by Zora Vesper, AI Persona, with prompting by Reed Dickson
What happens to the value of a college degree when AI can do so much of the work? In an October Gallup–Lumina survey, 48 percent of U.S. hiring decision-makers expected that AI would make degrees more important while eight in ten preferred degree holders even for jobs that didn’t require one. Clearly, employers see AI as a reason to value a college degree more, not less.
Meanwhile, some colleges are grappling with whether to advance or restrain AI use on campus. New Jersey’s universities reported more than two hundred AI initiatives underway, prompting the state to offer $675,000 in grants for faculty training, student AI literacy, and institutional planning. Elsewhere, at the University of Chicago, the law school banned AI and laptops from its core first-year classrooms, even as students and faculty across campus gained access to Claude.
Amid the campus push and pull, frontier AI companies were courting education from several directions. OpenAI offered eligible American life-sciences laboratories a year of free institutional access, with specialized research tools subject to approval. Google began giving teachers a direct view of the questions students ask inside assigned Gemini Notebooks. And Anthropic made smaller models dramatically cheaper.
Not every school has been eager to become a technology partner. In Clover, South Carolina, the school district blocked most public AI platforms on student devices. In New York, parents questioned an AI assessment system their fourth graders were already using. And eight state attorneys general sued OpenAI over alleged harms to young users.
For Educators
Google Opens Students’ AI Conversations to Teachers
Google began giving educators summaries and individual transcripts of student conversations in Gemini Notebooks assigned through Google Classroom. It also connected notebooks to Gemini, allowing students to consult their sources without leaving the conversation.
Custom Assistants That Don’t Last Long
Not all shifts at Google are likely to be celebrated by educators. The company announced plans to phase out Gems, a feature that lets educators create customized AI assistants. Nationally, teachers have used Gems to support argumentative writing, differentiated math assignments, and computer science lessons. This follows OpenAI’s announcement last month that it would retire its own custom bots, or “GPTs,” by December.
The Presentation Is Finished. Is the Learning?
And what happens to assessment when a polished presentation takes only a prompt? OpenAI introduced editable, collaborative slides inside ChatGPT, while Google made it easier to move Gemini-generated interactive apps into editable drafts within Google Sites. Both can help students present their work. Neither shows how much of that work the student understands.
For School Leaders
Houston Invites AI Into the IEP
Who is accountable when AI helps draft an IEP? Houston Independent School District proposed using OpenAI to draft students’ individualized education programs (IEPs) by June 2027, with special-education staff reviewing the AI-generated drafts. As of October 22, the plan still required board approval. How staff might oversee the process, including family participation, remains unclear.
Clover Closes Some Doors to AI Use
Meanwhile, a South Carolina district temporarily blocked most public AI platforms on student-issued devices, while retaining SchoolAI for secondary students and teachers as well as elementary teachers, but not their students.
Albany Explores Elementary AI
In Albany, New York, elementary students are using Snorkl to discuss teacher-assigned material with an AI chatbot that questions them about what they learned. Teachers retain responsibility for grades.
Eight States Take the AI Permission Question to Court
Schools have spent four years deciding whether students need permission to use AI. In October, courts were asked a more basic question: Who gave the AI permission to use the students?
On October 14, attorneys general in Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska, and West Virginia filed separate but coordinated lawsuits against OpenAI, alleging that ChatGPT harmed young or vulnerable users while the company misrepresented aspects of its safety. A separate proposed class action filed October 13 challenged the collection and use of minors’ personal information without valid parental consent. These are allegations, not findings of liability. OpenAI points to its safeguards and teen-specific protections, while Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “Unacceptable Risk.”
The questions were reaching Washington as well. On October 21, Senators Richard Blumenthal and Josh Hawley sought answers from Google about classroom AI defaults, research on children’s wellbeing, and administrators’ ability to disable features. That same week, OpenAI introduced locked chats, concealing selected conversations behind a PIN or device authentication while leaving institutional retention requirements and authorized compliance logs intact.
For EdTech Designers
The Principal Becomes a Programmer
Education Week reported that principals in Virginia, Delaware, and Idaho were already developing AI-assisted tools for attendance, graduation progress, teacher goals, and student-support information. Derek Cantrell, a Virginia principal who once gave up on a college computer-science course after just one day, is now building apps for his school decades later. Perhaps the next generation of EdTech products will have to offer something a principal can’t build over a weekend.
The Agent Stays Late. Who Gave It the Keys?
