What uses AI, and what does not
The record itself — courses, outcomes, offerings, grades, measures, narratives, mappings, imports from Assessment+ files, every program report and the self-study sections — needs no AI at all. The AI is there to read documents and draft text you would otherwise type:
- Filling a course in from its syllabus, and program outcomes or objectives from a web page or document.
- Course reports: Recommended Course Outcomes, Recommended Outcome Alignment, the Outcome Alignment Review, the Course Outcome Measurement Assessment, the Syllabus Evaluation, and parts 3 and 4 of the Course Assessment Report.
- Drafting narratives (✨ AI Suggestions…), evaluating a measure, drafting a course description.
- Reading old Excel workbooks or last year's report document, articulations and transfer evaluations.
The ADA Report runs entirely on this computer and never uses AI. Nor do any of the program reports.
Your key
- Choose an AI company: Claude (Anthropic), ChatGPT (OpenAI), Gemini (Google), Mistral or Grok (xAI).
- Get an API key from that company's developer site — console.anthropic.com (Claude), platform.openai.com (OpenAI), aistudio.google.com (Gemini), console.mistral.ai (Mistral) or console.x.ai (xAI). Sign in, find API keys, and create one. Most companies ask you to add a payment method or buy credit before a key works. A key is a code that lets a program use the AI on your account; the company bills you for what you use, usually a few cents a report. It is not the same as a chat subscription.
- In Assessment+: Tools ▸ AI Settings, choose the provider and paste the key. AI Settings has its own help on getting a key for each provider.
Keys are kept by AI Manager in Documents\AiManager, encrypted for your Windows account, and shared with the publisher's other programs (CourseBuilder+, Gamify+ …). They are never in the department folder or a backup, so they stay yours even when the folder is shared. Tools ▸ AI Usage and Cost shows what has been used.
Student privacy: nothing identifying is sent
Assessment+ is built on one rule: what goes to an AI must carry no student identifying information.
- Gradebooks are never sent. A gradebook is read on your computer to count grades and measures. Where an analysis needs to know about it, only the column headings and their points go — never a row, a name, an ID or a score.
- What Assessment+ assembles itself — outcome statements, grade counts, measure results — is checked anyway. If a student's details ever turned up in it, the request is stopped and reported as a fault in Assessment+.
- What you give it — a syllabus, a transcript, a pasted document, narratives you wrote — is checked for names, ID numbers and email addresses before it is sent. What was found is shown on the Student Information Found screen, each value with the line it is on. It is removed from the copy that is sent (your own files are never changed), or you send nothing.
- A value that is not a student — a word in an outcome that looks like a name — can be ticked Not a student. That tick is for that exact text, and lasts only until Assessment+ is closed.
- The check reads patterns, and patterns miss things. So you are also told, once for each provider, before anything is sent, and the screens that send carry a reminder. Read what you send, and never name a student in a narrative.
- Every AI screen has View Prompt, which shows exactly what would be sent. AI Prompt.
The Assessment+ privacy policy says the same in full.