← All education signals
Governance

The Human Mark

Oklahoma law bars AI as the main basis for a grade and makes a teacher review its output. In Dallas schools, human rescoring lifted about a third of resubmitted tests.

What is it?

AI may mark student work, but a human must own the grade. Oklahoma bars AI as the primary basis for grading and requires teachers to review its outputs. In Texas, many machine-scored state essays rose when humans rescored them, according to district figures reported locally.

Why now?

Oklahoma’s bill passed both chambers without a vote against and was signed on 12 May 2026. Texas districts paying for human rescoring saw many essays lifted, and four states now require district AI policies, as machine scoring spreads into classrooms and tests.

Who’s building it?

State legislatures, education agencies, scoring-engine vendors, school districts and the parents gaining a right to opt out.

Oklahoma LegislatureTexas Education AgencyScoring enginesSchool districtsParents
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — scoring-engine vendors and the agencies deploying them.
Buyers — students and districts whose ratings rise or fall on a machine’s mark.
Gatekeepers — state legislatures writing human review into law for schools.
What’s at stake
Whether a machine can give the grade that decides a student’s placement, promotion or a school’s rating.
Whether human review is real or a signature.
Whether paid rescoring becomes a routine check on machines.
What’s unsettled
How often machine marks fall short across Texas, not just two districts.
What counts as the primary basis once a teacher reviews.
Whether other states copy Oklahoma’s grading bar or stop at requiring district policies.
02 · How the demand matureda progression, not a timeline
Novelty
Scoring engines promise essay marking at scale with fewer human graders.
Retention
Texas keeps machine scoring, cutting graders from about 15,000 to about 5,000.
Willingness to pay
Districts pay for human rescoring and see about a third of Dallas resubmissions lifted.
Category
Human review becomes a legal feature of any AI used for grading.
03 · Where the market goesthe forks, not a prediction
→The reviewed machine: AI still drafts every mark, and a teacher signs off each grade that feeds a high-stakes decision.
→More states copy Oklahoma and bar AI as the primary grader in law.
→Review becomes a formality, and teachers approve machine marks at speed without real checks.
→Rescoring results push Texas to publish statewide rates of machine error.
The forks turn on whether a human reviewer has the time and standing to overturn a machine’s mark, or only to approve it.
04 · The Assessment deskadjacent market, moving now
The Detector Doubt
A machine flag stops counting as proof.
A court annulled a finding built on a Turnitin AI score.
NY Supreme Court (FindLaw)
The Supervised Page
Exam rooms now ban glasses that carry AI.
The College Board bans smart glasses at the SAT from March 2026.
Inside Higher Ed
The Bias Audit
Automated decisions about people now face outside checks.
Colorado replaced bias audits with a narrower transparency regime.
Colorado General Assembly
05 · The Wire
01
Oklahoma law bars AI as the primary basis for student grades
Oklahoma Legislature ↗
02
The Oklahoma Senate passed SB 1734 unanimously, 42-0.
Oklahoma Senate ↗
03
Four states now require district AI policies; the grading bar is Oklahoma’s.
K-12 Dive ↗
04
Texas hired almost 10,000 fewer STAAR essay graders from 2023 to 2024
The Austin Chronicle ↗
05
Texas STAAR reading rescore requests reached 64,489, from 21,620 a year earlier
The Texas Tribune ↗
06
Oklahoma House passed SB 1734 89-0; governor approved it 12 May 2026
Oklahoma Legislature ↗
The panel on this signal
They agreed on
–The grading bar is one state’s law; other states require policies, not a ban on machine grading.
–Texas rescore figures are district reports, not statewide.
–Oklahoma allows AI marking as long as a person reviews and owns the grade.
The psychometrician disagreed on
–Rescores requested by districts expecting gains are a biased sample; only blind double-marking shows which marker drifts.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.