Lighthouse AdvisorySLED AI Adoption Intelligence

K-12Issue 224 records

K–12

Fresh K-12 signal after the 26 Sep dual-path/AI-memory edition centers on access-without-engagement, consent/supervision forcing live-tool rework, and teacher-AI workload skepticism. Hechinger (21 Sep) reports the Tennessee Khanmigo cluster RCT’s practical punchline: Khan Academy practice helped modestly, but the Socratic AI assistant added nothing because students quit when it withheld answers—Sal Khan concurs and is shipping in-workflow, AI-optional redesigns. Flagler Schools (23 Sep) shows a midsize Florida district translating statewide parent opt-in, PreK–5 review, companion bans, and unsupervised-use prohibitions into board policy work, warning that existing Khanmigo use may fail the unsupervised test unless vendors offer AI-off modes and kill-switches before July 2027. Tes’s Bett/YouGov poll (27 Aug; newly salient) finds UK teachers’ weekly AI use doubled to 52% while only 35% work fewer hours and enthusiasm to use more fell to 47% from 68%—a caution against US workload-ROI sales decks. Montana’s Aug 2026 AI Guidelines v2 (not previously archived) supply SEA language: preserve productive struggle, treat assisted task gains as non-proof of independent mastery, and refuse AI as counseling, with companion tools on a stricter track. Net for SLED: demand unaided mastery evidence and engagement depth for student tutors; inventory live tools against supervision/consent rules now; and measure teacher-AI success in hours and quality, not logins.

Evidence records
4
Cross-source patterns
3
Evidence classes
3 independent reporting1 standards or public-body guidance
Outcomes
2 cautionary2 emerging
Source freshness
2 new this fortnight2 recent
Research completed
2026-09-28

Choose a role to see its takeaway beside every record in the ledger.

Patterns across the evidence

Synthesis · Lighthouse Advisory interpretation

3 patterns, each supported by at least two sources

  1. Student AI access can rise without engagement or independent mastery

    Hechinger’s Khanmigo coverage shows students abandoning Socratic help so AI adds no incremental learning over non-AI practice, while Montana’s guidelines warn that assisted task gains do not prove independent mastery, retention, or transfer. Together they push SLED buyers from license metrics toward engagement depth and unaided demonstration.

    Operating questionBefore scaling any student AI tutor, can learners show the skill without the tool—and do usage logs show sustained help-seeking rather than one-off clicks?

    Supporting evidenceThe Hechinger ReportMontana Office of Public Instruction / Montana Digital Academy Frontier Learning Lab

  2. Teacher AI adoption is not proof of workload-hour ROI

    Tes/Bett polling shows doubled weekly teacher AI use with only about one-third working fewer hours and cooling enthusiasm, while Montana treats educator prep-AI as comparatively lower risk only when verification and local time savings are real. US teacher-first rollouts need hour-level and quality gates, not adoption dashboards alone.

    Operating questionDoes our teacher-AI success metric include total weekly hours and lesson quality—or only prep minutes and login counts?

    Supporting evidenceTes (Times Educational Supplement)Montana Office of Public Instruction / Montana Digital Academy Frontier Learning Lab

Evidence ledger

Full record · every source keeps its link and limitations

4 records
  1. Independent reportingCautionaryNew this fortnight

    Hechinger: Tennessee Khanmigo study finds AI tutor added no gains over non-AI practice as students refused Socratic help

    Jill Barshay reports that a University of Toronto two-year study in 18 Tennessee middle schools found students assigned Khan Academy with Khanmigo improved modestly versus usual remediation, but gains matched prior non-AI Khan Academy practice because students largely stopped using the Socratic tutor after it withheld answers and asked questions. Sal Khan publicly accepted the engagement diagnosis and described product changes (in-workflow help after errors; optional AI-off).

