Police and Sheriff's Patrol Officers

33-3051.00
Median wage $76,210/yr670,520 employed (US)Rank #858 of 923 scored · top 93% by substitution

Maintain order and protect life and property by enforcing local, tribal, state, or federal laws and ordinances. Perform a combination of the following duties: patrol a specific area; direct traffic; issue traffic summonses; investigate accidents; apprehend and arrest suspects, or serve legal processes of courts. Includes police officers working at educational institutions.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution12
Exposure13
Augmentation41

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

30 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%13

panel mean rating 1.5/5 → substitution pressure 13/100

Technical feasibility todayw 20%13

panel mean rating 1.5/5 → substitution pressure 13/100

Cost vs. human wagew 15%12

panel mean rating 1.5/5 → substitution pressure 12/100

Adoption barriersw 20%inverted — strong barriers lower the score7

panel mean rating 4.7/5 (barrier strength) → substitution pressure 7/100

Sector adoption velocityw 10%12

panel mean rating 1.5/5 → substitution pressure 12/100

Task breakdown (30 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.

Record facts to prepare reports that document incidents and activities.

62

CI 43–82 · exposure 70 · augmentation 88 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Law enforcement is relatively digitized and data-driven compared to many sectors, with body cameras, dispatch systems, and digital filing already standard in many jurisdictions. Pilot programs and early deployment of AI transcription and report-assist tools are documented across major US police departments, indicating rapid adoption in this sector.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a traditionally slow-adopting, highly regulated sector with cautious technology rollout, though a few high-profile AI report-writing pilots are emerging.
Augmentation potentialclaude-haiku-4-5-202510015/5AI transcription and structured report generation substantially augment officer productivity by reducing manual typing and form-filling time while improving accuracy and consistency. Officers remain in the loop for verification, redaction, and judgment, but AI transforms the time cost of this routine task.
Augmentation potentialclaude-sonnet-54/5AI tools can meaningfully speed up drafting of narrative reports from officer notes or recordings, letting officers focus on verification and edits rather than writing from scratch.
Task automatabilityclaude-haiku-4-5-202510015/5Recording incident facts and generating structured reports can be almost entirely automated using AI systems that extract information from audio/video recordings, witness statements, and event logs, then populate standardized report templates. Current speech-to-text, NER (named entity recognition), and document generation systems can handle this end-to-end with >50% time savings at equal or higher accuracy and consistency.
Task automatabilityclaude-sonnet-53/5AI transcription and drafting tools can convert body-cam audio or officer dictation into structured incident reports, but factual accuracy, legal terminology, and context verification still require human review, so only partial time savings are realized.
Adoption barriersclaude-haiku-4-5-202510013/5While incident reports may face chain-of-custody and evidentiary standards, the *recording* of facts (as opposed to investigation and judgment) faces moderate friction rather than hard legal barriers. Agencies prefer human review for accuracy and liability, and some jurisdictions may require officer sign-off, but no strict licensing barrier prevents AI assistance or automation of the data-capture phase.
Adoption barriersclaude-sonnet-54/5Police reports are legal documents used in court, requiring sworn officer certification and accountability, creating strong liability and procedural barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI-based speech recognition, text extraction, and report templating cost a fraction of the officer hourly wage (typically $30–$50/hour loaded) to process and generate a report; API costs for transcription and document generation are typically under $1–$5 per incident, making AI at least an order of magnitude cheaper.
Cost vs. human wageclaude-sonnet-53/5AI drafting reduces time spent on paperwork significantly, but departments still pay for officer review, software licensing, and integration, keeping costs roughly comparable rather than drastically cheaper overall.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (transcription services, AI-assisted report generation, body camera data extraction systems) are already in use by law enforcement agencies. While some integration friction and need for human review of sensitive details remains, mature systems reliably capture and structure incident facts at scale in production environments.
Technical feasibility todayclaude-sonnet-53/5Products like Axon's Draft One and similar AI report-writing tools are deployed in some police departments today, but adoption is narrow and accuracy/liability concerns limit reliability at scale.

Provide road information to assist motorists.

52

CI 25–79 · exposure 50 · augmentation 50 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Police departments are digitizing dispatch and alerts, but actual adoption of AI systems to replace officer road-information provision remains limited; most agencies still rely on officer discretion and verbal communication rather than algorithmic routing of information to motorists.
Sector adoption velocityclaude-sonnet-54/5Consumer navigation and traffic apps have achieved deep, fast adoption across the general public, substituting for much of this informational function already.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-assisted tools (real-time traffic feeds, automated hazard alerting, mobile information systems) can help officers provide more timely and complete road information, raising officer productivity in this narrow task while the officer retains decision-making authority.
Augmentation potentialclaude-sonnet-53/5AI-based traffic data and mapping tools can help officers quickly answer motorist questions or reroute traffic, though this is a minor part of patrol duties.
Task automatabilityclaude-haiku-4-5-202510012/5AI could assist with basic information dissemination (e.g., automated alerts about road conditions, closures, or incidents), but the task inherently involves real-time situational assessment, judgment about relevance to specific motorists, and human communication that current systems cannot fully automate end-to-end with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-54/5Providing directions or road condition info is largely informational and can be handled by GPS apps, chatbots, and automated traffic systems with minimal human involvement for most routine inquiries.
Adoption barriersclaude-haiku-4-5-202510014/5Police officers' road information delivery is embedded in broader patrol and community-contact functions with high human-contact and judgment requirements; legal and organizational frameworks expect sworn officers to exercise discretion on what information to provide in specific contexts, creating friction against pure automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement to give road directions, though officers may combine this with other duties like traffic control or safety checks that keep a human present.
Cost vs. human wageclaude-haiku-4-5-202510012/5Automating road information delivery would require infrastructure investment, integration with dispatch and traffic systems, and human oversight; these costs may approach or exceed the value of the low-skill informational labor being displaced, especially in lower-volume departments.
Cost vs. human wageclaude-sonnet-55/5Automated map/traffic apps cost pennies per query compared to an officer's loaded wage and time spent on roadside assistance.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some automated systems exist for broadcasting road conditions and traffic alerts, but reliable end-to-end performance requires human judgment about which information matters to which motorists, verification of conditions, and adaptive communication that current AI products do not handle reliably in production.
Technical feasibility todayclaude-sonnet-54/5Navigation apps like Google Maps and Waze already provide real-time road information, traffic conditions, and rerouting at massive scale, reliably serving this function today.

Verify that the proper legal charges have been made against law offenders.

42

CI 20–64 · exposure 53 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Law enforcement and prosecution remain traditionalist, slow-to-digitize sectors with strong hierarchical oversight cultures and risk aversion around core legal functions. Adoption of AI-assisted charging verification is nascent and confined mostly to pilot programs in forward-thinking jurisdictions.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a slow-adopting, highly regulated sector with limited production AI use for legal charge verification specifically.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially assist human prosecutors and law enforcement by automating statute cross-referencing, surfacing relevant precedents, and highlighting charging inconsistencies, meaningfully accelerating the verification workflow while the human retains final authority and judgment.
Augmentation potentialclaude-sonnet-53/5AI can help cross-check statutes, precedents, and charge codes, aiding officers in verifying accuracy faster, though final judgment remains human.
Task automatabilityclaude-haiku-4-5-202510015/5Current AI systems can reliably cross-reference arrest details against legal statutes, case law, and charging guidelines to verify proper charges. Large language models and rule-based systems can parse offense descriptions, apply charge determination logic, and flag discrepancies with >50% time savings compared to manual legal research, though human review remains standard practice.
Task automatabilityclaude-sonnet-52/5This requires applying legal judgment to specific facts and jurisdictional statutes, which AI can support with research but cannot reliably finalize end-to-end given liability and nuance.,
Adoption barriersclaude-haiku-4-5-202510014/5Legal liability and judicial authority create hard barriers: prosecutors and law enforcement bear legal responsibility for charges filed, and courts may require human sign-off on charging decisions. Regulatory frameworks tie charging authority to licensed attorneys, limiting full automation even if technically feasible.
Adoption barriersclaude-sonnet-55/5Charging decisions have direct legal and constitutional consequences, requiring sworn officers and prosecutors with legal authority to verify and file charges.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI inference and integration costs for legal document review are modest relative to the loaded wage of paralegals or junior prosecutors who currently perform charge verification; savings are substantial once systems are deployed, though initial setup carries overhead.
Cost vs. human wageclaude-sonnet-52/5Human oversight (officer/prosecutor review) is still required, so AI mainly supplements rather than replaces the cost of trained personnel doing this task.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI products exist that support legal research and charge verification (e.g., legal AI platforms, prosecutorial software with charge recommendation modules), but they operate within narrow scopes and require significant human oversight; no mature system fully autonomously certifies charge legality in production across jurisdictions.
Technical feasibility todayclaude-sonnet-52/5Legal research and charge-code lookup tools exist, but no deployed product independently verifies charge accuracy in live police workflows at scale.

Relay complaint and emergency-request information to appropriate agency dispatchers.

