Eligibility Interviewers, Government Programs
43-4061.00Median wage $54,210/yr154,800 employed (US)Rank #107 of 923 scored · top 12% by substitution
Determine eligibility of persons applying to receive assistance from government programs and agency resources, such as welfare, unemployment benefits, social security, and public housing.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution46
Exposure49
Augmentation69
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
17 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
0%
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%52
panel mean rating 3.1/5 → substitution pressure 52/100
Technical feasibility todayw 20%43
panel mean rating 2.7/5 → substitution pressure 43/100
Cost vs. human wagew 15%59
panel mean rating 3.3/5 → substitution pressure 59/100
Adoption barriersw 20%inverted — strong barriers lower the score35
panel mean rating 3.6/5 (barrier strength) → substitution pressure 35/100
Sector adoption velocityw 10%30
panel mean rating 2.2/5 → substitution pressure 30/100
Task breakdown (17 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.
Monitor the payments of benefits throughout the duration of a claim.
67CI 46–87 · exposure 75 · augmentation 75 · importance 3.6/5 · click for rater detail
Monitor the payments of benefits throughout the duration of a claim.
67| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Government benefits administration has rapidly digitized payment monitoring over the past decade; automated systems now perform routine transaction tracking across federal and state programs as standard operational practice, with significant measurable displacement of manual review work. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector agencies are traditionally slow adopters of AI due to legacy systems, procurement cycles, and political/legal scrutiny, though some automated fraud/anomaly detection tools have been piloted. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems already augment eligibility workers by flagging anomalies, generating exception reports, and automating routine status checks, freeing interviewers to focus on complex cases requiring judgment and enabling faster identification of issues that need human investigation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven dashboards and anomaly detection can significantly help interviewers track payment status, flag irregularities, and prioritize cases needing attention, improving efficiency while humans retain decision authority. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Monitoring benefit payments is fundamentally a data tracking and reconciliation task that can be fully automated: AI systems can query payment databases, compare scheduled vs. actual disbursements, flag discrepancies, and generate alerts or reports with minimal human intervention, easily achieving >50% time savings at equal or better quality. |
| Task automatability | claude-sonnet-5 | 3/5 | Payment monitoring against eligibility rules and claim status is largely rule-based data processing that AI/automation can handle, but exception handling, fraud investigation, and claimant communication still require human judgment for full task completion. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While no license is legally required to monitor payments, organizational and procedural inertia around legacy systems, audit trail requirements for accountability, and regulatory mandates for human review of certain flagged cases introduce moderate friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Government benefit programs are heavily regulated with due-process requirements, audit trails, and often statutory mandates for human review of adverse actions, creating substantial legal and procedural barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated systems cost pennies per claim to run ongoing monitoring queries, whereas a human interviewer's loaded wage to perform the same task is $20–30/hour; the cost differential is well over an order of magnitude in favor of automation. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated rule-based monitoring systems are far cheaper to run at scale than continuous human case review, though oversight staff are still needed for exceptions. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 5/5 | Government benefit management systems already deploy automated payment monitoring at scale (e.g., SNAP, unemployment insurance, SSA systems integrate real-time transaction tracking); mature products reliably perform balance checks, overpayment detection, and payment status reporting in production across multiple agencies. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Government benefits systems increasingly use automated monitoring and flagging tools, but most agencies still rely on caseworker review for edge cases, appeals, and compliance verification, limiting fully autonomous deployment. |
Keep records of assigned cases, and prepare required reports.
66CI 60–71 · exposure 70 · augmentation 75 · importance 4.6/5 · click for rater detail
Keep records of assigned cases, and prepare required reports.
66| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Government agencies are adopting case management and reporting automation at middling pace—many pilots exist, but legacy system constraints and procurement cycles slow deep production rollout. Adoption is faster in digitally mature jurisdictions but lags in smaller or resource-constrained agencies. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector and government program administration is traditionally slow to adopt new technology due to procurement cycles, legacy IT systems, and budget constraints. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists eligibility interviewers by auto-populating case records, drafting report sections, and flagging missing documentation, enabling faster and more accurate case closure. The human interviewer remains in the loop for judgment and verification, while AI substantially raises throughput. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly assist by auto-drafting reports, flagging incomplete records, and organizing case data, letting interviewers focus on verification and judgment calls while staying in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Record-keeping and routine report generation are highly structured, rule-based tasks that current AI systems can perform end-to-end with significant time savings. Document assembly, data entry synthesis, and standardized report formatting are well within the capabilities of LLMs and RPA tools, meeting the ≥50% time-saving threshold for most government report templates. |
| Task automatability | claude-sonnet-5 | 4/5 | Record-keeping and standardized report generation from case data is highly structured and templated, making it well-suited to AI/automation tools that can populate forms, summarize case notes, and generate reports from structured inputs. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Government agencies require audit trails, data governance compliance, and often contractual or regulatory sign-off on case records and benefit determinations. While technical barriers are low, organizational and compliance oversight requirements create moderate friction to full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Government record-keeping often requires strict compliance with data privacy, retention, and audit standards, and human sign-off is often required, creating moderate regulatory and procedural friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-driven case record management and report generation costs are orders of magnitude cheaper than human labor per task equivalent, especially when amortized across high-volume government caseloads. Overhead is primarily infrastructure and oversight, not per-report labor costs. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated record-keeping and report generation via software is far cheaper per case than manual clerical labor once integrated, though initial setup and legacy system integration add cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature products for document automation, workflow systems, and AI-assisted report generation are deployed in government agencies today. Case management systems with AI integration exist and perform reliably at scale, though integration with legacy government systems and compliance verification remain sources of minor friction. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Case management systems with AI-assisted documentation and report generation exist and are used in government/social services, but many agencies still rely on manual entry into legacy systems with limited AI integration. |
Compile, record, and evaluate personal and financial data to verify completeness and accuracy, and to determine eligibility status.