    The Hechinger ReportUnited States (Tennessee middle-school remediation trial)September 21, 2026

    Why it matters, evidence and limitations
    Why it matters
    Directly informs US district procurement of student-facing AI tutors: access and 'Socratic' framing are insufficient without sustained engagement and unaided transfer evidence.
    Evidence and measured results
    Independent education journalism summarizing NBER/EdWorkingPapers RCT results and on-record Khan Academy response; not a substitute for the primary paper already archived elsewhere.
    Limitations and uncertainty
    Journalistic secondary source; remediation-block population; product versions evolving; primary RCT already in SLED archive under edworkingpapers URL.
    Hechinger Report Proof Points Role takeaways & implementation
  2. Independent reportingEmergingNew this fortnight

    Flagler Schools begins aligning policies to Florida AI rules; Khanmigo may conflict with unsupervised-use ban

    Flagler Schools reports early work to amend internet-safety policy for Florida’s Sept 16 AI rules (parent notice/opt-in with non-AI alternative; PreK–5 extra review; no SEL/companion AI; staff training; unsupervised student AI banned; staff must disable AI without vendor help). Officials say a 2024 board policy and taskforce participation leave them comparatively prepared, but note Khanmigo—used especially in grades 6–12—could conflict with the unsupervised-use prohibition; a UF Lastinger/Khan Academy grant partnership was discussed as Khan Academy prepares AI-optional modes.

    Palm Coast Observer / Observer Local NewsUnited States — Flagler County, FloridaSeptember 23, 2026

    Why it matters, evidence and limitations
    Why it matters
    Shows midsize Florida district operationalizing statewide AI constraints and exposing live-product conflicts—useful beyond megadistrict pause narratives.
    Evidence and measured results
    Local education reporting with on-record CTO and instructional technology coordinator quotes from a Sept 22 board workshop.
    Limitations and uncertainty
    Local reporting; final local policy not yet adopted; News Service of Florida contributed statewide context already covered in prior SLED Florida editions.
    Palm Coast Observer Role takeaways & implementation
  3. Independent reportingCautionaryRecent

    Tes: Bett/YouGov finds UK teacher weekly AI use doubled but only 35% work fewer hours; enthusiasm cools

    Cerys Turner reports Bett’s Working It Out analysis of a YouGov survey of 1,033 UK teachers: weekly AI use rose to 52% from 26% year-over-year; among AI users 51% say workload eased but only 35% work fewer hours (55% same hours). Share expecting to use AI more fell to 47% from 68%. Secondary teachers report high suspicion of AI-assisted pupil work; most want coursework redesigned rather than left unchanged.

    Tes (Times Educational Supplement)United Kingdom (transferable caution for US SLED workforce claims)27 August 2026

    Why it matters, evidence and limitations
    Why it matters
    Newly relevant scrutiny against US teacher-first AI ROI pitches: adoption and task-time savings are not hour reductions without workload policy. Why now amid US districts scaling teacher AI copilots.
    Evidence and measured results
    Education-press report of nationally weighted YouGov polling for Bett; self-reported hours; UK policy context (DfE) differs from US.
    Limitations and uncertainty
    UK sample; August publication older than last k12 run—included for newly salient contradiction of workload claims; not causal RCT.
    Tes Magazine Role takeaways & implementation
  4. Standards or public-body guidanceEmergingRecent

    Montana OPI AI Guidelines v2: preserve productive struggle; assisted performance ≠ mastery; companion AI is not support

    Montana’s August 2026 Version 2.0 guidelines (advisory) tell districts to adopt evidence-driven AI use, preserve productive struggle, keep humans in the loop, and protect student data. They warn that better assisted task performance does not prove independent mastery, retention, or transfer; urge a clear continuum of AI not-permitted / bounded / expected per activity; and instruct districts not to treat AI as counseling or crisis support, with separate review for companion-style tools. Immediate actions: cross-functional AI team, PD before student use, community communication; vendor checklist covers FERPA/COPPA, bias, engagement dark patterns, accessibility, and privacy agreements.

    Montana Office of Public Instruction / Montana Digital Academy Frontier Learning LabUnited States — Montana (advisory for local districts; transferable principles)August 2026 (Version 2.0)

    Why it matters, evidence and limitations
    Why it matters
    Why now: supplies independent SEA language matching this edition’s engagement-vs-mastery and companion/supervision themes for districts writing first policies.
    Evidence and measured results
    Official state guidance PDF opened in full; advisory, not statute; twice-annual revision process disclosed; AI assisted drafting with human responsibility noted.
    Limitations and uncertainty
    August dated—older than last run but not previously archived; Montana-specific legal anchors; rural capacity assumptions may differ for large urban LEAs.
    Montana AI in K-12 Education Guidelines v2 Role takeaways & implementation

How to read this edition

Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.

Independent reporting
Independent reporting with attributable sources but without a formal evaluation design.
Standards or public-body guidance
Normative or advisory guidance from a standards body or public institution.