits, letting preparers focus on judgment calls and client-specific complexities while the system flags most standard optimizations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can identify many common deductions and credits through tax code knowledge, the task requires understanding highly individual client circumstances, income sources, and life events. Current AI systems lack the contextual judgment and client-specific knowledge integration to reliably minimize taxes end-to-end; they excel at calculation and lookup but not at the strategic, personalized tax optimization that defines this task. |
| Task automatability | claude-sonnet-5 | 3/5 | AI tax software can identify standard deductions/credits and optimize straightforward returns, but complex situations (multi-state, business income, ambiguous credit eligibility) still require human judgment and verification, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax preparation has material regulatory barriers: the tax preparer must be appropriately credentialed (CPA, EA, or attorney) and legally responsible for the return filed. Liability for tax errors is asymmetric and costly. Client preference for a qualified human advisor, IRS regulations, and state licensing requirements create hard friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | While a licensed preparer must sign returns and take liability for accuracy, the actual deduction/credit identification isn't legally restricted to humans, so barriers are moderate but not prohibitive. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | A full end-to-end AI system for tax optimization would require expensive specialized models, verified tax database integrations, and significant oversight by licensed preparers. These costs, combined with the need for human sign-off and error liability, approach or exceed the blended cost of a mid-level tax preparer handling the same work. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Once developed, software-based deduction/credit optimization runs at near-zero marginal cost per return compared to a preparer's hourly billing, though integration and review overhead reduce the savings for complex cases. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Tax software exists for basic return preparation and deduction matching, but no production system reliably performs the full optimization task—determining the *best* combination of adjustments, deductions, and credits for a specific client's situation. Products assist but require substantial human tax expertise and judgment to finalize strategy. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like TurboTax, H&R Block AI assistants, and specialized tax software already surface deductions and credits at scale, but they still rely on human review for edge cases and accuracy guarantees, so reliability is uneven for complex filings. |
Furnish taxpayers with sufficient information and advice to ensure correct tax form completion.
42CI 28–56 · exposure 38 · augmentation 88 · importance 4.6/5 · click for rater detail
Furnish taxpayers with sufficient information and advice to ensure correct tax form completion.
42| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Tax prep software (TurboTax, H&R Block) has adopted AI for form-filling and some advice, but adoption remains on simpler returns and pilots for complex scenarios. Regulatory caution and professional licensing requirements have slowed deeper agent-based automation compared to less-regulated fields. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Tax software and financial services are among the faster-adopting sectors for AI-driven customer guidance, with major platforms already deploying conversational AI tax assistance widely. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI excels at summarizing tax code, identifying likely deductions, cross-checking form consistency, and flagging missing information, meaningfully raising preparer productivity on information delivery and initial compliance review while the human retains legal and advisory responsibility. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools substantially speed up preparers' ability to look up rules, draft explanations, and answer client questions, significantly boosting productivity while the preparer retains responsibility for accuracy and sign-off. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can draft initial tax information and highlight common deductions, but cannot reliably advise on complex situations (business structure, state-local nexus, passive activity rules) that require human judgment and legal knowledge. Material gaps remain in capturing individual circumstances that determine correct form completion. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can answer many common tax questions and draft explanations, but nuanced advice requiring judgment about a taxpayer's full financial situation, edge cases, and liability still needs human review, so only part of this task meets the 50% time-saving bar. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax advice carries regulatory and liability exposure; unlicensed systems giving incorrect advice can expose both preparer and platform to IRS scrutiny and malpractice claims. Most jurisdictions expect a licensed tax preparer to review and sign off, creating a hard barrier to end-to-end automation without human verification. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Tax advice carries liability concerns and in some contexts requires credentialed preparers (e.g., signing returns, representing clients before IRS), creating moderate barriers even though general information-sharing is less restricted. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI systems reduce time on routine information gathering and initial form review, but the marginal cost of human tax preparer oversight for advice quality and liability coverage remains substantial, keeping total cost per task near or above a loaded preparer wage. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-driven Q&A and guidance tools cost a small fraction of a preparer's hourly rate for routine informational tasks, though oversight costs narrow the gap somewhat for complex advice. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While chatbots can provide general tax information and some form-filling assistance exists, no deployed product reliably advises taxpayers on sufficient and correct completion without substantial human review. Error rates remain material, especially for nonstandard situations, and liability risk prevents fully automated advice. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like TurboTax's AI assistant, tax chatbots, and LLM-based advisors exist and are used at scale, but they still have material error rates on complex situations and often escalate to human preparers. |
Interview clients to obtain additional information on taxable income and deductible expenses and allowances.