36

CI 25–48 · exposure 38 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Police and emergency services are traditionally cautious adopters of new technology, especially in life-critical functions. While some agencies have piloted AI-assisted transcription or call screening, widespread production deployment of AI-driven dispatch remains rare and slow, constrained by risk aversion, union agreements, and the critical nature of the work.
Sector adoption velocityclaude-sonnet-52/5Public safety and government dispatch operations are historically slow to adopt new digital tools due to legacy systems, budget constraints, and cautious procurement cycles.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist dispatchers by transcribing calls in real time, suggesting categories or priority levels, and flagging keywords—raising their speed and accuracy on high-volume tasks. However, the core judgment (urgency, routing, coordination) remains with the human, so augmentation is real but not transformative.
Augmentation potentialclaude-sonnet-54/5AI-assisted transcription, automatic call classification, and CAD suggestions can meaningfully speed up and improve accuracy of information relay while officers still make final calls.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can parse and categorize incoming complaints with reasonable accuracy, this task requires real-time triage, judgment about urgency and jurisdiction, and handling of nuanced or unclear situations. Current systems cannot reliably handle the full task end-to-end without substantial human oversight, and the consequences of errors (incorrect routing, missed emergencies) prevent the 50% time-saving threshold from being met in practice.
Task automatabilityclaude-sonnet-53/5Routing structured complaint/emergency data between dispatch systems can be automated via CAD integrations and voice-to-text triage, but ambiguous or urgent situational judgment calls still require human relay.'
Adoption barriersclaude-haiku-4-5-202510014/5Strong regulatory and legal barriers apply: emergency services operate under strict liability frameworks, many jurisdictions require a licensed dispatcher to certify dispatch decisions, and the consequences of error (death or injury) create asymmetric liability. Public safety also typically requires human accountability and authority, limiting full automation even where technical performance might permit it.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement for the relay function itself, but public-safety liability, need for verified accuracy, and inter-agency protocol requirements create real friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integrating AI into a 911 dispatch system requires significant infrastructure investment, continuous human oversight, and redundancy for safety-critical operations. The loaded wage of a dispatcher is modest, and the all-in cost of AI (development, integration, liability insurance, human fallback) currently exceeds what a dedicated dispatcher costs.
Cost vs. human wageclaude-sonnet-53/5Software-based relay/dispatch integration can be cheaper than added staffing for routine handoffs, but liability and system integration costs keep the ratio only moderately favorable.
Technical feasibility todayclaude-haiku-4-5-202510012/5Some AI dispatch-support tools exist to assist with call transcription and initial categorization, but no mature product reliably handles the full intake-to-dispatch workflow independently. Deployed systems still depend on human operators to verify critical details, make final routing decisions, and handle edge cases or emergencies, so production reliability remains limited.
Technical feasibility todayclaude-sonnet-53/5CAD (computer-aided dispatch) systems and interoperable radio/text relay tools already exist and are used in many jurisdictions, though full reliability across agencies and edge cases is inconsistent.

Evaluate complaint and emergency-request information to determine response requirements.

25

CI 25–25 · exposure 25 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Police dispatch centers adopt AI tools incrementally for data entry and call classification assistance, but true autonomous evaluation of response requirements remains rare in production. Most adoption is augmentative rather than replacement-driven, reflecting organizational conservatism and risk aversion in public safety.
Sector adoption velocityclaude-sonnet-52/5Public safety/law enforcement is a traditionally slow-adopting sector for AI in operational decision-making, with pilots for call analytics but limited scaled deployment for autonomous decision-making.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist dispatchers by automatically extracting key information, suggesting response codes, and flagging high-priority calls, allowing dispatchers to make faster, more consistent decisions. The human dispatcher remains the decision-maker, with AI providing structured support for triage.
Augmentation potentialclaude-sonnet-53/5AI can help summarize incoming information, flag keywords, and suggest priority levels, assisting dispatchers/officers, though it doesn't yet transform the underlying judgment task.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with initial triage and categorization of complaints (risk assessment, routing), but the task requires contextual judgment about threat level, officer safety, and situational nuance that current systems cannot reliably handle end-to-end. Human review of final routing decisions is standard practice.
Task automatabilityclaude-sonnet-52/5Triage of complaint/emergency data involves rapid contextual judgment, ambiguous or incomplete information, and life-safety stakes that current AI cannot reliably resolve end-to-end without human dispatcher oversight.'
Adoption barriersclaude-haiku-4-5-202510014/5Significant legal and liability barriers exist: dispatchers are often legally accountable for response decisions, and incorrect triage can result in officer safety failures or civil liability. Union agreements, radio licensing rules, and public-safety regulatory frameworks also create friction against full automation.
Adoption barriersclaude-sonnet-54/5Emergency response determinations carry major liability and public-safety risk, and most jurisdictions require certified dispatchers/officers to make or confirm response-level decisions, creating strong regulatory and organizational barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI triage systems have upfront deployment costs, but dispatchers remain essential for judgment calls and liability reasons. The cost per evaluation is comparable to or exceeds human dispatcher wages when integration, monitoring, and required human oversight are factored in.
Cost vs. human wageclaude-sonnet-52/5AI triage tools require significant integration with legacy CAD/emergency infrastructure and mandatory human oversight, so total cost savings versus trained dispatchers/officers are modest, not order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510012/5While some dispatch AI systems exist for initial call classification and routing, they typically operate with human dispatchers retaining final authority due to high error cost. No mature product fully autonomously evaluates and determines response requirements without human oversight in production settings.
Technical feasibility todayclaude-sonnet-52/5Some CAD systems use AI-assisted call classification and priority scoring, but human dispatchers still make final determinations in virtually all production 911/emergency systems today.

Review facts of incidents to determine if criminal act or statute violations were involved.

23

CI 20–25 · exposure 25 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Law enforcement adoption of AI for legal determinations remains cautious and pilot-stage; agencies tend to use AI for supporting tasks (document review, pattern flagging) rather than replacing the core officer judgment on criminal charges.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a slow-adopting sector for AI in substantive decision-making due to legal, political, and liability concerns, despite growing use of AI in report writing and analytics.
Augmentation potentialclaude-haiku-4-5-202510013/5Current AI can assist meaningfully by organizing incident facts, suggesting relevant statutes, and highlighting inconsistencies, improving officer efficiency and reducing research time. However, the officer remains responsible for final determination.
Augmentation potentialclaude-sonnet-53/5AI can help officers quickly cross-reference statutes, summarize incident facts, and suggest possible charges, improving efficiency while the officer retains final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5Incident review requires nuanced legal judgment, context interpretation, and case-law knowledge. While AI can assist in flagging potential statute matches or organizing facts, the determination of whether a criminal act occurred demands human legal reasoning and discretion that current systems cannot reliably perform end-to-end.
Task automatabilityclaude-sonnet-52/5Determining whether facts constitute a criminal act requires legal judgment, contextual interpretation, and accountability that current AI cannot reliably replicate end-to-end, though it can assist in organizing facts against statutory elements.ract
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and organizational barriers exist: officers must be trained and authorized, liability for incorrect criminal determination rests with the department, prosecutorial standards require qualified personnel review, and civil-rights exposure creates institutional friction against full automation of this gatekeeping function.
Adoption barriersclaude-sonnet-55/5Determining criminal violations is a core law enforcement/legal function with strict authorization, chain-of-custody, and liability requirements that legally require sworn officer judgment and accountability.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI document-review and legal-lookup tools carry significant integration and oversight costs, and the loaded wage of a trained officer remains lower than the all-in cost of reliable AI systems plus mandatory human review and validation.
Cost vs. human wageclaude-sonnet-52/5AI tools could cheaply assist with statute lookup, but the human officer's judgment, liability, and final determination remain necessary, keeping overall cost savings limited.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed police product reliably makes independent criminal/statute determinations at scale. Tools exist for statute matching and document review assistance, but these require expert human validation; no production system substitutes for an officer's judgment without substantial oversight.
Technical feasibility todayclaude-sonnet-52/5Legal research and document analysis tools exist and can flag potential statute matches, but no deployed product independently makes reliable criminal determinations in live police workflows at scale.

Photograph or draw diagrams of crime or accident scenes and interview principals and eyewitnesses.

18

CI 11–25 · exposure 20 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Law enforcement has slowly adopted body cameras and documentation tools, but AI-driven investigation remains limited to evidence processing and administrative support. Core interviewing and witness assessment remain human-centric due to legal, accountability, and community trust requirements.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a traditionally slow-adopting, highly regulated sector with limited AI deployment beyond back-office analytics and evidence management, not field interviewing or scene documentation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI effectively augments this task through automated scene documentation (photo stitching, diagram generation), interview transcription and keyword extraction, and witness statement organization, allowing officers to focus on interpretation and follow-up questioning without replacing the investigative role.
Augmentation potentialclaude-sonnet-53/5AI can assist with 3D scene reconstruction, transcription of interviews, and report generation, meaningfully aiding officers even though it doesn't replace the on-scene judgment and interpersonal interview process.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with photographing scenes (automated documentation) and transcribing witness interviews, conducting investigative interviews and interpreting eyewitness credibility require human judgment, emotional intelligence, and legal safeguards. The task cannot achieve 50% time savings end-to-end without human investigator involvement in the critical interview and assessment phases.
Task automatabilityclaude-sonnet-52/5Photography and basic diagramming could be assisted by AI tools, but conducting witness interviews requires real-time human judgment, rapport-building, and adaptive questioning that current AI cannot perform in the field.'
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and liability barriers exist: investigative interviews must be conducted by law enforcement with proper Miranda warnings, chain-of-custody requirements, and potential courtroom testimony. Witness credibility assessment carries evidentiary weight requiring human professional judgment and legal accountability.
Adoption barriersclaude-sonnet-55/5Evidence collection and witness interviews are legally required to be performed by authorized law enforcement personnel for chain-of-custody and admissibility, creating hard legal and procedural barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI can reduce costs for documentation (photography, diagram generation, transcription), but human investigators remain essential and their salaries dominate task cost. The AI component savings are modest relative to a fully-loaded patrol officer's wage.
Cost vs. human wageclaude-sonnet-51/5The interviewing and physical scene work still require an on-site sworn officer, so AI cannot substitute at lower cost for the full task, only marginally speed up sub-components like report drafting.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed products exist for scene photography documentation and interview transcription, but no production system reliably handles the full investigative interview, witness credibility assessment, and legal chain-of-custody integration that defines this task. Current AI lacks the capability to independently conduct legally sound witness interviews.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously conducts crime scene interviews or fully replaces officer-led scene documentation; some AI-assisted diagramming/photogrammetry tools exist but are narrow and human-operated.

