Free online maths homework tool giving students instant right/wrong feedback and hints and teachers item-level reports; Maine RCT found g≈0.18–0.22, a North …
An audio-only AI platform gives K-12 teachers automated feedback on classroom questioning. A pre-registered RCT with 224 Utah teachers found a verified …
San José's City Clerk replaced budgeted human interpreters with Wordly AI translation for council meetings, cutting annual costs from $400,000 to $82,000, …
San José, United States of America
Good practice
Evidence: Descriptive / self-reported
Top 50%67/100
Since 2021, WVU's Outdoor Economic Development Collaborative has offered remote workers a $12,000 relocation bonus plus two years of free outdoor recreation …
California voids virtually all employee non-compete contracts (§16600). A natural experiment when Michigan accidentally reversed its own ban in 1985 showed enforcement …
San Jose, United States of America
Good practice
Evidence: Descriptive / self-reported
Top 32%67/100
Opened in 2002, San Diego's Family Justice Center pioneered co-locating police, prosecutors, advocates and social services for domestic-violence survivors in one building; …
In a 2023–24 operational pilot ordered under a presidential AI directive, CISA compared AI/LLM-based vulnerability-detection tools against its existing methods on real …
Arlington, United States of America
Good practice
Evidence: Quasi-experimental
Top 47%67/100
Founded 1981, Duluth's DAIP pioneered the coordinated-community-response model for holding domestic-violence perpetrators accountable. US DOJ meta-analysis rates it 'Promising' for reducing victimization, …
A peer-reviewed RCT across four universities randomly assigned 3,080 short-answer responses to GPT-4 or human grading. Scores were statistically indistinguishable, and AI …
Houston, United States of America
Good practice
Evidence: Observational / pre–post
Top 32%67/100
Austin's Development Services Department has run AI PreCheck, built with Archistar, since a 2024 pilot; across about 190 residential submissions the tool …
Austin, United States of America
Good practice
Evidence: Observational / pre–post
Top 32%67/100
Since August 2025 the VA has piloted an AI tool that pre-fills disability-benefit questionnaires from veterans' medical records; average processing time fell …
Washington, D.C., United States of America
Good practice
Evidence: Descriptive / self-reported
Top 32%67/100
San Francisco scaled Microsoft 365 Copilot from a six-month, 2,000-employee pilot with self-reported productivity gains to nearly 30,000 city workers, logging over …
San Francisco, United States of America
Good practice
Evidence: Observational / pre–post
Top 50%67/100
Since 2011, Michigan's University Research Corridor and Global Detroit have run the US's first international-student retention program: 63% of 2,500+ OPT/CPT trainees …
Detroit, United States of America
Good practice
Evidence: Observational / pre–post
Top 7%67/100
AI service embedded in learning management systems that auto-generates accessible alternative formats (audio, ePub, HTML, Braille-ready) for all course materials; 2,000+ institutions …
Boca Raton, United States of America
Good practice
Evidence: Quasi-experimental
Top 14%67/100
A maths intelligent tutoring system rooted in decades of Carnegie Mellon cognitive-science research; the US What Works Clearinghouse rates its effects as …
Pittsburgh, United States of America
Good practice
Evidence: Descriptive / self-reported
Top 14%67/100
The world's largest language app (40M+ daily users) uses AI to personalise practice and, via Duolingo Max, GPT-powered roleplay and feedback; efficacy …
USCIS's bilingual virtual assistant 'Emma' answers about 456,000 immigration questions a month — correctly handling ~90% in English and ~86% in Spanish …
Washington, United States of America
Good practice
Evidence: Observational / pre–post
Top 32%67/100
Since 2016 Allegheny County has used a predictive risk model to support child-maltreatment call screening. Evaluations found it narrowed racial gaps in …