33CI 25–41 · exposure 25 · augmentation 63 · importance 4.6/5 · click for rater detail
Interview clients to obtain additional information on taxable income and deductible expenses and allowances.
33| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Tax preparation remains a human-led, regulated professional service with slow AI adoption. Firms use AI for supporting tasks (document scanning, form population) but not autonomous client interview replacement; cultural and compliance norms favor licensed preparers conducting interviews. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Tax prep and accounting services have moderate digitization with growing use of intake chatbots and automated organizers, but many firms still rely on human interviews, especially for complex returns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by suggesting standard deduction questions, flagging missing line items, and organizing prior-year data during an interview, moderately improving preparer efficiency. However, the core interview task remains human-driven, so augmentation is useful but not transformative. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered intake questionnaires, document scanning, and chatbot pre-interviews can significantly speed up gathering standard information, letting preparers focus on complex or ambiguous items. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI can extract and organize financial documents and prompt structured questions, but cannot independently conduct genuine interviews requiring judgment, clarification of ambiguous situations, discovery of unreported income, and rapport-building with clients. Meaningful time savings remain far below 50% without substantial human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | Client interviews require adaptive questioning, judgment about ambiguous situations, and building trust to elicit complete/accurate financial disclosures, which current AI can partially support via structured intake forms/chatbots but not fully replace end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax preparation is regulated; the preparer bears liability for accuracy and completeness of client disclosures. Professional liability, client privilege expectations, and IRS enforcement create strong organizational and legal barriers to removing the human-client relationship from this information-gathering stage. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement mandates a human conduct the interview itself, though tax preparers often bear liability for return accuracy, creating some incentive to have humans verify information gathered. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI document processing and basic form-filling is modestly cheaper than transcription, but the human interview remains necessary to validate findings, handle edge cases, and ensure legal compliance. All-in costs remain comparable to or higher than a junior tax preparer conducting the interview. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-driven intake forms are cheap to run, but the need for human review and follow-up on ambiguous answers keeps blended cost closer to comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While chatbots and document-processing tools exist, no deployed product reliably conducts standalone client interviews with legal sufficiency or accuracy to extract complex, idiosyncratic tax situations. Products today require heavy human review and cannot replace the interview itself. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some tax software uses guided questionnaires and chatbots for intake, but these are narrow, scripted, and often require human follow-up for nuanced or unusual situations; no product reliably conducts full free-form client interviews. |
Answer questions and provide future tax planning to clients.
31CI 25–36 · exposure 25 · augmentation 75 · importance 4.2/5 · click for rater detail
Answer questions and provide future tax planning to clients.
31| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Despite digitization in accounting, tax firms remain cautious about delegating client advisory to AI alone due to compliance risk, client preference for licensed professional interaction, and the heterogeneity of tax situations that resists easy standardization. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Professional services and finance sectors are adopting AI assistants at a moderate pace, with pilots and embedded tools in tax software becoming common but full autonomous advisory still rare. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can effectively assist tax preparers by drafting preliminary answers, flagging planning opportunities, and automating scenario modeling, thereby raising preparer productivity while keeping the licensed professional in final judgment and client delivery. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI significantly boosts preparer productivity by quickly answering routine questions, drafting explanations, and flagging planning opportunities, while the human remains responsible for final advice. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can answer routine tax questions and generate basic planning scenarios, the task requires nuanced understanding of individual client circumstances, changing tax law, and personalized strategic advice that current systems struggle to deliver reliably end-to-end without human oversight and refinement. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can answer many general tax questions well, but future tax planning requires synthesizing client-specific financial details, judgment calls, and accountability that current systems cannot fully replicate end-to-end., |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Tax advice carries regulatory and fiduciary requirements; unlicensed AI systems cannot legally provide personalized tax planning without professional sign-off, and malpractice liability creates strong incentives for human authentication and accountability in practice. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement to answer general questions, but liability for erroneous tax advice and client preference for a trusted human advisor create meaningful friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI inference is cheap, but the integration cost, legal review, human oversight, and liability management for tax advice automation remain substantial—often exceeding the cost of a junior tax preparer handling routine questions. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI can cheaply handle routine Q&A, but complex planning still requires human review and liability coverage, keeping blended costs roughly comparable to human preparers for full-service engagements. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Products like ChatGPT can generate tax information, but deployed tax preparation software and AI assistants perform this task only partially and with material gaps in complex scenarios, requiring human tax professionals to validate and customize advice before client delivery. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Chatbots and AI tax assistants (e.g., TurboTax's AI helper) exist but are narrow in scope and prone to errors on nuanced planning scenarios, not yet reliable for comprehensive advisory work. |
Related occupations — Business & Financial Operations
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.