Process prisoners, and prepare and maintain records of prisoner bookings and prisoner status during booking and pre-trial process.

17

CI 14–20 · exposure 25 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Police departments operate under rigid procedural and liability constraints with strong organizational conservatism around prisoner handling. Even with digital systems in place, actual automation of booking tasks is minimal; human officers remain the gatekeepers by law and practice.
Sector adoption velocityclaude-sonnet-52/5Law enforcement and corrections are historically slow to adopt new technology due to legal, budgetary, and procedural constraints, with digitization of booking systems proceeding unevenly across jurisdictions.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by auto-filling routine fields, flagging missing information, or suggesting charge categories—moderately useful aids to officer productivity—but the core verification, custody decisions, and legal sign-off remain human-centered.
Augmentation potentialclaude-sonnet-53/5Digital booking systems, automated form-filling, and record management software meaningfully speed up documentation and reduce paperwork errors, aiding officers without replacing their judgment or authority.
Task automatabilityclaude-haiku-4-5-202510012/5While some sub-tasks like data entry and record-keeping could be partially automated, the task requires significant human judgment, legal compliance, identity verification, and interaction with the prisoner—elements that current AI systems cannot reliably handle end-to-end. Critical custody, safety, and constitutional requirements prevent meaningful time savings at equal quality.
Task automatabilityclaude-sonnet-52/5Parts of the record-keeping (data entry, form population, status logging) could be automated, but the physical processing of prisoners, verification, and legal chain-of-custody documentation require human judgment and physical presence., limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510015/5Prisoner processing is heavily regulated under constitutional law (Miranda, Fourth Amendment), state criminal procedure codes, and departmental policy. A human officer must legally perform, verify, and sign off on booking records; liability for errors in custody documentation creates hard legal and organizational barriers.
Adoption barriersclaude-sonnet-55/5Prisoner processing involves chain-of-custody, constitutional rights, and legal authority that only sworn law enforcement personnel can exercise, making this a hard legal barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI infrastructure, integration, legal compliance review, and mandatory human supervision for prisoner processing would exceed the wage cost of a trained booking officer, especially given liability and error-cost asymmetries in law enforcement.
Cost vs. human wageclaude-sonnet-52/5Booking software reduces clerical time but still requires sworn officers or trained staff to conduct identity verification, custody transfer, and legal documentation, so cost savings are moderate rather than transformative.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system reliably handles the full prisoner processing workflow today. Fragmentary tools exist for data entry and basic information management, but the legal, custodial, and verification requirements mean no production system performs this task reliably without extensive human oversight.
Technical feasibility todayclaude-sonnet-52/5Some jail management systems use digital booking software and biometric capture, but these are tools assisting officers rather than autonomous AI systems performing the booking process reliably without human control.

Monitor, note, report, and investigate suspicious persons and situations, safety hazards, and unusual or illegal activity in patrol area.

14

CI 7–20 · exposure 17 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While some police departments use predictive analytics and CCTV, widespread automation of the investigative and judgment components of patrol remains limited; adoption is pilot-stage rather than production-scale displacement of officer patrol roles.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a traditionally slow-adopting, heavily regulated public-sector field; AI tools (predictive policing, camera analytics) are used but full-task automation is not being pursued or piloted at scale.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist patrol officers by flagging suspicious patterns in surveillance data, providing real-time analytical support, and prioritizing high-risk areas, raising officer effectiveness in prioritizing where to direct attention and investigation.
Augmentation potentialclaude-sonnet-53/5AI-enabled surveillance cameras, license plate recognition, and report-drafting/transcription tools can help officers notice patterns and document incidents faster, improving productivity without replacing patrol judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with pattern recognition and data analysis (e.g., flagging unusual activity from camera feeds or dispatch records), but the task requires real-time judgment, physical presence, contextual understanding of situations, and legal authority to investigate—capabilities that current AI systems cannot replicate end-to-end with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-51/5Core task requires physical presence, real-time perception, judgment under uncertainty, and legal authority to intervene; current AI cannot patrol, observe in-person, or exercise discretion in the field.
Adoption barriersclaude-haiku-4-5-202510015/5Significant legal and institutional barriers exist: only licensed law enforcement officers have authority to stop, question, and investigate suspicious activity; liability for false accusations or improper investigation is high; and public accountability and human judgment are legally mandated components of lawful policing.
Adoption barriersclaude-sonnet-55/5Patrol duties involve sworn legal authority, use-of-force decisions, and constitutional/due-process obligations that require a licensed, accountable human officer.
Cost vs. human wageclaude-haiku-4-5-202510012/5Surveillance and analytics tools are non-trivial capital expenses, and human patrol officers remain cheaper to deploy for the integrated monitoring, judgment, and legal authority functions required; AI cost-benefit is marginal at best.
Cost vs. human wageclaude-sonnet-51/5Replacing a mobile, decision-making patrol officer with AI is not currently feasible at any cost; supplementary sensor systems add cost rather than replacing the officer's role.
Technical feasibility todayclaude-haiku-4-5-202510012/5While CCTV monitoring systems and crime prediction algorithms exist in some deployments, no current product reliably performs the full investigative and judgment component of this task autonomously; most implementations require substantial human oversight and validation.
Technical feasibility todayclaude-sonnet-52/5Some deployed tools (camera analytics, license plate readers, gunshot detection) assist surveillance in fixed areas, but no product autonomously patrols and investigates suspicious activity like an officer.

Monitor traffic to ensure motorists observe traffic regulations and exhibit safe driving procedures.

13

CI 5–20 · exposure 17 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite decades of technology availability, roving traffic patrol remains staffed by human officers. Adoption of automated enforcement is limited to fixed cameras in narrow jurisdictions, reflecting strong organizational, legal, and public resistance to AI-driven traffic stops.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a traditionally slow-adopting sector for full automation of enforcement actions, though camera-based monitoring tools have seen gradual uptake in specific niches like speed and red-light enforcement.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist through dashcam analysis, speed-violation alerts, and vehicle-data parsing to highlight violations or risk patterns, improving an officer's situational awareness during patrols. However, the assistance is partial; human judgment on enforcement discretion and safety remains essential.
Augmentation potentialclaude-sonnet-53/5AI-enabled cameras, license plate readers, and predictive analytics can help officers identify violations and prioritize patrols, offering meaningful but partial productivity assistance.
Task automatabilityclaude-haiku-4-5-202510012/5Traffic monitoring requires real-time detection of violations and context-dependent judgment about driving safety. While AI can assist with detection (speed enforcement cameras, lane-keeping analysis), end-to-end enforcement—including discretionary stops, verbal interaction, and situation assessment—remains far outside current automation capability and would require solving autonomous presence in traffic, which is not deployed.
Task automatabilityclaude-sonnet-51/5Traffic monitoring by patrol officers requires physical presence, situational judgment, pursuit capability, and legal authority to stop and cite drivers—none of which current AI systems can perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Traffic enforcement is a core sworn law enforcement function. Officers must be licensed by the state, the act of stopping vehicles carries liability and warrant requirements, and ticket issuance is a legal act requiring human judgment and accountability. These hard legal and liability barriers prevent substitution.
Adoption barriersclaude-sonnet-55/5Traffic stops, citations, and enforcement actions require sworn, legally authorized officers with arrest powers; this is a heavily regulated public safety function with strict human authority requirements.
Cost vs. human wageclaude-haiku-4-5-202510011/5Fixed traffic cameras (one narrow component) are capital-intensive with high false-positive rates requiring human adjudication. The full monitoring task—roving patrols, enforcement decisions, community interaction—has not been costed at scale against patrol officers because deployment does not exist, making AI substantially more expensive in practice.
Cost vs. human wageclaude-sonnet-52/5Fixed camera systems can be cheaper per violation captured, but they require legal infrastructure, human review, and cannot substitute for the full range of officer duties, so overall cost comparison favors AI only in narrow slices.
Technical feasibility todayclaude-haiku-4-5-202510012/5Limited commercial systems exist: fixed speed cameras and red-light detection work narrowly in controlled settings, but roving patrol monitoring (the core of this task) is not demonstrably performed by any deployed product. Human officers remain the standard; no production system replaces on-road traffic law enforcement.
Technical feasibility todayclaude-sonnet-52/5Automated traffic cameras and speed/red-light enforcement systems exist and are deployed, but they only cover narrow violation types and cannot replace the officer's active patrol, judgment, and interception role.