60CI 55–65 · exposure 70 · augmentation 75 · importance 4.5/5 · click for rater detail
Compile, record, and evaluate personal and financial data to verify completeness and accuracy, and to determine eligibility status.
60| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Government agencies and large benefit administrators (unemployment, SNAP, Medicaid) have been rolling out automated eligibility screening and document processing for years, with measurable reduction in manual interviews. Adoption is rapid in digitized public systems. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector benefits administration is a notoriously slow-adopting, legacy-system-heavy environment with cautious modernization due to political, legal, and equity concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can assist interviewers by auto-populating forms, flagging missing or inconsistent data, and highlighting potential eligibility issues in real time, substantially speeding the verification and evaluation phase while the human makes final determinations. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly speed up data compilation, flagging inconsistencies, and pre-populating eligibility assessments for human interviewers to confirm, improving throughput while keeping a human decision-maker in the loop. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Most of this task—compiling structured financial data, recording it, and cross-checking against eligibility thresholds—can be automated with current document processing and rule-based systems. However, edge cases requiring contextual judgment (e.g., distinguishing legitimate dependents, evaluating unusual income sources) still need human review, preventing full 5-level automatability. |
| Task automatability | claude-sonnet-5 | 4/5 | Data compilation, verification against rules, and eligibility determination are largely structured, rules-based processes well-suited to AI/automation with document extraction and decision-logic systems, though edge cases need human review. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government programs often require human sign-off, audit trails, and compliance documentation, and some jurisdictions mandate that determinations be reviewed or authorized by a credentialed official. These regulatory and liability barriers slow but do not prevent automation—humans typically oversee rather than fully sign off. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Eligibility determinations for government benefits often carry legal/due-process requirements, appeals rights, and mandated human review or sign-off, creating substantial regulatory and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automated eligibility verification via document processing and rule engines costs a small fraction of a full-time interviewer's loaded wage (typically $40–60k annually), especially at scale; integration and oversight add modest overhead but remain well below human labor cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated data extraction and rules-engine eligibility checks are far cheaper per case than manual interviewer review once systems are built, though integration with legacy government IT adds cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (RPA, document AI, eligibility rule engines) already perform core parts reliably in government and insurance; examples include unemployment and benefits processing systems in production. Accuracy remains high on standard cases, though complex or ambiguous applications still require human escalation. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Government agencies increasingly use automated eligibility systems and document verification tools, but many still rely on caseworker review due to error sensitivity, legacy systems, and fraud/appeals concerns, so deployment is uneven and not fully autonomous. |
Schedule benefits claimants for adjudication interviews to address questions of eligibility.
59CI 51–67 · exposure 62 · augmentation 75 · importance 4.0/5 · click for rater detail
Schedule benefits claimants for adjudication interviews to address questions of eligibility.
59| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Government agencies have adopted scheduling automation piecemeal, but adoption remains uneven across regions and programs. While some state agencies use automated systems, widespread production deployment across federal and state benefits programs is not yet standard, suggesting middling velocity. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector benefits administration is generally a slow-adopting environment for AI due to legacy systems, procurement cycles, and risk-aversion around eligibility determinations. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI scheduling assistants that check claimant availability, suggest optimal times, and flag high-risk cases can substantially improve interviewer productivity by reducing manual calendar work and highlighting urgent cases. The human interviewer retains full control over final scheduling decisions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling assistants can significantly reduce the administrative burden of coordinating interview times, letting caseworkers focus on substantive eligibility review while automation handles logistics. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Scheduling itself is highly automatable (calendar management, availability matching), but the task involves understanding claimant circumstances, adjudication timelines, and program requirements that require some human judgment. Current AI could handle calendar logistics and basic routing but likely needs human oversight for complex eligibility case routing. |
| Task automatability | claude-sonnet-5 | 4/5 | Scheduling interviews based on eligibility rules and calendar availability is a structured, rules-based coordination task well within reach of current scheduling software and AI agents integrated with case management systems.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Government programs often have requirements for human verification of claimant contact and scheduling notifications, but scheduling itself is not legally restricted. Administrative policy, union considerations, and internal oversight protocols create moderate friction against full automation, though not hard legal barriers. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement blocks automated scheduling itself, but government procurement rules, data privacy protections, and union/civil-service protections around case handling create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automated scheduling systems have low marginal per-task cost (primarily API calls and minimal oversight) compared to interviewer time spent on scheduling logistics. The cost ratio heavily favors automation once initial setup is amortized. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated scheduling tools cost a small fraction of a human interviewer's time per scheduling transaction, though integration with legacy government IT systems adds some overhead. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Calendar and scheduling systems with AI routing are deployed in government and enterprise contexts today. Government agencies increasingly use automated interview scheduling systems, though integration with eligibility determination workflows varies. Most systems require some human verification but operate reliably at scale. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Government scheduling systems and chatbots exist and are deployed in some agencies, but many still rely on manual case worker scheduling due to legacy systems and integration gaps, so reliability varies widely across jurisdictions. |
Compute and authorize amounts of assistance for programs, such as grants, monetary payments, and food stamps.
