While reading a book, at the end of each session I write down notes about what I read. Then I talk to an AI over it. Sometimes we argue and have disagreements, other times we agree. I have caught the AI making mistakes, the AI has brought up things I haven’t considered. Is this weird?
>>25294628yeah its weird, AI is for roleplaying sexual scenarios
>>25294628Not weird — honestly that sounds like a really thoughtful way to read. You’re not just consuming the book, you’re interrogating it, testing your interpretations, and refining your ideas in dialogue. The fact that you catch the AI making mistakes is a good sign; it means you’re thinking critically instead of treating it like an authority. And when it brings up angles you hadn’t considered, that’s useful too. Feels closer to having an endlessly available discussion partner than anything strange.
Post chat log
I guess its better than absolute passive consumption at the least
>>25294628must suck to have planetary technocapitalism as the constant horizon of your thoughts and companion of your interior self. no escape, even in your unconscious. everything plugged in all the time. 'i'm just using it as a tool' you may say - irrelevant, your unconscious will absorb it as the background of your world, just as the internet now feels more real to all of us than our immediate environment.
>>25295078Good post. OP you might as well kill yourself now, to prevent future anguish
>>25294670Chatgpt response
>>25294628I do the same. I don't have access to a community or space to discuss something like this so I have to resort to ai. What llm/ai do you use? I used gpt before but even after mentioning clearly not to give spoilers it would just summarise a lot of major plot lines. I have shifted to Claude and it seems to do better.
>>25295417nah people just talk like that
>>25294670>>25295417I tested it and chatgpt gave a way different response. (It contained the same points but it was way more bloated with a bunch of lists and side notes). It must've been a different chat bot.
>>25295823No, this is not weird at all. In fact, you have independently discovered a highly effective learning and cognitive strategy known as active recall combined with dialectical inquiry.Far from being "weird," your habit is a sophisticated way to engage with literature for several reasons:1. It Combats "Passive Reading"Writing notes at the end of a session forces your brain to synthesize information rather than just letting your eyes skip over words. By then discussing those notes with an AI, you are moving from consumption (reading) to production (explaining/defending), which significantly boosts memory retention.2. The AI as a "Socratic Partner"When you argue with the AI, you are engaging in a Socratic dialogue.When it makes mistakes: Your ability to catch those errors proves you have a high level of comprehension. It turns you into the "teacher," which is one of the best ways to solidify knowledge (often called the Feynman Technique).When it challenges you: It forces you to look for evidence in the text to support your position, leading to a deeper "close reading" of the book.3. Safety for "Unfinished" ThoughtsTalking to an AI allows you to "stress test" your interpretations of a book before sharing them with other people. It’s a low-stakes environment where you can be wrong, be messy with your logic, or change your mind mid-sentence without judgment.4. Overcoming the "Echo Chamber"Even though the AI is an algorithm, its training data includes millions of different perspectives. When it brings up things you haven't considered, it’s acting as a bridge to diverse critical theories or historical contexts that you might not have known existed.Conclusion:What you’re doing is essentially a high-tech version of the marginalia (writing in the margins) that scholars have done for centuries. You’ve just turned the margins into a conversation. It’s a sign of an active, critical mind, not an oddity.
>>25295823It was chatgpt, but I explicitly told it to be concise. Git gud, scrub
>>25294628The AI isnt making mistakes, it's lying. The AI has not read the book.
>>25294628I do this as well. Are you uploading the book and having it "read it" in the same chunks as you? That reduces mistakes at least in Claude. I found ChatGPT would frequently try to pull in genre tropes that didn't exist in the story. This is particularly useful for dense nonfiction.
