{
  "source": "Cooked Index — occupational AI risk register",
  "page": "https://cookedindex.com/jobs/educational-instruction-and-library-workers-all-other/",
  "methodology": "https://cookedindex.com/methodology",
  "notice": "Verdicts are re-examined as evidence accumulates. Re-fetch before relying on this; the page above always carries the current score.",
  "scored_at": "2026-08-11",
  "model": "claude-opus-5",
  "occupation": {
    "title": "Educational Instruction and Library Workers, All Other",
    "soc_code": "25-9099",
    "category": "Education",
    "us_employment": 118590,
    "median_annual_wage": 50890
  },
  "verdict": "EXPOSED",
  "risk_resistance": 41,
  "contested": false,
  "near_boundary": false,
  "dimensions": {
    "task_resistance": 9,
    "embodiment": 10,
    "liability_shield": 3,
    "trust_premium": 11,
    "judgment_accountability": 8
  },
  "reasoning": {
    "task_resistance": "Roughly half the workday — drafting worksheets, adapting texts to reading level, writing resource guides, pulling citations for a student's paper, logging attendance and progress notes — is text-in/text-out work a model does in seconds, while the other half (sitting beside a student who won't start their essay, running a small-group study skills session, setting up a maker space or testing room) doesn't compress, which is why this lands at 9 rather than in the resistant teens.",
    "embodiment": "You are on a campus or in a library building: shelving and processing materials, moving AV carts and laptops, staffing a tutoring desk, proctoring a testing room, supervising a group during an after-school block — physical but in controlled, climate-conditioned, predictable spaces, which puts it at 10 rather than the 15+ of a field technician or trades instructor.",
    "liability_shield": "Most of these positions require a bachelor's or less with no state credential, no teaching license, and no professional certification a state board could pull — the certified teacher of record or the head librarian signs off on grades, IEP paperwork, and collection decisions, so the 3 reflects a real absence of statutory gatekeeping, not merely weak protection.",
    "trust_premium": "An 11 recognises that the student who keeps coming back to your table, or the teacher who routes their toughest reader to you, is choosing you specifically — but the assignment is often term-length, students rotate, and the district can post the role and fill it without the returning cohort collapsing, so the relationship supports the work without being the deliverable.",
    "judgment_accountability": "You decide which intervention a struggling reader gets next, when to escalate a behaviour or welfare concern, and how to reallocate a session when three students show up needing different things — real discretion inside a curriculum, a district handbook and a mandated-reporting chain that specifies the escalation, which is why it sits at 8 and not in the mid-teens where the final call is yours to own."
  },
  "rationale": "This catch-all bucket holds instructional support staff, learning-center and tutoring coordinators, curriculum and media assistants, and program aides; the modal worker splits time between screen work (building lesson materials, cataloging resources, scheduling, tracking student progress, answering reference and referral questions) and in-person contact with students on a campus or in a library. The screen half is exactly what generative AI does cheaply — worksheet generation, reading-level adaptation, resource summaries, routine research help. The in-person half — sitting with a struggling student, running a study skills group, keeping a program physically operating — holds, but it carries no license and no signature requirement, so the only real moat is presence and relationship.",
  "outlook": "Headcount likely flattens or drifts down as AI absorbs materials prep and reference lookup, while the staff who remain shift toward hands-on student contact and program operations on site.",
  "what_would_raise_it": {
    "levers": [
      {
        "dimension": "task_resistance",
        "change": "Task-mix shift: if worksheet generation, reading-level adaptation, cataloging and routine reference are absorbed by district-licensed AI tools (Khanmigo-type deployments, ILS AI discovery layers), the residual day becomes small-group intervention, behavior de-escalation, attendance/truancy follow-up and MTSS/RTI progress monitoring with a live child — work current models cannot do. Watch for job postings that drop 'creates instructional materials' and add 'delivers Tier 2 intervention'.",
        "plausibility": "already happening",
        "would_add": 4
      },
      {
        "dimension": "liability_shield",
        "change": "State or district rules making a named, certified human the required signer on IEP/504 service logs and progress data used in eligibility decisions — i.e., barring AI-generated progress notes from being entered without a credentialed staff attestation. IDEA documentation audits and state paraprofessional certification (ESSA Title I para requirements) are the existing hooks; several states are drafting AI-use rules for special-education documentation.",
        "plausibility": "plausible",
        "would_add": 4
      },
      {
        "dimension": "liability_shield",
        "change": "Mandatory-reporter and child-supervision duties being formalized into the role description with personal legal exposure (state child-abuse reporting statutes already name school employees), plus state student-data-privacy rules (Illinois SOPPA, NY Ed Law 2-d) that require a designated human to approve any AI tool touching student records.",
        "plausibility": "already happening",
        "would_add": 2
      },
      {
        "dimension": "trust_premium",
        "change": "State legislation requiring a human educator of record and prohibiting AI as the sole provider of instruction or tutoring hours — bills of this shape have appeared in several statehouses and in teacher-union contract bargaining (UTLA, Chicago Teachers Union AI language). If high-dosage tutoring funding is conditioned on human-delivered contact minutes, buyer preference becomes a purchasing rule.",
        "plausibility": "plausible",
        "would_add": 3
      },
      {
        "dimension": "trust_premium",
        "change": "Parent-facing demand: private learning centers and library programs explicitly marketing 'no AI tutor, a person sits with your child', and public library systems adopting ALA-aligned policies that reference questions receive human review. Observable in center advertising and in library AI policy statements.",
        "plausibility": "plausible",
        "would_add": 2
      },
      {
        "dimension": "judgment_accountability",
        "change": "Being formally seated on MTSS/RTI or threat-assessment teams where the role owns the recommendation to escalate a student — several states now mandate multidisciplinary behavioral threat assessment teams (e.g., Virginia, Texas SB 11, Florida) and support staff are named members. Ownership of an escalation call under ambiguity is a real accountability transfer, not a title change.",
        "plausibility": "already happening",
        "would_add": 3
      }
    ],
    "ceiling_note": "This is a catch-all SOC with no license, no scope-of-practice statute and no protected title, so the liability_shield route is capped: attestation duties can be added, but personal malpractice liability of the kind physicians or engineers carry is not on any visible path. The realistic ceiling is mid-50s, and it depends almost entirely on the job being redefined around in-person intervention and escalation rather than material production — a redefinition that in practice often comes with headcount cuts to the material-production half."
  },
  "adjudication": null,
  "employment_history": {
    "points": [
      {
        "y": 2019,
        "emp": 129040,
        "wage": 40310
      },
      {
        "y": 2020,
        "emp": 142550,
        "wage": 41570
      },
      {
        "y": 2021,
        "emp": 155950,
        "wage": 46300
      },
      {
        "y": 2022,
        "emp": 129170,
        "wage": 47650
      },
      {
        "y": 2023,
        "emp": 113490,
        "wage": 49800
      },
      {
        "y": 2024,
        "emp": 114640,
        "wage": 48400
      },
      {
        "y": 2025,
        "emp": 118590,
        "wage": 50890
      }
    ],
    "from": 2019,
    "to": 2025,
    "change_pct": -8.1,
    "comparable_from": 2019,
    "spans_soc_revision": false
  },
  "pivots": [],
  "license": "https://cookedindex.com/terms"
}