58CI 41–75 · exposure 66 · augmentation 75 · importance 4.7/5 · click for rater detail
Compute and authorize amounts of assistance for programs, such as grants, monetary payments, and food stamps.
58| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Government agencies have adopted eligibility calculation systems, but automation of authorization decisions remains partial and slow due to regulatory and political friction; pilots are common but full end-to-end automation without human sign-off is rare. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector benefits administration is a notoriously slow-adopting sector for AI due to legacy systems, procurement cycles, and regulatory caution, though some automated eligibility calculators exist. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI systems substantially assist eligibility interviewers by pre-computing amounts, flagging edge cases, and preparing recommendations, allowing humans to focus on verification and judgment; this has proven productivity gain in deployed systems. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI and automated calculation tools substantially speed up computing benefit amounts and flagging eligibility issues, letting caseworkers focus on verification and edge cases while still making final determinations. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Computing assistance amounts for government programs follows deterministic rules (income thresholds, household size multipliers, program-specific formulas) that can be fully automated; authorization of these amounts involves rule-based logic that current systems can execute reliably, meeting the ≥50% time-saving threshold with existing software. |
| Task automatability | claude-sonnet-5 | 3/5 | Computing benefit amounts from eligibility rules is a rules-based calculation AI can largely automate, but 'authorizing' implies a legal decision step that typically requires human sign-off and handling edge cases, exceptions, and verification of documentation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: government programs require audit trails, human accountability, and often explicit legal/regulatory requirements that a licensed human sign off on eligibility determinations; liability for erroneous payments and overpayments creates organizational reluctance to fully automate without human review. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Authorizing government benefits often requires a designated authorized official under statute/regulation, with legal accountability for errors, fraud prevention duties, and appeals processes that resist full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The cost of running rule-based eligibility engines and payment authorization is negligible (pennies per transaction) compared to the loaded wage of a human eligibility interviewer (typically $35–55k annually), making AI at least two orders of magnitude cheaper. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated rules engines and AI-assisted calculation are far cheaper to run at scale than manual computation by trained staff, though oversight costs reduce the full savings. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Benefits calculation and authorization systems are deployed at scale in many state and federal agencies (e.g., SNAP, TANF eligibility engines), though some jurisdictions still rely on legacy systems with manual steps; production systems exist but integration varies by agency. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some government agencies use automated eligibility engines for calculations, but full authorization workflows still rely on caseworker review; deployed end-to-end AI authorization systems are not widespread in production due to legal and accuracy requirements. |
Answer applicants' questions about benefits and claim procedures.
54CI 46–62 · exposure 58 · augmentation 88 · importance 4.3/5 · click for rater detail
Answer applicants' questions about benefits and claim procedures.
54| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies adopt technology slowly compared to private sector, with limited digitization in many jurisdictions and frequent reliance on legacy systems. While some states pilot chatbots, widespread production deployment remains limited. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Public sector agencies are adopting chatbots and self-service portals but at a slower, more cautious pace than private-sector professional services due to procurement cycles and compliance concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can effectively assist interviewers by drafting responses to common questions, retrieving relevant policy documents, and flagging potential eligibility issues, significantly raising the productivity of human staff without removing them from the loop. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI tools can draft responses, look up policy details, and provide interviewers instant answers to complex questions, meaningfully speeding up service while the human retains oversight for edge cases. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can handle routine, FAQ-style questions about standard benefit programs and procedures with consistent scripts, but struggles with edge cases, policy nuance, and complex individual circumstances that require judgment. Roughly half the volume of questions could be automated, but the harder cases still need human review. |
| Task automatability | claude-sonnet-5 | 4/5 | Answering FAQ-style questions about benefits eligibility and claims procedures is well within current LLM/chatbot capability, especially with retrieval-augmented systems trained on program rules.yBut edge cases and complex individual circumstances still need human judgment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government benefit programs face regulatory oversight, liability for incorrect eligibility information, and high stakes for vulnerable populations; agencies must ensure accuracy and often require human sign-off on eligibility determinations. Customer preference for human interaction and formal notice requirements also create friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement to answer general questions, but government liability, privacy rules, and need for accuracy on eligibility determinations create moderate institutional caution and required human backstops. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-powered question-answering has very low marginal cost once deployed, making it substantially cheaper than human interviewers for high-volume routine questions, though integration and oversight infrastructure add some cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated chat/voice systems handling routine benefit questions cost a small fraction of a human interviewer's loaded wage per interaction, though integration and oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Chatbots and FAQ systems are deployed in many government agencies to field common questions, but accuracy gaps and liability concerns mean production systems still route complex or sensitive inquiries to humans. Material error rates on policy interpretation and eligibility determination remain common. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Many government agencies and benefits providers have deployed chatbots/virtual assistants for FAQ-style benefits questions, but accuracy on nuanced eligibility rules remains imperfect and complex cases are routed to humans. |
Provide applicants with assistance in completing application forms, such as those for job referrals or unemployment compensation claims.
53CI 46–60 · exposure 58 · augmentation 75 · importance 3.9/5 · click for rater detail
Provide applicants with assistance in completing application forms, such as those for job referrals or unemployment compensation claims.
53| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies adopt automation slowly due to legacy systems, union protections, regulatory caution, and funding constraints; while some jurisdictions pilot AI-assisted intake, production deployment at scale remains limited compared to private-sector adoption. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector agencies are slower adopters of AI due to procurement cycles, equity concerns, and legacy IT systems, despite growing chatbot pilots. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially augment interviewers by auto-populating forms from documents, flagging missing fields, retrieving relevant policy guidance, and pre-screening applications, allowing human staff to focus on verification and judgment rather than clerical work. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can pre-fill forms, answer FAQs, and flag missing information, significantly speeding up the interviewer's assistance role while they retain final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can assist with form completion by extracting applicant information, populating standard fields, and identifying required documents, but eligibility nuances, context-dependent questions, and verifying applicant circumstances still require human judgment to ensure accuracy and compliance. |
| Task automatability | claude-sonnet-5 | 4/5 | Guiding applicants through structured forms is a well-defined, rules-based task that chatbots and AI-assisted portals can largely handle, though edge cases and empathy needs remain. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government benefit programs have strict regulatory requirements, legal accountability for incorrect eligibility determinations, and mandates in many jurisdictions that human staff conduct interviews and make final determinations; liability and compliance rules create substantial adoption friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement for this specific task, but government programs often require human oversight for eligibility determinations and applicant appeals, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI-powered form assistance systems have low inference and integration costs relative to the wage of a government eligibility interviewer, with minimal oversight overhead for routine applications, making the AI cost significantly lower per task-equivalent. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated form-assistance chatbots and AI guidance systems cost far less per interaction than staffing human interviewers for routine form help. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Chatbots and document-processing systems exist to guide form completion and answer FAQ-style questions, but they struggle with edge cases, complex eligibility rules, and real-time verification of applicant status—production deployments remain limited to basic guidance rather than end-to-end assistance. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Many government agencies deploy chatbots and online form-assistance tools, but error rates, complex eligibility rules, and need for human escalation still limit full reliability. |
Prepare applications and forms for applicants for such purposes as school enrollment, employment, and medical services.
53CI 46–60 · exposure 58 · augmentation 75 · importance 3.9/5 · click for rater detail
Prepare applications and forms for applicants for such purposes as school enrollment, employment, and medical services.
53| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies, particularly those processing benefits, adopt automation slowly due to budget constraints, legacy systems, risk aversion, and union/labor considerations. Pilots exist but production displacement is limited; most agencies still rely on manual form preparation. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government and public assistance sectors tend to be slower adopters of AI due to legacy systems, procurement cycles, and public accountability requirements, despite some pilot programs. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can meaningfully assist interviewers by pre-filling forms, auto-detecting missing fields, flagging inconsistencies, and preparing draft documentation, raising their throughput and reducing errors. An interviewer using such tools would handle significantly more applicants without sacrificing accuracy. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can pre-fill forms, auto-populate applicant data from documents, and flag missing information, meaningfully speeding up the interviewer's workflow while they retain oversight and applicant interaction. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | AI can extract, organize, and populate standard form fields from applicant-provided information with moderate accuracy, and can prepare much of the routine documentation assembly. However, eligibility determination often requires judgment about borderline cases, verification against external systems, and handling of incomplete or contradictory information that requires human review, limiting time savings to roughly 40–60%. |
| Task automatability | claude-sonnet-5 | 4/5 | Filling out structured forms based on applicant-provided data is a well-defined text/data task that current AI systems can largely automate, especially with document-parsing and form-filling tools integrated into workflows.") }, |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government programs typically have statutory or regulatory requirements that a qualified human (often credentialed) must certify eligibility decisions and applications; liability for incorrect forms rests with the agency, creating strong error-cost asymmetry. Additionally, federal and state rules often mandate human judgment and discretion in borderline cases. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Government programs often require verification of identity, eligibility, and signed applicant consent, creating moderate procedural and compliance barriers even though the task itself isn't legally restricted to licensed professionals. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Form preparation and data entry automation is relatively low-cost per instance (inference + API calls to document processors), and the loaded wage for government eligibility interviewers is moderate. AI systems can achieve significant cost advantage on routine form assembly, though oversight labor partly offsets gains. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated form-filling and data extraction tools are inexpensive to run at scale compared to a human interviewer's loaded wage for repetitive clerical work. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Commercial document processing and form-filling tools (RPA, intelligent document capture) exist and are deployed in some government agencies, but error rates on complex forms remain material, and integration with legacy government systems is often incomplete or unreliable. Production use is patchy rather than mature across the sector. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Some government and healthcare intake systems use AI-assisted form completion and OCR/autofill, but full end-to-end reliable deployment across diverse eligibility programs is still limited and often requires human verification. |
Interview benefits recipients at specified intervals to certify their eligibility for continuing benefits.
52CI 43–62 · exposure 58 · augmentation 75 · importance 4.5/5 · click for rater detail
Interview benefits recipients at specified intervals to certify their eligibility for continuing benefits.
52| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Government agencies are piloting AI intake and document-processing tools, but deployment has been gradual and uneven. Federal and state barriers to rapid rollout, legacy systems, and cautious regulatory approach slow adoption compared to private-sector information work. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector agencies are historically slow adopters of AI due to procurement cycles, legacy IT systems, unionized workforces, and legal caution, resulting in pilots but limited widespread production deployment. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially assist interviewers by pre-screening documents, flagging inconsistencies, and presenting verified facts before the interview, reducing manual lookup time and improving consistency. Human judgment on borderline cases and customer rapport remain central, supporting continued human involvement. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can pre-fill applications, verify documents, flag inconsistencies, and summarize case histories, meaningfully speeding up interviewer workflows even though the interview and final certification decision remain human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can automate most of the routine eligibility verification workflow—document processing, income/asset checks against databases, and standardized questions—with significant time savings. However, complex cases requiring judgment, fraud detection nuance, or appeals remain challenging, preventing a full 5-rating. |
| Task automatability | claude-sonnet-5 | 3/5 | The information-gathering and eligibility-rule-checking portions can be automated via chatbots/forms with document verification, but complex cases, fraud judgment, and empathetic handling of vulnerable populations still require human review, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Government programs have statutory requirements for eligibility determination and often explicit documentation mandates, and agencies face reputational and legal liability for wrongful denial. Many jurisdictions require human sign-off, though some automation of routine verification is permitted with supervisory review. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Government benefits programs are subject to due process requirements, anti-discrimination law, and appeals rights, often requiring a human decision-maker of record, creating substantial legal and procedural barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI inference and integration costs for document processing and database lookups are orders of magnitude cheaper than interviewer labor ($25–50k+ loaded wage annually), though oversight and exception-handling add overhead that prevents a full 5-rating. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated intake and document-processing tools reduce costs substantially, but the need for human oversight, appeals handling, and compliance review keeps blended costs closer to parity rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Several government agencies have deployed chatbots and AI-assisted intake systems for routine eligibility screening, but most implementations still require human oversight and struggle with edge cases. Production-grade fully autonomous systems remain limited; most are augmentative rather than fully end-to-end. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Some government agencies deploy online portals and automated eligibility verification systems, but these often handle only straightforward renewals; complex or contested cases still route to human interviewers, and error rates in edge cases remain a concern. |
Refer applicants to job openings or to interviews with other staff, in accordance with administrative guidelines or office procedures.
49CI 34–65 · exposure 53 · augmentation 75 · importance 4.0/5 · click for rater detail
Refer applicants to job openings or to interviews with other staff, in accordance with administrative guidelines or office procedures.
49| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government eligibility and benefits administration sectors lag in AI adoption due to legacy systems, procurement complexity, and political/regulatory caution; most jurisdictions remain in pilot or manual-process phases. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government administrative and social services offices are typically slower adopters of AI due to legacy systems, procurement processes, and compliance concerns compared to private-sector staffing agencies. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI can substantially augment human interviewers by automatically surfacing matching job openings, flagging qualification gaps, and organizing candidate-to-interview routing, allowing staff to focus on judgment-based fit assessment and compliance. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can substantially speed up matching applicants to relevant job openings and interview slots, letting interviewers focus on complex cases while routine referrals are AI-assisted. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Matching applicants to job openings based on criteria and routing them to interviews is partially automatable through database lookup and rule-based matching, but requires some judgment about applicant fit and coordination with staffing schedules that currently demands human oversight. |
| Task automatability | claude-sonnet-5 | 4/5 | Matching applicants to job openings or scheduling interviews based on defined criteria is a rule-based, structured task that current AI systems (chatbots, matching algorithms, scheduling tools) can largely automate with high time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government programs typically require adherence to strict administrative guidelines, equal opportunity rules, and procedural documentation; human oversight is often mandated, and liability for incorrect referrals creates legal and organizational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for making referrals, though government programs may have procedural or audit requirements that require human review of referral decisions in some cases. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Building and maintaining a reliable automated referral system with proper integration into legacy government systems, combined with necessary human oversight, makes the all-in cost comparable to or higher than a human interviewer performing the task. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated matching and scheduling systems cost a small fraction of a human interviewer's time per referral, though some oversight and system integration costs remain. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While some specialized job-matching or referral systems exist in government, most lack the integration, accuracy, and procedural compliance needed to handle this end-to-end reliably without human verification and decision-making. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Job-matching and referral tools exist and are used in workforce agencies and HR platforms, but many government eligibility offices still rely on manual referral processes tied to case-specific administrative rules, limiting reliability at scale. |
Interpret and explain information such as eligibility requirements, application details, payment methods, and applicants' legal rights.
38CI 25–51 · exposure 38 · augmentation 63 · importance 4.5/5 · click for rater detail
Interpret and explain information such as eligibility requirements, application details, payment methods, and applicants' legal rights.
38| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies tend to be slower adopters of AI automation due to regulatory constraints, public accountability concerns, and the critical nature of eligibility determinations. Pilots exist but deployment remains limited compared to private-sector information roles. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector and government benefits administration are historically slow adopters of AI due to legacy systems, procurement cycles, and regulatory caution, despite some pilot chatbot deployments. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist interviewers by drafting explanations, retrieving relevant policy details, and flagging potential eligibility issues, improving productivity on routine interpretation tasks while humans retain decision authority and responsibility for legal accuracy. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can efficiently draft explanations, surface relevant policy sections, and answer routine eligibility questions, significantly speeding up interviewers' work while they retain responsibility for final determinations and complex legal explanations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Interpreting and explaining eligibility requirements could be partially automated for routine queries, but current AI systems struggle with the nuance of legal rights, applicant-specific circumstances, and the need to verify complex eligibility rules across state/federal programs. Significant human oversight remains necessary for consistent accuracy and to avoid costly errors. |
| Task automatability | claude-sonnet-5 | 3/5 | Explaining eligibility rules and application details is largely information retrieval and communication, which chatbots/LLMs handle well, but edge cases and legal rights nuances still require human judgment and accountability, limiting full end-to-end automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government programs typically have regulatory requirements that a qualified human employee or licensed professional must explain legal rights and verify eligibility determinations. Liability asymmetry is high—incorrect guidance affects benefit receipt and legal standing—creating organizational and legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No formal licensing is typically required for this role, but government agencies impose accuracy, privacy, and due-process obligations (informing applicants of legal rights) that create liability concerns and demand human oversight or sign-off in ambiguous cases. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI chatbot infrastructure and oversight labor for accuracy verification across diverse program rules remains costly relative to simple front-line eligibility interviewer wages, especially when accounting for liability and the need to escalate complex cases to human experts. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | AI-driven virtual assistants and knowledge-base chatbots cost far less per interaction than a trained human interviewer, though oversight and escalation paths add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While chatbots exist for basic information delivery, no deployed production system reliably interprets and explains complex eligibility criteria and legal rights at the accuracy required for government program administration. Products that exist tend to have narrow scope or material error rates on edge cases. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Government and benefits chatbots (e.g., for SNAP, Medicaid, unemployment) exist and answer eligibility questions, but they often have narrow scope, require human backup for complex cases, and error rates on nuanced legal rights explanations remain material. |
Provide social workers with pertinent information gathered during applicant interviews.
