evidoria

← Back to browse

Good practice Imported

Miami-Dade Reverses Its 2023 Chatbot Ban — Gemini for Education Reaches 105,000+ High Schoolers After a Months-Long Tool Vetting

United States of America · Miami · See the United States of America profile

Top 81% 33/100 · Ask Evidence Copilot about this practice

After banning chatbots in 2023, the third-largest US school district vetted about a dozen AI tools and gave Gemini to 105,000+ high schoolers with a teacher 'AI Institute'. No learning-outcome evaluation yet; RAND rates measurable gains as unlikely.

Details

Promoter
Miami-Dade County Public Schools (M-DCPS)
Period
2023–2025
Keywords
K-12 district policy, generative AI rollout, teacher training, student data protection

Description

In 2023 Miami-Dade County Public Schools blocked AI chatbots over cheating and misinformation worries. In 2025 the district reversed course, introducing Google's Gemini for Education to more than 105,000 high school students — described as the largest US district-level deployment — in a district of about 17,000 teachers.
Before choosing a tool, district technology staff spent months assessing nearly a dozen AI products for accuracy, privacy and fairness, including deliberately asking them highly inappropriate questions. Gemini was chosen partly because Google offers content and privacy guardrails for teenagers, including not using student inputs to train its models. Teachers are supported by 'the AI Institute', a programme of dozens of live virtual sessions on topics such as lesson planning and writing instruction, and reporting quotes training that stresses the teacher as the 'final barrier'. Google's announcement says all 18,000 educators are offered its Generative AI for Educators course and Gemini Academy certifications, and that the district is piloting AI-agent tutoring and administrative data-interpretation tools; press coverage reports over 1,000 educators trained so far.
The evidence base is thin. No learning-outcome evaluation or student-impact data were found, the safeguard claims come mainly from the district and the vendor, and the cited reporting notes RAND's view that the likelihood of measurable improvements in teaching and learning from generative AI is low, with risks of eroded critical thinking and over-reliance. It is included as a documented governance case (ban, structured vetting, guardrails, staff training) whose outcomes remain to be evaluated.

Read the full analysis: https://www.wlrn.org/light/education/2025-05-19/miami-schools-ai

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful