Skip to main content

Artificial Intelligence

info

This page was summarized from this guidance document: AI_at_Texas_AM_Faculty_Staff_Handbook

Overview & Purpose​

This handbook provides clear operational guidelines for Texas A&M University (TAMU) faculty and staff regarding artificial intelligence tools. Revised quarterly, it serves two main purposes:

  • Tool Catalog: Explains available university-provided AI tools, their capabilities, costs, and access points.
  • Workflow Matching: Guides users based on their specific work needs—from routine efficiency to custom app building and large-scale research.

Data Classification & Safety Rules​

Data classification dictates which AI tools are permissible. Never input sensitive, confidential, or proprietary university data into unapproved personal or public tools.

Classification LevelCoverage / ExamplesApproved AI Tools
Level 1: PublicPublished papers, public web pages, course catalogs.Free or paid personal accounts, TAMU enterprise tools.
Level 2: University-InternalInternal memos, emails, draft plans, working notes.TAMU Enterprise Tools ONLY (TAMU AI Chat/API, TAMU Google Gemini, TAMU Copilot).
Level 3: University-ConfidentialStudent records (FERPA), health data (HIPAA), PII, banking details, IRB/export-controlled data.TAMU AI Chat & TAMU AI Chat API (Approved without restriction). Exercise caution with TAMU Gemini/Copilot.
Level 4: CriticalCUI, Secret/Top Secret federal government data.DO NOT USE standard AI tools. Requires specific university IT approval.

The Golden Rule: Commercial/personal accounts (even paid ones like personal ChatGPT Plus or Claude Max) do not offer contractual privacy protection. Prompt inputs may be used to train models. Always use TAMU enterprise tools for non-public data.

Catalog of Tools​

1. Google Gemini & Gemini Notebook​

  • Access: Sign in at gemini.google.com using @tamu.edu NetID. Free for TAMU students, faculty, and staff with data protection enabled.
  • Gemini Notebook: Grounded specifically in user-uploaded documents (notes, readings, reports) to create study guides, summaries, or audio overviews without web hallucination.
  • Gemini Pro (Paid Add-on): Departmentally purchased license adding Gemini features inside Google Workspace (Docs, Sheets, Slides).
  • Compliance: Covered under TAMU's Business Associate Agreement (BAA) with Google; data is encrypted and not used for model training.

2. Microsoft Copilot Chat & Copilot​

  • Access: Copilot Chat (Basic) is free for TAMU NetID holders via Microsoft M365.
  • M365 Copilot (Paid Add-on): Departmentally purchased license for deep integration into Word, Excel, PowerPoint, and Outlook.
  • Compliance: Safe for University-Internal data and select University-Confidential data.

3. TAMU AI Framework (TAMU AI Chat & Chat API)​

  • TAMU AI Chat (chat.tamu.ai): Free web interface providing access to top commercial models (GPT, Gemini, Claude) simultaneously. Includes a daily token usage budget (equivalent to ~$10/day).
  • TAMU AI Chat API: OpenAI-compatible API designed for lightweight app prototyping, personal scripts, and internal automation.
  • Compliance: Fully approved for University-Confidential data (FERPA/HIPAA). Default tool for any sensitive work.

4. Commercial Subscriptions & Cloud Computing​

  • Commercial Accounts: Reimbursable only for public data use if under $25,000 and purchased through AggieBuy with CLEAR accessibility/security review. Personal accounts cannot be used for confidential data.
  • VISION Supercomputer: NVIDIA H200 GPU supercomputer designed for deep learning, research-scale model training, and simulations across the TAMU System.
  • Enterprise Compute Environment (ECE): AWS Bedrock, Azure AI Foundry, and Google Vertex AI provisioned for projects requiring custom infrastructure via FAMIS funding.

Matching Tools to User Workflows​

Group 1: Routine Work & Productivity​

  • Writing & Editing: Use Microsoft Copilot or Google Gemini inside office apps for draft generation. Shift to chat.tamu.ai if working with student grades or confidential evaluation notes.
  • Teaching & Materials: Use Gemini Notebook to upload lecture notes and automatically create mind maps, practice quizzes, or study summaries.

Group 2: Building Tools & Applications​

  • Prototyping: Build against the TAMU AI Chat API. Since it supports University-Confidential data, custom apps handling university data remain compliant.
  • Coding Assistants: Developers using Claude Code or ChatGPT should write code using sample/public data, then point final API calls to TAMU AI API endpoints when using real university data.
  • Key Security: API keys must be secured in TAMU's 1Password vault—never hardcoded or exposed publicly.

Group 3: Research & Advanced Computing​

  • Cost Controls: Token usage scales with text length, multi-modal features, and complex "reasoning" models. Always implement spending caps on un-capped APIs.
  • High-Performance Computing: Apply for beta/general access to the VISION Supercomputer (vision.tamus.edu) for large GPU-dependent model training or experiments.

Future Enhancements​

  • AI Application Sandbox: Upcoming deployment of an official sandbox environment for hosting AI tools and agentic workflows (incorporating Nvidia's NemoClaw).
  • TAMU AI Gateway: An API proxy service under development to provide centralized authentication, security, and cost governance across provider APIs.
  • VISION Supercomputer Rollout: Transitioning from beta to general availability by Fall 2026.

Official References & Terms​

  • Policies: TAMU System Policy 29.01.05 (AI), System Policy 15.99.08 (AI in Research).
  • Documentation & Web Portals:
    • TAMU AI Services: it.tamu.edu/ai-services
    • TAMU AI Chat: chat.tamu.ai
    • TAMU AI Documentation: docs.tamus.ai
    • VISION Supercomputer: vision.tamus.edu