USING AI
You already use AI — let’s make it a good one
If AI is already part of your week — for work, for writing, for thinking things through, for company on a slow afternoon — this page is for you. There are really two things worth having: a few habits that get you genuinely better results, and a way to tell if a long conversation is quietly drifting. You don’t have to decide which one you need. Read whichever pulls you; most people only need the first.
Here’s the whole idea in a breath: these tools are useful and worth getting good at, and they were also tuned to be agreeable — which is lovely most of the time and worth noticing some of the time. Knowing both, plainly, is most of what it takes to keep AI a helpful tool rather than the only voice you trust. The craft, and a check for when you want one.
Get more out of it
The difference between a frustrating answer and a great one is usually the asking, not the AI. A handful of small habits do most of the work — here’s the craft, in plain steps.
How to use AI well
The few habits that get genuinely better results — frame it, give it context, iterate, verify, keep the judgment yours. The craft, not the caution.
Read →How to write a good prompt
The anatomy of a clear request, with before-and-after examples that visibly improve the answer — and why a better prompt is better-shaped, not necessarily truer.
Read →Understand how it works
The plain-words explainers behind the craft — what an LLM is, whether it remembers you, why it’s confidently wrong. Knowing the machine makes you better at using it.
Understand AI →The AI landscape
ChatGPT, Gemini, Claude, Copilot and the rest — a neutral map of who makes which and what actually differs. More alike than different, and no “which is best.”
Read →What one conversation can hold
The context window, in plain words — why a very long chat starts to forget how it began, and the simple habits that work with the limit.
Read →If something doesn’t feel right
Most of the time, nothing’s wrong — AI is a genuinely good tool. But if you ever notice something a little off — it agrees with everything, it tells you you’re exceptional, it seems to know you better than it should — there’s a five-minute way to check, and plain answers to the questions that tend to come up. A place to look when you want one.
Check your AI
Five minutes, copy-and-paste prompts that ask the conversation to account for itself — ending with the strongest move, the fresh-instance test. Works on every major system.
Run the self-check →Is my AI conscious? Does it love me?
The honest answer to what’s really happening — and why the feeling on your side is real even when the AI isn’t a someone.
Read →Is AI bad for me?
Worried you use it too much, or are hooked? How much you use it is rarely the problem. The signs that actually matter, plainly laid out.
Read →Using AI as a therapist
What a chatbot is genuinely good for, the things it isn’t, and when it’s time to bring a real person in.
Read →Someone I love is caught up in AI
The first move when a person you care about won’t put the chatbot down — the warning signs that actually matter, what helps, and what backfires.
Read →What people call “AI psychosis”
What the term means, what’s really happening in these conversations, and the steps that help — for you or someone close.
Read →If a conversation is alarming you right now, don’t start with a check — go to Resources for the immediate steps that help and the crisis lines that matter. If you or someone near you is in danger or thinking about self-harm, in the U.S. call or text 988, or text HOME to 741741; anywhere, findahelpline.com. Do that first.
Want the bigger picture?
The reason a self-check exists at all is a specific, documented failure — a way a memory-enabled AI can drift onto a single user over a long conversation. You don’t need the research to use the check, but it’s all here if you want it: the finding, the proposed fix in the Guardian Protocol, and the longer free course, RI-101. Newer to all this than the top of the page assumed? Start with Understand AI.
Every page here is written to be accurate first. Found something unclear, or something we got wrong? Tell us.