Inform citizens of community services and recommend options to facilitate longer-term problem resolution.

13

CI 0–25 · exposure 13 · augmentation 38 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement agencies are laggards in AI adoption for citizen-facing service tasks; most remain reliant on traditional patrol officer training and community policing models. There is minimal evidence of production deployment of AI systems for this specific counseling and recommendation function.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a relatively slow-adopting sector for AI in citizen-facing interactions, with pilots for administrative tasks but little production use for direct community engagement recommendations.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist by providing quick access to curated lists of community services or eligibility screeners that an officer reviews before speaking with a citizen, but the core work—assessing needs, building rapport, and recommending tailored solutions—remains fundamentally human.
Augmentation potentialclaude-sonnet-53/5AI can help officers by providing quick access to updated databases of community resources, social services, and referral options to inform their recommendations, improving efficiency without replacing the human interaction.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires nuanced human judgment, empathy, and contextual understanding of individual circumstances to recommend appropriate community services. Current AI systems lack the relational trust, local knowledge integration, and ability to adapt to complex social situations that citizens expect and need from law enforcement interactions.
Task automatabilityclaude-sonnet-52/5While providing basic information about community services could be automated (e.g., via chatbots), the judgment-based recommendation of appropriate options based on situational assessment requires human contextual understanding that current AI cannot reliably replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Strong barriers exist: citizens expect and in many cases require face-to-face interaction with a human officer; legal accountability for referrals rests with the officer; and trust-building is essential to this function. Regulatory expectations and community standards strongly favor human judgment in sensitive social service recommendations.
Adoption barriersclaude-sonnet-54/5This task is embedded in policing duties requiring sworn authority, situational judgment, and accountability; liability and public trust concerns mean a human officer must be the one engaging with citizens even if AI assists behind the scenes.
Cost vs. human wageclaude-haiku-4-5-202510011/5Implementing AI for this task would require significant integration with local service databases, continuous oversight by human officers to verify recommendations, and liability management—making the total cost comparable to or exceeding direct human performance.
Cost vs. human wageclaude-sonnet-52/5AI information lookup tools are cheap, but the overall task requires an officer present anyway for the interaction, so there's minimal marginal cost savings from adding AI to this specific sub-task.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product reliably performs this task end-to-end in production at law enforcement agencies. While chatbots can provide generic information about services, they cannot assess individual needs, build trust, or make contextually appropriate recommendations the way patrol officers do.
Technical feasibility todayclaude-sonnet-52/5Some jurisdictions use chatbots or referral databases for general public information, but no deployed product performs the officer's contextual assessment and personalized recommendation function in the field.

Conduct community programs for all ages concerning topics such as drugs and violence.

13

CI 0–25 · exposure 8 · augmentation 38 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Police departments are low-digitization, risk-averse institutions with entrenched human-centered processes; adoption of AI for community engagement remains negligible and faces cultural and legal resistance.
Sector adoption velocityclaude-sonnet-52/5Law enforcement and community policing functions show slow, cautious AI adoption, mostly limited to administrative support rather than public-facing engagement tasks.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist with preparing educational materials or data visualization on drug/violence statistics, but offers minimal productivity gain for the core task of delivering an engaging, trusted community program.
Augmentation potentialclaude-sonnet-53/5AI can assist officers by generating presentation content, translating materials, and tailoring messaging for different age groups, improving prep efficiency while humans still deliver the sessions.
Task automatabilityclaude-haiku-4-5-202510011/5Conducting community programs requires live interaction, relationship-building, trust-establishment, and real-time adaptation to diverse audience responses—capabilities that current AI systems cannot perform end-to-end in meaningful ways today.
Task automatabilityclaude-sonnet-52/5AI can help draft curricula and materials, but live delivery of community engagement programs requires human presence, adaptability, and trust-building that current AI cannot replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Strong legal and institutional barriers exist: only authorized law enforcement personnel can credibly speak on behalf of police departments; community trust, liability, and organizational policy require human officers to conduct these programs.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement for AI to assist content creation, but community trust, public safety messaging authority, and department policy favor sworn officers delivering these programs.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of deploying an AI system with sufficient reliability and integration overhead would far exceed the cost of a police officer delivering a program, especially given liability and engagement requirements.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate content and materials, but the actual program delivery still requires paid officer time, so overall cost savings are limited.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI products reliably conduct live community education programs for all ages; this task requires human presence, authority, and social presence that AI cannot authentically substitute.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product independently runs community outreach programs; this remains a human-delivered, in-person function.

Investigate traffic accidents and other accidents to determine causes and to determine if a crime has been committed.

10

CI 0–20 · exposure 13 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Law enforcement adoption of AI for investigative tasks remains slow and cautious, largely limited to data analytics and evidence management pilots. Cultural resistance and legal compliance concerns in this highly regulated sector limit production deployments.
Sector adoption velocityclaude-sonnet-51/5Law enforcement is a slow-adopting, physically grounded sector with minimal AI deployment in core investigative fieldwork.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by organizing scene documentation, flagging inconsistencies in witness statements, and surfacing relevant precedent cases, thereby helping investigators work more efficiently. However, the scope of assistance remains bounded by the human's decision-making role.
Augmentation potentialclaude-sonnet-53/5AI can help with accident reconstruction modeling, report writing, and data lookup, providing moderate assistance while the officer remains central to on-scene judgment.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI can assist with evidence synthesis, documentation, and initial scene analysis from photos/video, but cannot independently investigate causation or determine criminality. The task requires judgment calls about human intent, liability, and complex causal reasoning that still demands human investigators.
Task automatabilityclaude-sonnet-51/5Investigating accidents requires physical scene presence, interviewing witnesses, collecting physical evidence, and exercising legal judgment—none of which current AI can perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Legal liability and criminal procedure standards require a licensed officer to conduct the investigation, make determinations of fault, and potentially arrest suspects. Chain-of-custody rules and court admissibility create hard regulatory barriers to autonomous investigation.
Adoption barriersclaude-sonnet-55/5Only sworn, legally authorized officers can conduct accident investigations, gather evidence with chain-of-custody integrity, and make probable-cause determinations, making this a hard legal barrier.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for accident scene analysis and evidence processing cost significant amounts to integrate and maintain, but still require substantial human investigator oversight. The all-in cost per investigation resolved remains higher than for human-led investigation.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this task, so cost comparison favors the human officer by default.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tools exist for evidence review and scene reconstruction aids, no deployed product reliably performs the core investigative judgment (determining cause and criminality) end-to-end. Most deployments remain advisory rather than autonomous.
Technical feasibility todayclaude-sonnet-51/5No deployed product conducts autonomous accident investigations or determines criminal culpability; AI tools at best assist with report drafting or diagramming.

Question individuals entering secured areas to determine their business, directing and rerouting individuals as necessary.

9

CI 0–18 · exposure 13 · augmentation 25 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Public-sector law enforcement agencies adopt automation slowly and conservatively. This task involves interpersonal judgment and security decisions that agencies have shown little inclination to delegate to AI, with most deployment remaining in low-risk, structured settings like badge scanning rather than behavioral assessment.
Sector adoption velocityclaude-sonnet-51/5Policing is a slow-adopting, physically-grounded sector with minimal AI agent deployment for direct citizen interaction and security screening duties.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist with identity verification (facial recognition, credential checking) and alert officers to anomalies, but the core task—questioning and deciding whether to route or reroute individuals—remains fundamentally officer-dependent. The augmentation value is limited to peripheral support.
Augmentation potentialclaude-sonnet-52/5AI can assist with access logs, facial recognition alerts, or credential verification support, but offers limited direct augmentation to the interactive questioning and judgment call itself.
Task automatabilityclaude-haiku-4-5-202510012/5While AI systems can recognize faces and analyze access credentials, the nuanced judgment required to assess intent, detect deception, and make real-time security decisions about directing or rerouting individuals cannot be reliably automated today. The task requires contextual reasoning and dynamic interaction that current AI cannot replicate at a 50% time-saving threshold with equal quality.
Task automatabilityclaude-sonnet-51/5This requires real-time physical presence, verbal interaction, judgment about intent/deception, and enforcement authority that current AI systems cannot perform end-to-end in the physical world.
Adoption barriersclaude-haiku-4-5-202510015/5This task has strong legal and regulatory barriers: patrol officers must be licensed, sworn personnel with authority to enforce access control and security protocols. Liability for security decisions, criminal authority, and organizational policy typically require a human officer to legally perform or sign off on access decisions.
Adoption barriersclaude-sonnet-55/5This is a core law-enforcement duty requiring sworn authority, use-of-force potential, and legal accountability, making it one of the most protected tasks against automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5The infrastructure, computing, and human oversight required to deploy AI for this task would likely exceed the cost of a patrol officer's time, especially when error costs (security breaches, false denials) are factored in. The all-in cost remains uncompetitive.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this exact task, so AI cost is not comparable; any partial tech (turnstiles, cameras) requires human backup, making all-in cost higher than a single officer for this function.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs this task end-to-end. AI access-control systems exist but typically operate in narrow, controlled settings (badge readers, turnstiles) and cannot substitute for the human judgment and presence required in this interpersonal, security-critical task.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously questions and reroutes individuals at secured areas with police authority; access control tech (badges, cameras) exists but does not replace the interactive judgment task itself.

