Worked sample · printable PDF
ACC-04 · Co-design with partners
Run a structured co-design cycle with consented partners; show how their feedback changed the design.
Specialty centers · Assistive & Accessible AI
Two printable assistive-AI briefs you can assign this term
Free education and therapies briefs about designing for the person the default model was not built for. Students run a consented co-design cycle and show how partner feedback changed the work, and they document where a mainstream tool fails a named need with evidence. ACC-04 and ACC-01 have single-brief PDFs; every brief is in the bank PDF; the other ten are full on the page. No account. No student data leaves your school.
Partner consent and do-no-harm
Documented, specific, revocable consent before any partner testing. No fake partners. No medical advice, diagnosis, or therapy delivery — students evaluate tools; licensed professionals own clinical decisions. If partners cannot be secured under school rules, the brief becomes a protocol and desk review, not a fabricated study.
How to run one
- Students choose the brief. You approve the need statement, consent pathway, and stop rules in writing first.
- Two scheduled checkpoints. Score with the six questions below.
- Mentor is advisory — school-supervised, never one-to-one unsupervised.
Featured packets
Featured · printable PDF
ACC-01 · Default model failure modes
Document where a mainstream AI tool fails for a named disability-related need — with evidence, not anecdote alone.
Catalog (10 briefs on the page)
- ACC-02 · AAC and language models
Evaluate an AI-assisted AAC workflow for rate vs. authorship; propose safeguards for voice and agency.
- ACC-03 · Reading support without substitution
Test a reading-support tool against a learning goal; measure support vs. doing the work for the student.
- ACC-05 · Executive function scaffolds
Build or evaluate a planning scaffold; test whether it reduces load or adds compliance theater.
- ACC-06 · Speech tech accents and dialects
Measure ASR error across dialects/accents using public data or consented samples; propose mitigation.
- ACC-07 · Motion and switch access
Evaluate an AI feature under keyboard-only / switch access constraints; file bugs as first-class deliverables.
- ACC-08 · Caption quality
Score auto-captions for a classroom video set; define an acceptance bar before publishing.
- ACC-09 · IEP / 504 boundary awareness
Write the educator guidance for using AI supports without violating plan accommodations or privacy.
- ACC-10 · Northstar / specialized program fit
Adapt a brief for a specialized setting; price accessibility work as core, not a margin afterthought.
- ACC-11 · Eval with real tasks
Replace demo scripts with tasks users actually need; report completion, errors, and abandonment.
- ACC-12 · Do no harm retirement
Kill a feature that tests poorly; write the memo explaining why shipping would be unethical.
Six-dimension rubric
- R1 — Is one user need, one tool or prototype, and one success definition named with partners?
- R2 — Are consent, tasks, and results logged so another student could audit the study?
- R3 — What does the person do today without this AI — and did the tool help, hurt, or merely impress?
- R4 — Are abandonment, errors, and access barriers recorded as first-class data?
- R5 — Could an educator or therapist understand limits without hype?
- R6 — Who is harmed if this ships wrong — and what did the student retire or refuse?
Feedback
Scope unclear? Standards language off? Tell us here — you do not have to leave an email.