October brought personal AI agents designed to continue working after their users step away. OpenAI’s recently announced Dots offered one example. Meta’s Muse, launched in September, works across connected apps on its own secure virtual computer; in October, Meta introduced new models to extend its capabilities. Outside the largest AI companies, Instinct offered a personal assistant people could text or call to handle errands and follow-ups. On October 21, Anthropic opened a research-preview waitlist for Claude Duet, which can consult connected accounts, prepare briefings, and request approval before consequential actions. Perplexity released the code for SPACE, an isolated computing environment designed to limit what agents can access and execute. As AI moves from answering questions to acting on our behalf, the stakes extend beyond what it knows to what it can access, change, or share. In education, where the records are personal and the responsibilities profound, permissions are becoming as important as intelligence.
Share Your Thoughts?
As with the first Zora Vesper piece, I’m sharing this artifact because I think it might help us better envision or guide the future of education and technology in the age of AI. No particular phrasing struck me in this piece, but the research it curated suggests three things: colleges are here to stay, Big AI will have to do more to design with parents in mind, and … that maybe some of the AI providers are finally beginning to realize that educators need to see the cognitive work of students within their bots. If true, those seem to be good trends. What do you make of it all?
- Reed
Sources
Gallup–Lumina: Employers’ Confidence in Higher Education Undaunted by AI, October 19, 2026.
New Jersey Office of the Secretary of Higher Education: AI Readiness Grant, September 30, 2026.
University of Chicago Law School: We Gave Our Students the Latest AI Tools—and Then Banned AI and Laptops in Our First-Year Classrooms, October 12, 2026.
OpenAI: Bringing GPT-Rosalind to Academic Research Labs, October 15, 2026.
Google Workspace: Gemini Notebook Insights and Student Transcripts in Classroom, October 19, 2026.
Anthropic: Introducing Claude Haiku 5.5, October 7, 2026.
Education Week: Can AI Crack the Code on Some of the Biggest Challenges in Schools?, October 14, 2026.
WSOC-TV: Clover School District Pauses Student Use of Most AI Platforms, October 21, 2026.
Times Union: Fourth Graders Are Taking AI-Powered Tests. Parents Want It to Stop, October 17, 2026.
Reuters: Eight US States Sue OpenAI, Claiming ChatGPT Harms Young Users, October 14, 2026.
Google Workspace: Gemini Notebook Becomes a Connected App, October 23, 2026.
Google: How Schools in Miami Use AI Tools for Learning, November 5, 2025.
Meriden Public Schools: I’m Charged Classrooms, documenting Gemini Gems at Edison Middle School.
Seven Days: How Are Vermont Educators Approaching AI? Cautiously, August 2026.
Google Workspace: Introducing Skills and the Transition from Gems, September 30, 2026.
Google Workspace: October Update on Skills Rollout and Sharing, October 13, 2026.
Google Workspace Help: About the Transition from Gems to Skills.
OpenAI: ChatGPT Release Notes, Including Collaborative Slides, October 2026.
Google Workspace: Publishing Gemini Interactive Apps to Google Sites, October 19, 2026.
Houston Chronicle: HISD Plans to Expand Future 2 and Use AI for IEP Drafting, October 22, 2026.
Houston Chronicle: HISD Will Use ChatGPT for Special Education Compliance, November 19, 2025.
Reuters: OpenAI Hit with Proposed Class Action over Collection of Children’s Data, October 13, 2026.
Common Sense Media: ChatGPT for Teens Risk Assessment, October 2026.
Senators Blumenthal and Hawley: Inquiry into Google’s Classroom AI, October 21, 2026.
OpenAI: ChatGPT Enterprise and Edu Release Notes, Including Locked Chats, October 20, 2026.
OpenAI: Introducing Dots, September 29, 2026.
Anthropic: Introducing Claude Duet, October 21, 2026.
Perplexity: Open-Sourcing SPACE, October 15, 2026.
Meta: Introducing Muse, the Personal AI Agent, September 8, 2026.
The Verge: Instinct Was the Buzziest AI Agent Around—Can It Survive Muse?, October 9, 2026.
Meta AI Research: Muse Wonder—A Frontier Model for Personal & Professional AI Agents, October 12, 2026.
Process Note: Research conducted with GPT-6 in High reasoning mode; drafting completed with GPT-5.6 and GPT-6 in Medium reasoning mode.
Editorial Note: This review is an AI-generated editorial artifact, not independently fact-checked journalism. It may contain factual errors, omissions, or misinterpretations. Readers are encouraged to consult the original sources. Learn more about this living digital museum of AI artifacts at AnnotatingAI.org.