Identify, pursue, and arrest suspects and perpetrators of criminal acts.

3

CI 0–5 · exposure 5 · augmentation 38 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement adoption of AI for arrest decisions remains minimal and heavily restricted. Most deployments are pilot-stage and narrowly scoped (database searches, dispatch optimization), with significant institutional and legal resistance to autonomous suspect apprehension.
Sector adoption velocityclaude-sonnet-51/5Law enforcement is a low-digitization, physically embedded sector with minimal deployment of autonomous agents for arrest-related functions; adoption is confined to investigative tools, not action-taking.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist with pattern analysis and information retrieval (suspect databases, contextual data), but cannot meaningfully augment the core judgment and physical execution of pursuit and arrest, which remain fundamentally human-dependent.
Augmentation potentialclaude-sonnet-53/5AI can assist via facial recognition, predictive analytics, license plate readers, and dispatch support to help identify and locate suspects, improving officer efficiency without replacing the physical task.
Task automatabilityclaude-haiku-4-5-202510011/5This task involves real-time decision-making in high-risk, dynamic physical environments requiring immediate judgment calls about threat assessment, use of force, and suspect interaction. Current AI systems cannot reliably perform end-to-end suspect identification, pursuit, and arrest in the field without constant human oversight and intervention.
Task automatabilityclaude-sonnet-51/5This task requires physical presence, real-time judgment, use of force decisions, and physical apprehension of suspects—none of which current AI systems can perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5This task has multiple hard legal barriers: only sworn law enforcement officers are authorized to arrest, use force is heavily regulated, liability for wrongful arrest/injury is severe, and judicial oversight of arrest procedures is mandated. These requirements cannot be bypassed by automation.
Adoption barriersclaude-sonnet-55/5Arrest authority is a legally licensed police power requiring sworn officers, use-of-force accountability, and legal chain of custody—hard regulatory and liability barriers prevent automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure required for reliable outdoor pursuit, threat assessment, and safe arrest procedures (combined computer vision, robotics, liability) far exceeds the cost of a patrol officer's salary. Integration and oversight costs are prohibitively high relative to the task's value.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute capable of performing physical pursuit and arrest, so cost comparison is inapplicable/AI is far more expensive in effect since it cannot deliver the outcome at all.
Technical feasibility todayclaude-haiku-4-5-202510012/5While narrow components exist (facial recognition, predictive policing), no deployed product reliably performs the full task of identifying, pursuing, and arresting suspects in production. Existing tools have documented high error rates, legal challenges, and require significant human verification and decision-making.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs identification, pursuit, and physical arrest of suspects; AI is at most used for peripheral surveillance or facial recognition leads, not the task itself.

Investigate illegal or suspicious activities.

3

CI 3–3 · exposure 0 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Law enforcement agencies are exploring AI for data analysis, facial recognition, and case management assistance, but core investigative authority remains centralized in human officers. Adoption of full investigation automation is negligible due to legal and accountability constraints.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is adopting AI tools like predictive analytics and license-plate readers, but core investigative fieldwork remains largely untouched by automation.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist investigators by analyzing evidence databases, flagging patterns, organizing case files, and suggesting leads, which can improve investigator productivity. However, the human officer must retain control of the investigation and decision-making process itself.
Augmentation potentialclaude-sonnet-53/5AI can assist with data analysis, pattern detection, report drafting, and surveillance footage review, aiding officers without replacing the investigative task itself.
Task automatabilityclaude-haiku-4-5-202510011/5Investigating suspicious activities requires human judgment, legal authority, witness interaction, scene assessment, and discretionary decision-making that AI cannot currently perform end-to-end. Current AI lacks the contextual understanding, authority to make investigative decisions, and ability to handle real-world ambiguity at the quality and scope required.
Task automatabilityclaude-sonnet-51/5Investigating suspicious activity requires physical presence, real-time judgment, confrontation, and legal authority that current AI cannot replicate end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Only licensed law enforcement officers with legal authority can legally investigate crimes and make enforcement decisions. Investigation is embedded in statutory authority, liability frameworks, and constitutional protections that mandate human judgment and accountability in law enforcement.
Adoption barriersclaude-sonnet-55/5Investigations require sworn, legally authorized officers with arrest powers, chain-of-custody responsibilities, and accountability structures that cannot be delegated to AI.
Cost vs. human wageclaude-haiku-4-5-202510011/5Automating investigation would still require significant human oversight, legal compliance infrastructure, and ongoing verification. The marginal cost of AI assistance does not yet approach orders of magnitude savings compared to trained officers who are legally required to perform this function.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this task, so no meaningful cost comparison favors AI; human officers remain the only viable option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs criminal investigation end-to-end. AI can assist with data analysis or pattern detection, but investigation fundamentally requires human officers with legal standing, the ability to interview witnesses, make arrests, and exercise sworn authority.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently investigates crimes or suspicious activity; AI is at most used for tips analysis or pattern flagging, not the investigative task itself.

Provide for public safety by maintaining order, responding to emergencies, protecting people and property, enforcing motor vehicle and criminal laws, and promoting good community relations.

1

CI 0–3 · exposure 0 · augmentation 38 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of full automation is essentially zero; police departments remain primarily human-staffed. While some AI tools (predictive dispatch, license-plate readers, data analysis) are in use, these augment rather than replace patrol and enforcement. Organizational and legal resistance remains very high.
Sector adoption velocityclaude-sonnet-52/5Law enforcement is a slow-adopting, highly regulated, physically-grounded sector with limited AI deployment beyond back-office analytics.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist with dispatch optimization, crime prediction, and data analysis to help officers allocate time, but offers limited direct assistance during the core tasks of patrol, emergency response, enforcement, and community interaction. Augmentation is marginal relative to the full scope of the task.
Augmentation potentialclaude-sonnet-53/5AI tools (predictive analytics, license plate readers, report drafting, dispatch optimization) assist officers in parts of their work without touching the core enforcement/safety task.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical presence, real-time decision-making in unpredictable environments, de-escalation, use of force judgment, and community interaction—none of which AI can perform end-to-end today. Current systems cannot substitute for patrol, emergency response, or law enforcement presence.
Task automatabilityclaude-sonnet-51/5This is a broad, physical, real-world task requiring judgment, use of force decisions, and physical presence; current AI cannot perform patrol, arrests, or emergency response.
Adoption barriersclaude-haiku-4-5-202510015/5Severe legal and regulatory barriers exist: police authority to detain, search, and use force is strictly licensed to human officers; liability for errors is substantial and asymmetric; state and local law mandate human law enforcement; and public safety depends on human judgment and accountability in life-critical situations.
Adoption barriersclaude-sonnet-55/5Policing requires sworn, licensed authority, legal use-of-force powers, and accountability structures that legally must reside with a human officer.
Cost vs. human wageclaude-haiku-4-5-202510011/5The task requires armed, trained human officers with liability coverage, vehicles, and real-time physical presence. Even if narrow subtasks (dispatch optimization) were automated, the core patrol and enforcement functions cannot be meaningfully replaced by AI at lower cost than human labor.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical, authority-laden task, so cost comparison favors humans entirely.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product can reliably perform patrol, emergency response, or law enforcement duties independently. While AI assists with dispatch prediction and data analysis, no production system performs the core task of providing public safety through presence and enforcement.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously performs patrol officer duties; AI is at most used for dispatch support or analytics, not the core task.

Render aid to accident survivors and other persons requiring first aid for physical injuries.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of AI automation for this task is near-zero because it requires physical presence and human judgment in high-stakes, variable emergency contexts. No sector is moving toward autonomous first-aid provision.
Sector adoption velocityclaude-sonnet-51/5Law enforcement field response to physical emergencies has essentially zero AI/robotic displacement in practice; adoption is confined to administrative or analytic tasks, not physical first aid.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could marginally assist a responder via real-time guidance (e.g., protocol reminders or symptom checklist support), but the task is predominantly physical and judgment-driven, limiting meaningful augmentation potential. Such tools exist but are not transformative.
Augmentation potentialclaude-sonnet-52/5AI could offer minor support (e.g., dispatch guidance, medical instruction apps, or wearable sensors relaying vitals) but does not materially transform the hands-on task itself.
Task automatabilityclaude-haiku-4-5-202510011/5Rendering first aid to accident survivors requires physical intervention, situational assessment, and direct human contact that current AI systems cannot perform. Even tool-using agents cannot physically apply bandages, perform CPR, stabilize fractures, or make real-time clinical decisions at the scene.
Task automatabilityclaude-sonnet-51/5Rendering physical first aid requires hands-on manipulation of a person's body in unpredictable, urgent environments—something no current AI system can perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5First aid provision has strong legal barriers: only trained and certified humans can legally render aid in most jurisdictions, liability for incorrect treatment is severe, and direct physical human contact is non-negotiable. Regulatory and professional licensing protections are substantial.
Adoption barriersclaude-sonnet-55/5Physical intervention on injured persons is core to sworn officer duty and often legally/organizationally mandated, with liability and public-safety expectations requiring a human physically present.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems cannot replace the cost of a trained human first responder at the scene. The human wage is low relative to any AI infrastructure that might guide or supplement, making economic replacement infeasible.
Cost vs. human wageclaude-sonnet-51/5No AI system can substitute for the physical act of aid, so cost comparison is moot; a human officer is the only viable option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can autonomously perform first aid. While AI may assist in triage or guidance via chatbots, the task itself—hands-on medical intervention—remains entirely dependent on human responders with physical presence.
Technical feasibility todayclaude-sonnet-51/5There are no deployed products that physically render first aid to injured persons; this remains purely a human physical-action task.

