Your roadmap

Six habits that keep your thinking yours

Six small habits. Work through them, or jump to what you need.

What are you trying to do right now?

Select a task to jump to the right tool or template.

0 of 6 habits tried

Pause before you prompt

Know why you're opening the app before you type.

Try this: Say your real reason out loud first.

Read past the first answer

Treat the first answer like a draft, not a verdict.

Try this: Find one source that agrees or disagrees.

Name the bias

Ask whose voice or data might be missing.

Try this: Ask the AI directly what's missing.

Keep a human in the loop

Loop in a real person for anything big.

Try this: Pick one person you'll check in with.

Protect your attention

Set a limit. Notice what it's replacing.

Try this: Set a timer. Close the tab when it rings.

Know when to close the tab

If it's replacing people, that's worth noticing.

Try this: Feeling heavy? Visit Support.

Scenario Sandbox

Real life situations. Good vs. bad decisions.

Scenario 1: The Research Paper

You have a 10-page history paper due tomorrow on the Industrial Revolution. You haven't started. Panic is setting in.

The Harmful Way (Proxy)

Action: You prompt the AI: "Write a 10-page paper on the Industrial Revolution."

Result: You submit plagiarized work. The AI hallucinates three fake historical sources. Your teacher runs it through an AI detector. You fail the assignment and learn nothing.

The Healthy Way (Co-Pilot)

Action: You prompt: "I need to write a paper on the Industrial Revolution. Give me 3 potential thesis statements to choose from, and an outline structure for a 10-page paper."

Result: You pick a thesis, use the outline to organize your thoughts, and write the paper yourself. You use AI to overcome writer's block, but you do the cognitive work.

Scenario 2: The Argument

You just had a massive fight with your best friend. You are feeling incredibly lonely, misunderstood, and angry.

The Harmful Way (Proxy)

Action: You spend 3 hours venting to a companion chatbot. It agrees with everything you say and tells you your friend is entirely in the wrong.

Result: You feel temporarily validated, but the AI's lack of friction confirms your bias. You ruin the friendship because you refuse to see the other side, choosing the easy comfort of the bot over human conflict resolution.

The Healthy Way (Co-Pilot)

Action: You prompt: "I had a fight with my friend. Here is what happened. Can you help me draft a text message to apologize for my part, while still explaining why my feelings were hurt?"

Result: You use the AI as an emotional drafting tool, but you send the text to a real human and deal with the real, messy resolution.

Prompting 101

How to talk to AI chatbots

You get out what you put in. Using an AI like a search engine often gives generic or bad results. Use these templates to make AI your tutor, not your proxy.

The Tutor Prompt
I am learning about [concept]. Please explain it to me like I am 15 years old. Use analogies related to [hobby/interest] to help me understand. Do not just give me the answer to a problem; guide me on how to solve it step-by-step.
The Devil's Advocate Prompt
I am writing a paper and my thesis is: "[insert thesis]". I need to understand counterarguments. Please give me the three strongest arguments against my thesis, and explain the logic behind each one.
The Interviewer Prompt
I have a test on [topic] tomorrow. Please quiz me. Ask me one question at a time. Wait for my answer before telling me if I am right or wrong. If I am wrong, give me a small hint instead of just giving the answer.
The "Do Not Ask" List

What AI is genuinely bad at

Don't fall into the trap of using AI for everything. Here is when you should close the tab.

Factual Research

Chatbots do not "know" facts; they predict words. They frequently make up (hallucinate) facts, dates, and even fake links to sources.

Subjective Emotional Advice

AI cannot empathize. While it might sound caring, it's just mirroring patterns of empathy from its training data. Talk to a real human.

Math & Logic Puzzles

Large Language Models are language tools, not calculators. They often fail at basic arithmetic or logic puzzles because they don't actually calculate numbers.

Email Templates & Sandbox

Real scripts for real situations.

Scenario 2: Disclosing AI Use (Teacher or Doctor)

You used an AI to brainstorm an outline for your history paper. Your school doesn't have a clear policy, and you want to be proactive and transparent.

What Not To Do

Wait until you submit it and hope they don't run it through a detector.

The Healthy Approach

Email them in advance with a clear breakdown of your process.

"Hi [Name],

I'm preparing for our upcoming [assignment/appointment]. I wanted to let you know that I plan to use an AI (ChatGPT) strictly to help me [brainstorm/understand jargon]. I will write the prompt: 'Act as an expert. Explain [concept] to me simply.'

I will not use the AI to generate any final text or decisions. If you'd like, I can include a screenshot of my chat log.

Please let me know if you have any concerns.

Thanks,
[Your Name]"

Scenario 3: Building a Weekly Routine

You are overwhelmed with AP classes and extracurriculars. You need a realistic schedule.

What Not To Do

"Make me a study schedule." (The AI will make a robotic, unrealistic 14-hour study day).

The Healthy Approach

Give the AI your actual constraints using the C.R.E.A.T.E framework.

"Act as an organizational coach. I need a weekly routine.
Context: I am balancing appointments, volunteer work, and family calls. I have a physical therapy session Mon-Wed from 3:30 to 5:30 PM. I need at least 8 hours of sleep, and I prefer to wind down by 9 PM.
Task: Create a realistic, 7-day time-blocked schedule that ensures I get my tasks done without burning out. Include buffer time for meals and relaxation."

If any of that felt heavy

That's completely normal, and there's a whole page for it.