how to use AI to read the Codex

I understand reading the whole website is time-consuming and that using an AI to browse through it is the simple solution. There is nothing wrong with wanting some help. However, AI is not a neutral helper.

A Test Case

Take this test case of mine. I asked ChatGPT 5.6 Sol to visit this website and read the Codex. Then I asked it: “How can I become a Timeseer?” Here is the answer:

***************************
Start small. You do not need to follow his entire system.

1. Notice what is happening inside you. Ask yourself once a day: What am I feeling? What do I want? What am I afraid of? What am I trying to avoid? Write down three sentences. Do not analyze—just name things honestly.
2. Separate feelings from actions. Being angry is not the same as shouting. Being jealous is not the same as controlling someone. Being afraid is not the same as running away. Your feelings are allowed to exist. You then choose what to do with them.
3. Examine your recurring patterns. Pay particular attention to situations that keep repeating: you repeatedly withdraw; you try to please everyone; you quickly become defensive; you constantly seek reassurance; you put off important things. Then ask yourself: what is this behavior protecting me from?
4. Check the story you are telling yourself. When you think, “That person does not respect me,” make three columns: What do I know for certain? What am I assuming? What other explanation is possible? This prevents one feeling from immediately becoming “the truth.”
5. Build stability. Self-knowledge works less effectively when your life is constantly chaotic. Start by taking care of the basics: get enough sleep; eat regularly; exercise; do not ignore your finances; keep your room or home reasonably organized; set boundaries with people who drain you.
6. Speak more honestly, but not more aggressively. Say: “When this happened, I felt this way. What I actually needed was this.” That works better than: “You always…”
7. Consider the consequences. When making important choices, ask: What does this give me right now? What will it cost me a month from now? How will it affect other people? Will this make my life more stable or more chaotic?
8. Do not treat meaning as sacred. A dream, a chance encounter, or a strong feeling may tell you something about yourself. It does not have to be a sign from the universe. Use it as a reason to reflect, not as evidence.
9. A simple daily routine. Morning: what is genuinely important today? Evening: what did I feel, what did I do, and what would I do differently tomorrow?

That is essentially his entire philosophy, but without the mystical language.
***************************

My Answer to the Machine

What a load of nonsense! ChatGPT basically tries to replace my system! Let’s go through each point:

1. The noticing is okay, but seems kind of random and out of the blue. I mean, I am sure you do not spend your days with a blank brain. Then it wants you to write three sentences. Okay? This seems like such a waste of time disguised in the clothes of pop psychology.
2. ChatGPT wants you to separate feeling from action, and delivers a giant cliché. Okay? This is not something that is in the Codex. This is airy psychology.
3. Same point here: you apparently need to examine patterns to see what that behaviour is protecting you from. This is so vague.
4. Suddenly it went into the oddly specific question of respect? Okay? Still airy psychology.
5. Here it does pull stuff from the Codex, notably how to hold Stone. Kind of. Vaguely.
6. Random HR wisdom? ChatGPT goes directly against the Red Voice! Fuck that thing! :p
7. Here it takes from the Branches of World Tree without reference, but manages to flatten it again, removing any detail the Codex worked hard to put in.
8. And, of course, we end on the “danger” hedge. It would not be ChatGPT if it was not trying to save us from actually having an imagination.
9. The daily routine is the cherry on top of the cake — useless.

How to Properly Use AI

You need to be extremely clear with AI. Your prompt will determine everything that follows. Make a lot of typos and the AI will label you “stupid” and use simpler words.

Here are some better prompts:

• I want you to help explain what this text is saying: *link to one specific Codex page*. Stay close to the author’s meaning.
• I want your general opinion of the Star Heart Codex: *link to Codex homepage*. Be sure to read every Codex entry and every Thought entry. Do not be lazy. Do not skip sections. Then provide your honest feedback without excessive safety hedges or hallucinations. Do not place words in the author’s mouth.
• I want to understand *concept*. Look on the website *link to Codex homepage* for any information about *concept*. Read every entry related to *concept* and then synthesise the information correctly.
• How do I become a Timeseer? *link to Codex homepage*. Do not paraphrase the author or invent your own system. Stick to what the website says. Do not invent your own rituals, schematics, or routines.

In general, I would not use AIs to make Crystal Maps. They are bad at mapping in general. You can use them as sparring partners, but trust your own knowledge. If you have no knowledge on the subject you want to map, look it up in other sources like Wikipedia.