Testify in court to present evidence or act as witness in traffic and criminal cases.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5There is no adoption because this task is legally and constitutionally mandated to be performed by humans. No sector can or will attempt to automate courtroom testimony.
Sector adoption velocityclaude-sonnet-51/5Law enforcement and judicial systems show minimal movement toward replacing witness testimony with AI, given legal and evidentiary constraints.
Augmentation potentialclaude-haiku-4-5-202510011/5AI cannot meaningfully assist an officer in the act of testifying itself, though AI tools might help prepare case materials beforehand. The testimony itself offers no augmentation opportunity.
Augmentation potentialclaude-sonnet-53/5AI can help officers prepare testimony by organizing case notes, reports, and evidence summaries beforehand, though it does not participate in the testimony itself.
Task automatabilityclaude-haiku-4-5-202510011/5Courtroom testimony requires human presence, legal authority, and the ability to answer unpredictable cross-examination questions—functions that are legally and practically impossible for AI to perform. No AI system can serve as a legal witness or substitute for human testimony.
Task automatabilityclaude-sonnet-51/5Courtroom testimony requires a sworn human witness with firsthand perceptual and experiential knowledge; AI cannot legally or functionally substitute for this act.
Adoption barriersclaude-haiku-4-5-202510015/5This task is protected by hard legal barriers: only a sworn human witness can testify in court, and rules of evidence and criminal procedure explicitly require human testimony and cross-examination. Automation is legally prohibited.
Adoption barriersclaude-sonnet-55/5Testimony is a legally mandated act requiring the sworn, identifiable human officer; courts require personal firsthand testimony and cross-examination of a live witness, an absolute legal barrier.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot perform this task at any cost; the comparison is meaningless. A human officer's testimony is legally required and cannot be replaced by any technology.
Cost vs. human wageclaude-sonnet-51/5There is no AI alternative offering equivalent output, so cost comparison is moot—AI cannot deliver the required output at all.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product can testify in court or serve as a witness in legal proceedings. This is a uniquely human legal function that existing AI systems are fundamentally incapable of performing.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs live witness testimony; this is not a task category any AI product addresses.

Drive vehicles or patrol specific areas to detect law violators, issue citations, and make arrests.

0

CI 0–0 · exposure 0 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement agencies remain highly human-centric in patrol operations; automation pilots are extremely rare and limited to specific, controlled scenarios (parking enforcement only in very few cities), with no significant production displacement of patrol officers.
Sector adoption velocityclaude-sonnet-51/5Law enforcement is a highly physical, low-digitization sector with minimal AI-driven displacement of frontline patrol duties.
Augmentation potentialclaude-haiku-4-5-202510013/5AI systems can assist officers with dispatch routing optimization, real-time criminal database lookups, and predictive analytics on high-crime areas, improving patrol efficiency, but the officer remains essential for enforcement decisions and interactions.
Augmentation potentialclaude-sonnet-53/5AI tools like license plate recognition, predictive patrol routing, and body-cam analytics assist officers, but the core driving/detection/arrest task remains human-performed.
Task automatabilityclaude-haiku-4-5-202510011/5While autonomous vehicles can technically drive, the core task requires real-time judgment to detect violations, decide whether to issue citations, make arrests, and handle unpredictable public interactions—all requiring human authority and contextual discretion that current AI cannot reliably replicate end-to-end.
Task automatabilityclaude-sonnet-51/5Physical patrol, real-time judgment, use of force decisions, and arrests require embodied presence and legal authority that no current AI system can replicate.
Adoption barriersclaude-haiku-4-5-202510015/5Strong legal barriers: only licensed, sworn law enforcement officers are authorized to stop vehicles, issue citations, and make arrests. Public safety and due-process requirements mean a human must retain decision authority and legal accountability.
Adoption barriersclaude-sonnet-55/5Arrests and citations require sworn, legally authorized officers with use-of-force training and accountability; this is a hard legal/licensing barrier.
Cost vs. human wageclaude-haiku-4-5-202510011/5A fully autonomous patrol system would require massive infrastructure, liability insurance, and ongoing remote human oversight, making it substantially more expensive than a patrol officer's loaded cost in current deployments.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this full task, so no meaningful cost comparison exists; human officers remain the only viable option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product performs patrol, violation detection, citation issuance, and arrest authority without human officers. Autonomous vehicles exist in limited domains, but law enforcement tasks remain research-stage and require human operators.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously patrols, detects violations, and makes arrests; ADAS and license-plate readers only assist narrow sub-components.

Patrol and guard courthouses, grand jury rooms, or assigned areas to provide security, enforce laws, maintain order, and arrest violators.

0

CI 0–0 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite pilot programs with security robots in some venues, core patrol and law enforcement functions remain almost entirely human-staffed; adoption of AI for security is minimal and limited to narrow, non-substitutive monitoring roles.
Sector adoption velocityclaude-sonnet-51/5Law enforcement and physical security are slow-adopting sectors for AI-driven task replacement, with technology limited to surveillance support tools rather than patrol/arrest automation.
Augmentation potentialclaude-haiku-4-5-202510012/5While AI tools like analytics dashboards or sensor integration could assist human officers with data or scheduling, the core task of physical patrol and law enforcement involves human judgment and presence that cannot be meaningfully augmented by current AI.
Augmentation potentialclaude-sonnet-53/5AI can assist via surveillance analytics, license plate recognition, or predictive alerts that help officers prioritize patrol routes, but this only supports a fraction of the overall task.
Task automatabilityclaude-haiku-4-5-202510011/5Physical patrol, situational judgment, law enforcement discretion, and arrest authority cannot be automated; the task requires human presence, decision-making in ambiguous situations, and legal authority that cannot be delegated to machines.
Task automatabilityclaude-sonnet-51/5This is a physical presence, judgment, and use-of-force task requiring bodily patrol, real-time threat assessment, and legal authority to arrest; no AI system can perform this end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Law enforcement and courthouse security are heavily regulated; arrest authority, use of force, legal responsibility, and public safety require a licensed human officer with legal standing; substitution is legally prohibited.
Adoption barriersclaude-sonnet-55/5Arrest authority, use of force, and courthouse security are strictly limited to sworn, licensed law enforcement officers under statute, creating hard legal barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5A human patrol officer's loaded cost is already lower than the total infrastructure, liability coverage, and oversight required to deploy autonomous security systems that meet legal and operational standards.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing the full task, so any comparison favors the human officer who can legally and physically perform the entire function.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can autonomously patrol physical spaces, identify legal violations in real-time, make arrest decisions, or maintain courtroom security; these functions require human judgment and legal accountability.
Technical feasibility todayclaude-sonnet-51/5No deployed product patrols, guards, or makes arrests; robotic security patrol tools exist only in narrow, supervised pilot contexts, not as substitutes for sworn officers.

Execute arrest warrants, locating and taking persons into custody.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Warrant execution is deeply embedded in traditional law enforcement structures with minimal digitization and zero demonstrated AI adoption in production; it remains a human-intensive, judgment-heavy activity resistant to technological displacement. No sector data suggests movement toward AI-driven warrant execution.
Sector adoption velocityclaude-sonnet-51/5Law enforcement field operations are a low-digitization, physically-grounded sector with essentially no movement toward automating custodial arrests.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could marginally assist with pre-warrant intelligence gathering, suspect location prediction, or route optimization, but these are support functions, not the core task of apprehension. The human officer remains essential and AI assistance is limited to preliminary planning rather than execution itself.
Augmentation potentialclaude-sonnet-52/5AI can assist with locating suspects via data analysis, facial recognition, or predictive tools, but offers minimal help with the physical act of custody and confrontation itself.
Task automatabilityclaude-haiku-4-5-202510011/5Executing arrest warrants requires physical apprehension of suspects, real-time threat assessment, tactical decision-making, and legal compliance in dynamic, unpredictable environments—capabilities entirely outside current AI systems' scope. No meaningful part of this task can be automated by available technology.
Task automatabilityclaude-sonnet-51/5Physically locating and apprehending a person requires embodied presence, force judgment, and real-time physical action that no AI system can perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Only licensed law enforcement officers with proper legal authority and training are legally permitted to execute arrest warrants and take suspects into custody; this is a hard regulatory and liability barrier that forbids substitution with non-human agents. The human officer's personal judgment and accountability are mandated by law.
Adoption barriersclaude-sonnet-55/5Making arrests requires sworn law enforcement authority, use-of-force legal accountability, and constitutional/due-process safeguards that legally restrict this power to authorized human officers.
Cost vs. human wageclaude-haiku-4-5-202510011/5An armed patrol officer executing warrants has significant training, equipment, and liability costs; no AI system can substitute for these expenses, and the task cannot be partially offloaded to reduce human-equivalent costs. The comparison is not meaningful because AI cannot perform the core function at any cost.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this task, so cost comparison favors the human officer by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs warrant execution or suspect apprehension; the task is fundamentally dependent on embodied human agents capable of physical restraint, judgment, and legal authority. This remains research-stage conceptually and cannot be performed by any commercial system.
Technical feasibility todayclaude-sonnet-51/5No deployed product can physically execute an arrest warrant; this remains entirely research-fiction territory for autonomous systems.