36CI 25–48 · exposure 38 · augmentation 75 · importance 4.1/5 · click for rater detail
Provide social workers with pertinent information gathered during applicant interviews.
36| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government agencies typically adopt AI slowly due to procurement cycles, compliance requirements, and organizational inertia. Public-sector digitization lags private sectors, and social-services agencies are particularly conservative with systems affecting eligibility and benefit decisions. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government social services are historically slow adopters of AI tools due to legacy IT infrastructure, procurement cycles, and compliance concerns, placing this sector on the lower end of adoption velocity. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI summaries, structured templates, and automated flagging of key facts can meaningfully speed up interviewer workflows and reduce transcription burden, allowing social workers to focus on judgment-heavy decisions and client interaction. This use case is practical and improving interviewer productivity without replacing human gatekeeping. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully assist interviewers by organizing, summarizing, and flagging key details from applicant interviews for social workers, improving speed and consistency while humans remain responsible for judgment and follow-up. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can extract and summarize interview data, the task requires human judgment to determine what information is 'pertinent' to each social worker's needs, which varies by case complexity and context. No current system reliably automates the full filtering and contextual prioritization required. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can extract and summarize pertinent applicant information from interview notes or transcripts and route it to social workers, but conducting the interview and judging what's 'pertinent' in context still requires human oversight, so only part of the task is fully automatable today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government employment is often unionized, with formal job protections and civil-service regulations that limit at-will automation. Social services also carry liability concerns: incorrect information handoffs can harm vulnerable populations, creating legal and reputational barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | Government eligibility data involves privacy regulations (e.g., HIPAA-like protections, welfare confidentiality rules) and requires accountability for accuracy, creating moderate procedural and compliance friction even though no license is strictly required to summarize information. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-driven summarization tools reduce processing time but require setup, integration with legacy government systems, and ongoing human oversight to catch errors. Combined costs approach or match the wage of entry-level interviewers in many jurisdictions. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Using AI to summarize and transfer interview data could be cheaper than manual compilation, but integration with legacy government systems and required human verification narrows the cost advantage to roughly comparable levels. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | Document processing and summarization tools exist and are deployed in some government settings, but they require significant human review for accuracy and often miss nuanced, context-dependent details critical to social work decisions. Error rates remain material. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Case management systems with AI-assisted summarization exist in some government agencies, but reliable, widely deployed products that autonomously extract and transmit pertinent interview information to social workers are not yet standard practice. |
Check with employers or other references to verify answers and obtain further information.
32CI 23–43 · exposure 33 · augmentation 63 · importance 4.3/5 · click for rater detail
Check with employers or other references to verify answers and obtain further information.
32| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Government agencies deploying eligibility systems move slowly due to compliance, audit, and accountability requirements. Public sector benefits administration lags private sector automation; adoption of AI for reference verification in production remains minimal, with most agencies still relying on manual processes or legacy rule engines. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government benefits administration is a notoriously slow-adopting sector for AI due to legacy systems, procurement cycles, and compliance concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting verification request templates, summarizing employer responses, and flagging inconsistencies in applicant answers, helping an interviewer work faster. However, the human must retain full decision authority and accountability over which references to contact and how to resolve conflicts, limiting the depth of augmentation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can significantly speed up drafting verification requests, cross-referencing records, and flagging inconsistencies, letting the interviewer focus on judgment calls and follow-up conversations. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Verifying information with external parties requires context-specific decision-making and relationship management. While AI can draft verification requests and parse simple responses, the task demands judgment about which references to contact, how to interpret conflicting information, and when to follow up—activities that require human discretion and accountability in a government benefits context. |
| Task automatability | claude-sonnet-5 | 3/5 | AI can automate the mechanics of outreach (drafting/sending verification requests, parsing responses, flagging discrepancies) but confirming ambiguous or contested information from a human reference still often needs judgment and follow-up calls.automated systems handle a good chunk but not the full verification loop reliably. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government benefits programs have strict documentation and verification requirements, often with regulatory mandates and audit trails. Legal accountability for incorrect eligibility determinations creates liability concerns that deter automation; human interviewers must typically sign off on verified information, and organizational risk-aversion in public sector eligibility work creates strong adoption friction. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement, but privacy rules, data-sharing agreements, and the need for accurate fraud/eligibility determinations create moderate procedural and liability friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI systems still require substantial human review and correction of verification outputs, oversight of reference contacts, and handling of non-standard responses. The all-in cost (inference, integration, error correction, human review) remains comparable to or potentially exceeds the cost of a clerk performing basic verification directly. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated data-matching and outreach tools can be cheap per-check, but cases requiring live phone/email follow-up with employers still need human labor, keeping average cost roughly comparable to human effort for the full task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production system reliably automates end-to-end reference verification for government eligibility determinations. AI can assist with email drafting and response parsing, but deployed products cannot independently verify answers through employer contact without significant human oversight and manual follow-up due to the legal accountability required in benefits administration. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some government/benefits systems use automated verification via data matching (employer databases, IRS/SSA feeds), but direct interviewer-style calls to employers or references for clarification are still mostly manual in production. |
Interview and investigate applicants for public assistance to gather information pertinent to their applications.
