AICoP: Introducing Gemini & NotebookLM

Columbia AI Community of Practice Introduces Google Gemini and NotebookLM

Columbia University’s AI Community of Practice celebrated Halloween with the launch of Google Gemini and NotebookLM, two advanced AI tools now available through Columbia’s LionMail workspace. The event featured guest speaker Jillian Yoerges from Google, alongside demonstrations by John P. Martin and Spencer Ames from CUIT’s Emerging Technologies team.

What Are Gemini and NotebookLM?

Gemini and NotebookLM are part of Google’s new suite of generative AI tools, now included under Columbia’s existing Google Workspace for Education license. Both tools are free to use and are designed to help faculty, staff, and researchers enhance productivity, learning, and research through secure AI integration.

Gemini for Education offers conversational AI capabilities similar to ChatGPT, while NotebookLM functions as a research assistant that can analyze, summarize, and generate insights from uploaded documents, spreadsheets, and even YouTube videos.

A pilot program is also underway to integrate Gemini directly into CourseWorks (Canvas), allowing instructors to use AI tools seamlessly within their teaching environment.

Why It Matters

• Secure and compliant: Both tools operate within Google’s enterprise-grade security framework, ensuring that user data is encrypted, never used for training, and protected under the university’s Google Workspace agreement.
• Accessible for all users: Faculty, staff, and researchers can log in using their LionMail credentials.
• Free to use: No additional cost or subscription required.
• Supports multiple roles: Designed for use across teaching, research, and administration.

Key Features

Gemini

• Two models available: 2.5 Flash for general tasks and 2.5 Pro for advanced reasoning and coding.
• Tools include Guided Learning, Deep Research, Canvas, and Image Generation.
• Can generate content, write and debug code, summarize data, and automate workflows.
• Offers “prompt starters” and Canvas mode for iterative writing and testing.

NotebookLM

• Imports PDFs, Docs, Sheets, websites, and YouTube videos.
• Summarizes, compares, and cites from multiple sources automatically.
• Generates briefing documents, quizzes, flashcards, and audio or video explainers.
• Now supports Google Sheets, allowing users to analyze historical data, budgets, or research metrics.
• Creates interactive “studios” for generating reports or study materials.

Demonstrations

John Martin demonstrated Gemini’s versatility, showing how it can:

• Write automation scripts to organize local files.
• Analyze U.S. Census data using Python code generated by the AI.
• Leverage Canvas mode for real-time editing and iteration.

Spencer Ames showcased NotebookLM’s ability to synthesize information from diverse sources, generate citations, and even create interactive audio explainers. He also demonstrated Gemini’s image generation, producing consistent icons, designs, and personalized visuals for presentations and coursework.

Practical Use Cases

• Faculty: Summarize readings, build quizzes, or generate study materials.
• Researchers: Aggregate literature, draft reports, and compare papers.
• Administrators: Analyze budgets, process applications, or summarize meeting notes.

Security and Governance

Google’s AI tools come with enterprise-grade protections:

• No ads, data collection, or model training from user inputs.
• Compliance with SOC 1, SOC 2, COPPA.
• Managed entirely within Columbia’s existing Google infrastructure.

Support and Training

• CUIT hosts AI office hours for Gemini and NotebookLM on Fridays from 10 - 11 am. You can find the links here.
• Faculty and staff can register through the Emerging Technologies website.
• Additional training and prompt libraries are available through the CUIT knowledge base.

Takeaway

The launch of Gemini and NotebookLM marks another milestone in Columbia’s effort to provide secure, practical, and accessible AI tools for its community. By combining creative flexibility with institutional oversight, these tools help faculty, researchers, and administrators work smarter—while keeping privacy and responsibility at the forefront of AI adoption.

Try It Yourself

During the demo, John P. Martin showed how Gemini can generate working code that automatically organizes files on both Windows and macOS systems. The prompts below can be used directly in Gemini (set the model to 2.5 Pro) to recreate the same automation:

Windows File Organizer Prompt:

I'm not a coder, but my "Downloads" folder on my Windows PC is a complete mess. I don't have any special programming software installed. Could you write a simple Windows Batch Script (a .bat file) that can help me organize a folder? I'd like it to: Ask me for the folder path I want to clean up (like C:\Users\MyName\Downloads). Look at every file in that folder. Create new subfolders based on file type (e.g., "Images", "PDFs", "Documents", "Videos", "Other"). Move the files into the correct new folder. For example, all .jpg and .png files would go into the "Images" folder, and all .pdf files would go into the "PDFs" folder. Most importantly, please give me simple, step-by-step instructions on how to run this. I need to know how to save the code in Notepad (what to name the file, like organize.bat) and how I can just double-click that file to make it run. 

macOS File Organizer Prompt:

I'm not a coder, but my "Downloads" folder on my Mac is a complete mess. I don't have any special programming software installed. Could you write a simple Mac Shell Script (a .command file) that can help me organize a folder? I'd like it to: Ask me for the folder path I want to clean up (like /Users/MyName/Downloads). Look at every file in that folder. Create new subfolders based on file type (e.g., "Images", "PDFs", "Documents", "Videos", "Other"). Move the files into the correct new folder. For example, all .jpg and .png files would go into the "Images" folder, and all .pdf files would go into the "PDFs" folder. Most importantly, please give me simple, step-by-step instructions on how to run this. I need to know how to save the code in TextEdit (what to name the file, like organize.command) and how I can just double-click that file to make it run.

Faculty and staff interested in exploring CHAT or developing custom AI agents can contact the Emerging Technologies team at [email protected]

Link to slide deck can be found here.