Patrol specific area on foot, horseback, or motorized conveyance, responding promptly to calls for assistance.

0

CI 0–0 · exposure 0 · augmentation 38 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Despite interest in patrol technology and data analytics, actual autonomous physical patrol remains at the experimental fringe; law enforcement agencies retain human officers for frontline response and show no adoption velocity toward AI replacement of patrol itself.
Sector adoption velocityclaude-sonnet-51/5Policing is a physically-grounded, low-digitization sector with minimal AI-driven automation of frontline patrol duties.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can assist with dispatch optimization, crime prediction mapping, and data analysis to help officers prioritize patrols, but these augmentations are peripheral to the core task of physical presence and in-the-moment patrol response.
Augmentation potentialclaude-sonnet-53/5AI aids dispatch optimization, predictive patrol routing, and license plate/camera analytics, but the core foot/vehicle patrol and response remains human-performed.
Task automatabilityclaude-haiku-4-5-202510011/5Foot, horseback, or motorized patrol combined with real-time response to calls requires physical presence, situational judgment, and human discretion in diverse, unstructured environments—capabilities far beyond current AI systems. No automation can substitute for the human officer's responsiveness and decision-making in the field.
Task automatabilityclaude-sonnet-51/5Physical presence, real-time situational judgment, and use of force decisions are required; no AI system can independently patrol areas and respond to calls today.
Adoption barriersclaude-haiku-4-5-202510015/5Police patrol is fundamentally a role requiring a human officer with legal authority to detain, use discretion in force, and be held accountable—regulatory and legal frameworks mandate licensed human presence and decision-making. Liability for autonomous patrol decisions is prohibitive.
Adoption barriersclaude-sonnet-55/5Law enforcement authority, use of force, arrest powers, and legal accountability require a sworn, licensed human officer, creating hard legal and liability barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure cost of autonomous patrol vehicles, sensor systems, and continuous remote oversight would far exceed the loaded cost of human patrol officers, especially given liability exposure and current technology immaturity.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical task, so cost comparison favors the human by default; any robotic equivalent would require expensive hardware and oversight.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product today reliably performs autonomous physical patrol and emergency response; autonomous vehicles exist for controlled environments but not for adaptive, safety-critical police patrol with appropriate judgment and escalation handling.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs autonomous patrol and response in place of a human officer; robotic patrol trials are extremely limited and non-authoritative.

Supervise law enforcement staff, such as jail staff, officers, and deputy sheriffs.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement remains highly resistant to automation of supervisory functions due to accountability, liability, and the necessity of human judgment in personnel and operational decisions. Adoption of AI in supervisory roles is negligible in this sector.
Sector adoption velocityclaude-sonnet-51/5Law enforcement is a slow-adopting, highly regulated, physically-grounded sector with minimal AI integration into command and personnel supervision functions.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could modestly assist with scheduling optimization, performance analytics, or incident documentation review, but augmentation value is limited because supervision fundamentally requires human presence, authority, and real-time decision-making that AI cannot meaningfully enhance.
Augmentation potentialclaude-sonnet-52/5AI can assist with administrative aspects like scheduling, report review, or performance data tracking, but offers little help with the core interpersonal and authority-based supervisory task.
Task automatabilityclaude-haiku-4-5-202510011/5Supervision of law enforcement staff requires real-time judgment calls, personnel management, conflict resolution, and accountability decisions that demand human authority and contextual understanding of complex operational situations. Current AI systems cannot replace the discretionary, interpersonal, and legal authority components inherent in this task.
Task automatabilityclaude-sonnet-51/5Supervising law enforcement personnel requires in-person leadership, real-time judgment about high-stakes situations, and interpersonal authority that current AI cannot replicate or perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Law enforcement supervision is legally and organizationally protected: only authorized sworn personnel can supervise staff, command units, and make disciplinary decisions. Liability, chain-of-command requirements, and regulatory mandates create hard barriers to any form of automation.
Adoption barriersclaude-sonnet-55/5Supervisory authority in law enforcement is tied to sworn rank, legal accountability, chain-of-command structures, and use-of-force oversight responsibilities that legally require a human officer.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI tools for scheduling or analytics may assist supervision, but the core task—managing personnel, making staffing decisions, and providing accountability—cannot be economically automated. The cost of AI infrastructure would far exceed the value of partial support.
Cost vs. human wageclaude-sonnet-51/5There is no AI system performing this supervisory role, so no cost comparison favors AI; a human supervisor's wage is the only viable cost basis.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs law enforcement staff supervision in production. This task involves hiring, disciplinary decisions, shift management, and real-time operational oversight that remain firmly in human domain across all law enforcement agencies.
Technical feasibility todayclaude-sonnet-51/5No deployed product manages or supervises human law enforcement staff; this remains purely a human management function with no AI substitute in production.

Transport or escort prisoners and defendants en route to courtrooms, prisons or jails, attorneys' offices, or medical facilities.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement remains a physically embedded, human-contact-mandatory sector with low AI adoption for core operational tasks. Prisoner transport is a high-liability function unlikely to see rapid automation in practice despite technological possibility.
Sector adoption velocityclaude-sonnet-51/5Law enforcement physical custody functions show essentially no AI adoption; this is a highly physical, low-digitization task within a sector slow to automate core enforcement duties.
Augmentation potentialclaude-haiku-4-5-202510012/5GPS routing and communication tools offer minor assistance, but the core task—physically securing and escorting a detainee—offers limited scope for meaningful AI augmentation while an officer remains in control.
Augmentation potentialclaude-sonnet-52/5AI can assist with route planning, scheduling, or tracking logistics of transports, but offers minimal help with the core physical security and escort function.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical presence, real-time decision-making in unpredictable environments, and active custody/restraint of human subjects. Current AI cannot operate vehicles autonomously in general conditions, manage security protocols, or assume legal custody responsibility.
Task automatabilityclaude-sonnet-51/5This is a physical custody and security task requiring bodily presence, restraint capability, and legal authority; no AI system can perform physical transport or escort duties.
Adoption barriersclaude-haiku-4-5-202510015/5Legal custody of prisoners requires a licensed peace officer; liability for escape or harm falls on the officer in custody. Regulations and common law explicitly require a responsible human authority present, creating a hard legal barrier to full automation.
Adoption barriersclaude-sonnet-55/5Legal custody, chain-of-custody requirements, use-of-force authority, and public safety liability mean only sworn, authorized officers can legally transport prisoners.
Cost vs. human wageclaude-haiku-4-5-202510011/5Autonomous vehicle plus security systems would require significant capital and insurance overhead, while a patrol officer's cost is already embedded in payroll. The liability and specialized equipment make AI substitution more expensive than human labor today.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute for physical transport, so cost comparison is moot—human officers remain the only means of performing this task.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed system performs prisoner transport end-to-end. While autonomous vehicles exist in limited domains, none integrate the security, liability, legal chain-of-custody, and dynamic human-management requirements this task demands.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical prisoner transport or escort; this remains entirely human-executed with no automation in production.

Direct traffic flow and reroute traffic in case of emergencies.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement remains a human-centric, highly regulated sector with minimal AI adoption for front-line patrol and traffic control tasks. Adoption of AI in this domain is in early stages, driven by data analysis rather than operational automation.
Sector adoption velocityclaude-sonnet-51/5Physical, public-safety field work has extremely low AI adoption; law enforcement patrol functions remain almost entirely human-performed with minimal automation in the field.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist minimally via real-time traffic data feeds or route suggestions, but the core task of physically directing traffic and making split-second emergency decisions remains entirely human-dependent, limiting augmentation potential.
Augmentation potentialclaude-sonnet-52/5AI can help with traffic camera monitoring, predictive congestion analysis, or route rerouting suggestions to inform dispatch, but offers little direct assistance to the officer physically directing traffic.
Task automatabilityclaude-haiku-4-5-202510011/5Directing traffic flow requires real-time situational awareness, dynamic decision-making, and physical presence at the scene to communicate with drivers and respond to emergencies. Current AI systems cannot perform these tasks end-to-end, and no autonomous system today can safely manage live traffic without continuous human oversight.
Task automatabilityclaude-sonnet-51/5Directing traffic in real time requires physical presence, hand signals, real-time judgment amid chaotic emergency conditions, and interaction with drivers/pedestrians that current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Traffic direction during emergencies is legally and operationally a police function that requires a licensed officer's authority and judgment. Liability for accidents or misdirection, public safety requirements, and the need for human authority to communicate commands create hard barriers to automation.
Adoption barriersclaude-sonnet-55/5Traffic control during emergencies typically requires a sworn, authorized officer with legal authority to direct vehicles, enforce compliance, and make on-the-spot safety judgments, creating hard legal/licensing barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure, sensors, signage, and integration costs to automate traffic direction would far exceed the cost of a patrol officer performing the task, particularly given the safety-critical nature and need for occasional intervention.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical task, so cost comparison favors the human by default since AI cannot deliver the output at all.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs dynamic traffic direction and emergency rerouting autonomously. While traffic management systems exist, they are passive (signal coordination) rather than active direction, and emergency response requires human judgment and communication that current AI cannot reliably replicate in production.
Technical feasibility todayclaude-sonnet-51/5No deployed product physically directs live traffic or reroutes vehicles during emergencies in place of an officer; existing traffic-management AI only adjusts signals or provides route suggestions, not active scene control.

