Salvato in:
| Autore principale: | |
|---|---|
| Natura: | Recurso digital |
| Lingua: | inglese |
| Pubblicazione: |
Zenodo
2025
|
| Soggetti: | |
| Accesso online: | https://doi.org/10.5281/zenodo.15708778 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
Sommario:
- <p>This report examines the legal, institutional, and civil rights risks posed by AI-generated content detection tools in U.S. public education and universities with federal funding. Originally introduced to support educators in managing academic integrity, these systems now influence disciplinary decisions, often in the absence of transparent standards, regulatory oversight, or meaningful appeal processes.</p> <p>Drawing from technical literature, vendor documentation, and federal legal frameworks, the report demonstrates how detection models, built on linguistic norms and probabilistic inference, expose students with non-normative writing patterns to measurable risk. Particularly those who are neurodivergent, multilingual, or disabled. The analysis centers on three intersecting concerns: (1) The opaque, proprietary nature of detection algorithms; (2) disproportionate impact on neurodivergent, multilingual, and disabled students; and (3) potential institutional liability created when public schools delegate evaluative authority to private, unregulated software. </p> <p>Rather than categorically rejecting detection tools, this report offers a systems-level critique of their deployment in punitive settings. It highlights how these practices clash with constitutional guarantees, disability accommodations under federal law, and due process protections in educational environments.</p> <p>While detection technologies may have a legitimate role in academic workflows, their unregulated use in disciplinary contexts, without transparency, accommodation protocols, or qualified human oversight, create risk to equity and legal compliance. The impact of these technologies does not stem from the tools alone, but from public systems that enforce algorithmic outputs without accountability.</p>