Abstract
This disclosure describes FairSI, a realm in which students build, test and prove governed AI systems, and through which young people help define what fairness in AI should mean for them. Students prove that they are currently enrolled with a credential issued by their school, presented without revealing their name, school or age unless a rule requires it, one account per credential, re-verified each year. Each student works in an isolated playfield that runs the same governance as certified systems, including the runtime governor, the owner's stop and restore, identity and the signed record, but is fed only synthetic and open data, is separated from every production system, and permits attack-and-defence exercises only inside it and under published ethical rules. Each student's own work and notes are held under keys only the student holds, exportable and deletable with proof of deletion. The service is free, and its promises (no advertising based on students' data, no sale or sharing of that data, no training on students' work without their revocable consent, honest labels on AI-generated content) are enforced by the system's own controls and attested continuously, with every attestation and every exception recorded in a public registry, so that "free without making you the product" can be checked rather than trusted. Students may design fairness tests for AI systems, run them in the playfield and submit them through a recorded pipeline: a test is made reproducible, reviewed by peers and by experienced tutors, and, if adopted by an accredited body, becomes a candidate test case for certification, carrying a permanent attribution to its student authors. Students never certify anything themselves. An elected Student Council advises on the realm's priorities and reviews its charter each year, with its decisions and views recorded and published. Student interns build and help run the realm under tutors, with every change to its trust core reviewed before release and no access to other students' data. Educators may set course modes that allow explanation and feedback but not finished assignments, so the realm supports rather than undermines learning.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Wang, Edward D. H., "FairSI™: A Students' Realm for Governed AI, in which Verified Students Build, Test and Prove AI Systems in an Isolated Playfield Governed Exactly Like Certified Systems but Fed Only Synthetic and Open Data; Hold Their Own Work Under Their Own Keys; Receive Free Service Whose "Not the Product" Promises Are Enforced and Attested Rather than Merely Stated; Turn Their Own Fairness Tests into Candidate Certification Test Cases Through a Recorded, Attributed Pipeline; and Govern the Realm Through an Elected Student Council Whose Decisions Are Recorded", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12919