Notify patrol units to take violators into custody or to provide needed assistance or medical aid.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement agencies are conservative adopters of automation in decision-critical dispatch and custody functions. Pilot programs exist for routing optimization, but autonomous custody notifications remain essentially absent from production systems.
Sector adoption velocityclaude-sonnet-51/5n/a
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist by summarizing incident data or suggesting relevant prior contacts, but the dispatcher's judgment call on whether to order custody versus aid remains central and non-delegable.
Augmentation potentialclaude-sonnet-52/5n/a
Task automatabilityclaude-haiku-4-5-202510011/5This task requires real-time judgment about situational appropriateness, suspect dangerousness, and resource allocation—decisions that depend on contextual factors current AI cannot reliably assess. A human supervisor must evaluate whether custody or assistance is warranted in each unique scenario.
Task automatabilityclaude-sonnet-51/5This requires real-time tactical judgment, radio communication, and command authority during dynamic field situations that current AI cannot perform end-to-end."},"feasibility":{"rating":1,"rationale":"No deployed AI product independently notifies patrol units to take custody actions or dispatch medical aid; this remains a human dispatcher/officer function."},"cost_ratio":{"rating":1,"rationale":"There is no viable AI substitute performing this function, so cost comparison favors humans by default; any AI attempt would require full human oversight anyway."},"barriers":{"rating":5,"rationale":"Only sworn, authorized officers can direct custody actions or use-of-force related dispatch decisions, and legal/liability frameworks require human authority and accountability."},"adoption_velocity":{"rating":1,"rationale":"Law enforcement is a slow-adopting, highly regulated, physically-grounded sector with minimal AI deployment in live tactical decision-making."},"augmentation":{"rating":2,"rationale":"AI-assisted dispatch systems (CAD, transcription, translation aids) can support situational awareness, but the core judgment and notification action remains manual."}}gm/2,
Adoption barriersclaude-haiku-4-5-202510015/5This task is protected by law enforcement hierarchy, civil liability rules, and the requirement that a sworn officer exercise discretion in arrest decisions. Automating custody notifications without human judgment creates legal jeopardy.
Adoption barriersclaude-sonnet-55/5n/a
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI systems to monitor dispatch communications, assess situations, and trigger notifications would exceed the cost of a dispatcher, especially when factoring in legal liability and required human oversight.
Cost vs. human wageclaude-sonnet-51/5n/a
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably makes custody/assistance decisions or notifies patrol units autonomously. Such decisions carry high liability and legal consequences that require human authority and accountability.
Technical feasibility todayclaude-sonnet-51/5n/a

Place people in protective custody.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Police departments continue to rely on human officers for all protective custody operations; there is no meaningful AI adoption or automation in this core law enforcement function.
Sector adoption velocityclaude-sonnet-51/5Law enforcement's physical enforcement functions show minimal AI adoption; this is a core physical/legal duty with no displacement trend.
Augmentation potentialclaude-haiku-4-5-202510012/5AI might assist with risk assessment data or dispatch routing before an officer arrives, but cannot transform the core physical and judgment-intensive task of placing a person in protective custody.
Augmentation potentialclaude-sonnet-52/5AI can assist with related documentation, risk assessment flags, or database checks prior to the decision, but offers little help with the physical act of custody itself.
Task automatabilityclaude-haiku-4-5-202510011/5Placing someone in protective custody requires in-person physical restraint, transport, and real-time judgment about individual safety, mental state, and legal justification—capabilities entirely outside current AI systems' reach. No AI can perform this core operational and legal function.
Task automatabilityclaude-sonnet-51/5Placing a person in protective custody requires physical presence, legal authority, judgment about safety, and direct human interaction; no AI system can perform this act at all today.'
Adoption barriersclaude-haiku-4-5-202510015/5Placing someone in custody is a licensed law enforcement function requiring legal authority, chain-of-custody documentation, and personal accountability; only a sworn officer can lawfully execute this task, creating an absolute legal barrier.
Adoption barriersclaude-sonnet-55/5Taking someone into protective custody is a legally authorized use of police power requiring sworn officer status, statutory authority, and accountability structures that cannot be delegated to software.
Cost vs. human wageclaude-haiku-4-5-202510011/5The task requires human officers to be physically present and legally accountable; AI has no cost advantage when the human officer cost is effectively mandatory.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical, legally-authorized action, so cost comparison is moot—AI cannot deliver the output at any price.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can physically place a person in custody or perform the requisite on-scene assessment and decision-making; this remains fully human-dependent in all deployed systems.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs physical custody or detention of individuals; this remains an inherently physical, legally-authorized human act.

Serve statements of claims, subpoenas, summonses, jury summonses, orders to pay alimony, and other court orders.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement remains a laggard sector for task automation due to physical requirements, legal mandates, and the need for human judgment and authority. No meaningful AI displacement has occurred in document service.
Sector adoption velocityclaude-sonnet-51/5Law enforcement field duties involving physical presence and legal authority show negligible AI adoption for this specific function.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist marginally with route planning, database lookups, or document organization, but the core task—locating and serving an individual—remains fundamentally human-driven with limited augmentation potential.
Augmentation potentialclaude-sonnet-52/5AI could help with logistics like route planning, address verification, or documentation/record-keeping after service, but offers minimal assistance to the core physical task.
Task automatabilityclaude-haiku-4-5-202510011/5Serving legal documents requires physical presence at a specific location, identification of the correct person, and often interaction with potentially hostile or evasive individuals. No current AI system can perform these physical and interpersonal requirements end-to-end.
Task automatabilityclaude-sonnet-51/5Serving legal documents requires physically locating individuals, verifying identity, and completing in-person delivery with legal attestation—no AI system can perform this physical, real-world task.
Adoption barriersclaude-haiku-4-5-202510015/5Legal requirements mandate that a sworn officer or authorized process server physically serve most court documents and witness the delivery. This creates a hard regulatory barrier to full automation.
Adoption barriersclaude-sonnet-55/5Service of legal process typically requires a sworn officer or authorized process server, with strict legal and evidentiary requirements for valid service, creating hard legal barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot perform document service; the comparison is inapplicable. A human officer must be deployed, making any AI-only cost comparison meaningless.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical act of service at all, so any AI cost is irrelevant—human officers remain the only viable option, making AI effectively infinitely more 'expensive' in the sense of non-functional.
Technical feasibility todayclaude-haiku-4-5-202510011/5This task intrinsically requires a human officer to locate individuals, verify identity, and physically deliver court documents. No deployed AI product performs this task in any form.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical service of process; this remains an entirely human, field-based activity.

Locate and confiscate real or personal property, as directed by court order.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Law enforcement continues to rely on human officers for property seizure due to legal mandates and constitutional requirements. No meaningful automation or AI-driven displacement is occurring in this domain.
Sector adoption velocityclaude-sonnet-51/5Law enforcement's physical enforcement functions show essentially no AI adoption or displacement; this is a low-digitization, physical-world task.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could marginally assist by helping locate property records or cross-reference court documents before an officer's physical visit, but the core execution—locating and confiscating—remains fundamentally human and cannot be substantially augmented.
Augmentation potentialclaude-sonnet-52/5AI may help with locating property records, addresses, or asset databases beforehand, but offers minimal assistance during the actual physical confiscation task.
Task automatabilityclaude-haiku-4-5-202510011/5Locating and confiscating property requires physical presence, legal judgment about what qualifies under a specific court order, and discretionary decision-making in real environments. Current AI cannot execute physical actions or independently interpret nuanced legal directives in situ.
Task automatabilityclaude-sonnet-51/5This requires physically locating, confronting individuals, and executing legal seizure of property under a court order—a physical, judgment-laden law enforcement action that current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5This task is protected by hard legal barriers: only commissioned officers with legal authority can execute court orders and confiscate property. Liability for wrongful seizure, constitutional protections against unlawful search and seizure, and statutory restrictions on who may enforce court orders create insurmountable regulatory and legal barriers.
Adoption barriersclaude-sonnet-55/5Only sworn, authorized law enforcement officers have legal authority to execute court orders and use force/seizure powers; this is a hard legal and licensing barrier.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems cannot perform this task at all today, making any cost comparison academic. The human officer's loaded wage remains the only viable cost.
Cost vs. human wageclaude-sonnet-51/5AI has no capacity to execute this task at all, so any AI cost is irrelevant compared to the human officer who must physically perform the seizure.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system can physically locate property, assess legal compliance with a court order, or confiscate items. This task fundamentally requires human agents with legal authority and physical capability.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical property confiscation or asset seizure; this remains entirely a human law enforcement function.

Related occupations — Protective Service

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.