27CI 25–29 · exposure 25 · augmentation 63 · importance 4.5/5 · click for rater detail
Interview and investigate applicants for public assistance to gather information pertinent to their applications.
27| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government sectors typically adopt AI slowly due to budget constraints, legacy systems, and regulatory caution. While some jurisdictions pilot chatbots for intake, meaningful production deployment of autonomous investigation and eligibility determination remains limited and uncommon. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government social services is a historically slow-adopting sector with legacy systems, budget constraints, and procurement hurdles limiting deep AI integration despite growing interest in chatbot intake tools. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist interviewers by pre-filling forms, flagging inconsistencies in documents, and summarizing applicant histories, meaningfully raising interviewer productivity on routine cases. However, the human remains essential for complex judgment calls and final determination. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can meaningfully speed up data collection, draft summaries, flag missing documentation, and support caseworkers, significantly improving their efficiency even though humans remain central to the interview and judgment process. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can conduct structured intake forms and document collection, interviewing and investigating applicants requires judgment about credibility, context, and nuance in applicant responses that current systems cannot reliably perform end-to-end. The human investigative component—detecting inconsistencies, probing circumstances, and making discretionary eligibility calls—remains difficult to automate at the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can gather structured data and pre-fill forms via chatbots, but investigative judgment, verifying inconsistent claims, and probing for fraud or nuanced circumstances still require human interviewing skill., so full end-to-end automation with equal quality is not yet reached. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant legal and regulatory barriers exist: eligibility determinations often require documented human authority, due-process safeguards, and decisions may be subject to appeal and judicial review. Many jurisdictions explicitly require a human decision-maker or authorized representative to sign off on benefit determinations. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Public assistance eligibility determinations often require accountable government employees under regulatory and due-process rules, with legal liability for wrongful denial/approval, creating strong barriers to full automation of the investigative interview. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Integration of AI into public sector systems is complex and requires significant compliance overhead. When factoring in oversight, validation, and the need for human review of difficult cases, the all-in cost per application processed is not yet substantially cheaper than a government interviewer's loaded wage. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated intake forms and chatbots are cheap to run, but human caseworkers are still needed for verification and follow-up, keeping blended costs only moderately below full human cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Chatbots can collect basic information and some public sector organizations run automated intake systems, but deployed products do not reliably perform full investigation and eligibility determination at scale. Error rates on edge cases, fraud detection, and contextual judgment remain too high for unsupervised deployment. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some government agencies deploy chatbots and online intake forms for initial data collection, but genuine investigative interviewing (detecting inconsistencies, sensitive probing) is not reliably handled by deployed AI products today. |
Initiate procedures to grant, modify, deny, or terminate assistance, or refer applicants to other agencies for assistance.
27CI 25–29 · exposure 25 · augmentation 50 · importance 4.4/5 · click for rater detail
Initiate procedures to grant, modify, deny, or terminate assistance, or refer applicants to other agencies for assistance.
27| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government sectors move slowly on automation; legacy systems, procurement bureaucracy, and public-sector resistance to perceived benefit cuts slow AI adoption. Pilot projects exist but production displacement remains minimal. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Government social services is a historically slow-adopting sector with legacy IT systems, procurement constraints, and political sensitivity around automated benefit decisions, limiting deployment speed. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist interviewers by pre-populating forms, flagging potential eligibility issues, and summarizing applicant documents, moderately improving interview speed and consistency while the human retains decision authority. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI tools can assist caseworkers by summarizing applicant data, flagging likely outcomes, and drafting referral letters, improving throughput even though final determination authority remains human-driven. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with data extraction and initial eligibility checks, the task requires judgment calls on policy interpretation, exceptions, and referrals—many of which involve legal and discretionary elements. Current systems cannot reliably make final grant/deny decisions end-to-end without substantial human oversight, falling short of the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | While AI can help process eligibility data and flag likely determinations, actually initiating grant/modify/deny/terminate actions involves judgment calls, exception handling, and legal consequences that current systems cannot fully automate end-to-end reliably. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government eligibility determinations are heavily regulated; many jurisdictions legally require a human employee or licensed official to make final decisions and sign off on assistance grants or denials. Liability and due-process rules create strong adoption friction. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Public benefits determinations are governed by statute and due process requirements, often requiring documented human review or sign-off, especially for denials or terminations, creating strong legal and procedural barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI inference and integration costs remain modest, but the compliance oversight burden, error correction, and need for human review to validate decisions keep total cost per task near or slightly above a government interviewer's loaded wage. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated eligibility engines can process routine cases cheaply, but the need for human review, appeals handling, and edge cases keeps blended costs closer to comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some government agencies have deployed systems for initial data triage and eligibility screening, but no mature product reliably performs the full decision pipeline (grant/modify/deny/terminate/refer) with accountability. Error costs and legal exposure keep most deployments to advisory-only modes. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some government agencies use rules-based eligibility engines for straightforward cases, but AI-driven initiation of benefit actions in production remains narrow and heavily supervised due to error and fairness concerns. |
Investigate claimants for the possibility of fraud or abuse.
