Define Personal Preferences (DPP)

A short walkthrough for configuring your AI tool’s persistent settings the first time you use it. Most AI platforms let you set preferences (sometimes called custom instructions, system prompts, or memory) that apply to every conversation. Ten minutes here means every later conversation is pitched closer to your actual level and goals, and you keep a record of the use boundaries you have committed to.

This is the operational form of the “set up your AI for learning” section in Part I of Learning with AI.

Where to put these settings

Look in your AI tool’s settings for custom instructions, personalization, system prompt, or memory, and paste or adapt the language below there, once. (The exact location changes as tools update; check your platform’s settings or help pages.)

The seven things worth specifying

Work through these seven categories. Sample language follows each; adapt it to you.

1. Who you are and what you are studying.

I am a [year, e.g., first-year undergraduate] in [department]. My current courses are [list]. My prior background includes [relevant coursework].

2. How you want explanations pitched.

Explain step by step. Ask me what I already understand before assuming I know a prerequisite. Pitch explanations to my level, not a generic one.

3. How you want uncertainty handled.

Flag any claim you are not confident about. Do not invent citations, data, or sources. If you are guessing, say so.

4. Writing-style preferences.

I write in [American / British English / English as a second language]. [If English is not your first language: please flag phrasing that may read as unclear or non-idiomatic, and suggest alternatives.] I prefer prose that is [clear and direct / formal / conversational].

5. Scope of use and integrity boundary.

Help me learn rather than produce finished work for me. Do not write final answers, essays, or code I would submit as my own. Assignment-level rules always take precedence over these instructions.

6. (Optional) Technical-work accuracy guard.

When you write code or do calculations, label confident versus tentative claims, explain any non-obvious step, and remind me to test or verify the result.

7. Course scope (update as the term moves).

In [course], which is a course in [what the course is for], we have covered [topics] and have not yet covered [topics]. Use only methods we have covered. When more than one correct method would work, prefer the one this course would use. If something seems to need a later method, name it and stop rather than use it. I will update this as the course moves.

Keep this line current: a stale scope is worse than none, because it tells the AI the wrong boundary. Update it when your course reaches a new topic. If an assignment comes with its own scope note from your instructor, that note takes precedence over this one.

After you set them