Proposed evaluation framework
We propose to evaluate whether AltStatic improves everyday task completion, supports continued product use, and reduces resource use in selected AI-supported design and manufacturing workflows. These are outcomes to be tested, not benefits already demonstrated. This page reports no completed pilot results.
Define the comparison before collecting data
Each pilot should document its setting, intended participants, baseline method, numerical targets, measurement period, sample size, data sources, and review responsibilities before enrollment. The comparison must reflect equivalent tasks or production requirements. Repeated tasks by one person must not be reported as if they were independent participants. Missing follow-ups, unsuccessful outcomes, and changes to the plan belong in the results.
Four proposed outcome measures
Accessibility and participation
Measure successful tasks as a proportion of attempted tasks, alongside time, errors, reported effort, and the participant's preferred support level. Intended users, including autistic participants where relevant, should help define useful outcomes. More independence is not automatically better than chosen human support.
Reliable AI assistance
Evaluate a defined sample against approved information for accuracy, grounding, and the correct object and task. Report the number evaluated, pass criteria, useful outcomes, unsafe or unsupported responses, and appropriate human escalation. A fluent answer or a high interaction count is not proof of reliability.
Product longevity
Record documented maintenance, repair, or reuse and confirm whether products remain in service at a specified follow-up. Identify the participating population and products lost to follow-up. Describe AltStatic's contribution cautiously; continued use alone does not establish that the system caused an avoided replacement. Estimate avoided waste only with a documented comparison and supporting evidence.
Manufacturing efficiency
Compare material scrap, measured energy per accepted item, and rework rates across production runs with comparable designs, materials, volumes, and quality requirements. Account for added physical and digital resources where measurable, and disclose excluded processes and uncertainty. Do not present a partial measurement as a complete lifecycle or carbon assessment.
Participant safeguards before a pilot begins
- Use a written partner scope and accessible participant notice covering purpose, tasks, expected burden, foreseeable risks, data collected, recipients, retention, and a contact for questions.
- Explain voluntary participation and withdrawal, including what can be deleted and any limits once data has been aggregated or must be retained. Separate participation from optional photos, quotations, or publicity.
- Agree on approved content, human support, an incident-reporting path, and conditions for pausing unsafe or burdensome tasks.
- Determine whether research ethics review, parental permission, participant assent, or other authorization is required for the actual project. Autism or disability does not by itself mean someone lacks capacity to consent.
- Minimize sensitive information and protect small groups from identification. Browsing this page is not consent to participate.
Reporting and sustainable development goals
Results should state the pilot dates, methods, denominators, comparison, limitations, and whether findings support only feasibility or a stronger conclusion. These project-level measures relate to inclusion underSDG target 10.2, waste prevention and reuse undertarget 12.5, and industrial resource efficiency undertarget 9.4. They are not the official UN statistical indicators, UN endorsement, or certification.
Our proposed standard for success is measurable benefit to people and products, rather than simply more scans or AI interactions. Contact info@stateofstick.coto discuss a defined evaluation. Read our Privacy Policy andAI and assistive-use information.