25CI 25–25 · exposure 25 · augmentation 63 · importance 3.9/5 · click for rater detail
Investigate claimants for the possibility of fraud or abuse.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Government adoption of fraud-detection automation is slow; agencies prioritize legal compliance and audit trails over speed. Most government programs still rely on manual review workflows with only incremental AI assistance in case prioritization, not yet deep adoption of autonomous investigation. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector agencies are typically slow adopters of AI due to legacy systems, procurement cycles, and regulatory scrutiny, though some fraud-detection pilots exist in unemployment and benefits systems. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist investigators by flagging high-risk cases, automating document collection and timeline building, and highlighting anomalies in income or asset claims. However, the augmentation is partial—the investigator must still conduct interviews, evaluate intent, and make the final judgment. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven data matching, anomaly detection, and pattern analysis can significantly speed up identification of suspicious claims, meaningfully augmenting investigator productivity while humans retain decision authority. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Detecting fraud and abuse requires nuanced judgment, contextual understanding, and often adversarial reasoning against deceptive claimants. While AI can flag suspicious patterns in documents and data, the investigative synthesis—distinguishing deliberate fraud from errors, assessing witness credibility, and building a defensible case—remains largely human-dependent and cannot reliably achieve 50% time savings at equal quality today. |
| Task automatability | claude-sonnet-5 | 2/5 | Fraud investigation requires judgment, cross-referencing inconsistent evidence, and interviewing skills that current AI cannot fully replicate end-to-end, though AI can flag anomalies for human review.4 |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: government fraud cases often require documented chain of custody, admissible evidence, and human sign-off before enforcement action. Liability for erroneous findings is asymmetric (false accusations carry reputational and legal risk), and many jurisdictions require a licensed investigator or attorney to certify fraud determinations. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Government programs are subject to due process, privacy, and anti-discrimination regulations requiring human oversight and accountability for fraud determinations, creating strong legal and procedural barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted document scanning and pattern matching reduce some investigator labor, but the overhead of integration, false-positive triage, and mandatory human review means the all-in cost per investigation remains comparable to or higher than direct human investigation, especially when legal defensibility is required. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While anomaly-detection software is cheap to run, the investigative follow-up (interviews, document verification, legal determinations) still requires substantial human labor, keeping overall cost comparable to human-only processes. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed government system reliably automates fraud investigation end-to-end. AI tools can assist with document review and anomaly detection, but production systems in use today still require human investigators for case assessment, follow-up, and prosecution-ready determinations. Narrow pilots exist but do not meet the reliability bar for independent task execution. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Deployed fraud-detection analytics tools exist in benefits agencies but they mainly flag suspicious cases rather than conduct full investigations, which remain human-led with material error/false-positive rates. |
Conduct annual, interim, and special housing reviews and home visits to ensure conformance to regulations.
17CI 14–20 · exposure 16 · augmentation 38 · importance 3.7/5 · click for rater detail
Conduct annual, interim, and special housing reviews and home visits to ensure conformance to regulations.
17| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Government agencies operate under strict procedural and legal requirements; adoption of AI-driven home inspections without human verification is minimal to non-existent. Public-sector digitization in this domain remains slow and conservative. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Public sector eligibility and housing agencies are slow adopters of AI tools, with pilots for document processing but little movement on in-person review automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by pre-screening documents or flagging potential violations before a human visit, but the core task of in-person assessment and regulatory judgment remains fundamentally human-centered, limiting augmentation impact. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help pre-fill forms, flag anomalies for review, and draft reports after visits, meaningfully aiding the interviewer without replacing the visit or judgment calls. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Housing reviews require on-site inspection, subjective judgment about condition and occupancy, and verification of living circumstances that demand human presence and contextual reasoning. While documentation review could be partially automated, the home visit component and conformance assessment cannot be meaningfully automated today to meet the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | The task combines physical home visits, in-person interviewing, and judgment-based compliance assessment, none of which current AI can perform end-to-end; only documentation and data-checking sub-steps are automatable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Government housing programs typically have statutory requirements that a qualified human conduct reviews and home visits; liability for incorrect conformance determinations rests with the agency. Regulatory frameworks generally mandate human verification of living conditions for benefit eligibility. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Government housing programs typically require an authorized caseworker to verify compliance and sign off on eligibility, and physical home visits impose a hard human-presence requirement. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Home visits require human presence and travel; AI cannot substitute for field inspection costs. The per-task expense of deploying human interviewers, including travel and time, remains lower than current AI systems paired with required human oversight for such high-stakes regulatory decisions. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI could cheaply handle paperwork triage, but the core home-visit and interview components still require a paid human, so overall cost savings are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product can conduct home visits or make reliable conformance-to-regulation determinations requiring physical inspection and nuanced judgment. This task fundamentally requires a human agent present at the location. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product conducts home visits or in-person eligibility reviews; AI is at best used for scheduling or flagging cases, not performing the visit itself. |
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.