概要: An AI system developed by Infinite Campus and deployed by Nevada to identify at-risk students led to a sharp reduction in the number classified as needing support, dropping from 270,000 to 65,000. The reclassification caused significant budget cuts in schools serving low-income populations. The drastic reduction in identified at-risk students reportedly left thousands of vulnerable children without resources and support.
Editor Notes: Timeline notes and clarification: Before 2023, Nevada identified at-risk students mostly by income, using free or reduced-price lunch eligibility as the key measure. In 2022, this system classified over 270,000 students as at-risk. Looking to improve the process, Nevada partnered with Infinite Campus in 2023 to introduce an AI system that used more factors like GPA, attendance, household structure, and home language. The new system was meant to better predict which students might struggle in school. However, during the 2023-2024 school year, the AI cut the number of at-risk students to less than 65,000. This reclassification caused budget cuts in schools that depended on the funding tied to at-risk students, especially those serving low-income populations. By October 2024, the problem gained national attention.
Alleged: Infinite Campus developed an AI system deployed by Nevada Department of Education, which harmed Low-income students in Nevada , Nevada school districts , Mater Academy of Nevada と Somerset Academy.
インシデントのステータス
Risk Subdomain
A further 23 subdomains create an accessible and understandable classification of hazards and harms associated with AI
1.3. Unequal performance across groups
Risk Domain
The Domain Taxonomy of AI Risks classifies risks into seven AI risk domains: (1) Discrimination & toxicity, (2) Privacy & security, (3) Misinformation, (4) Malicious actors & misuse, (5) Human-computer interaction, (6) Socioeconomic & environmental harms, and (7) AI system safety, failures & limitations.
- Discrimination and Toxicity
Entity
Which, if any, entity is presented as the main cause of the risk
AI
Timing
The stage in the AI lifecycle at which the risk is presented as occurring
Post-deployment
Intent
Whether the risk is presented as occurring as an expected or unexpected outcome from pursuing a goal
Unintentional
インシデントレポート
レポートタイムライン
ネバダ州は長い間、国内で最も不均衡な学校予算を抱えてきた。同州の低所得地区は、裕福な地区に比べて生徒一人当たりに使える資金が約35％少なく、これはどの州よりも最大の格差である。
1年前、ネバダ州は外部の業者が提供する人工知能の助けを借りて、この疑わしい状況を改善しようとした。しかし、それは大騒動を引き起こした。
AIシステムは、学校で苦労する子供の数に関する同州の以前の推定値がはるかに高すぎると計算した。以前は、ネバダ州はすべての低所得の生徒を学業や社会的な問題の「リスク」が…
バリアント
「バリアント」は既存のAIインシデントと同じ原因要素を共有し、同様な被害を引き起こし、同じ知的システムを含んだインシデントです。バリアントは完全に独立したインシデントとしてインデックスするのではなく、データベースに最初に投稿された同様なインシデントの元にインシデントのバリエーションとして一覧します。インシデントデータベースの他の投稿タイプとは違い、バリアントではインシデントデータベース以外の根拠のレポートは要求されません。詳細についてはこの研究論文を参照してください
よく似たインシデント
Did our AI mess up? Flag the unrelated incidents
Machine Bias - ProPublica
· 15 レポート
Policing the Future
· 17 レポート
よく似たインシデント
Did our AI mess up? Flag the unrelated incidents
Machine Bias - ProPublica
· 15 レポート
Policing the Future
· 17 レポート