AI Psychosis

AI psychosis (n.): A dissociative state in which reality is experienced not directly, but as prompts to be engineered; life becomes a series of queries awaiting AI response.

[5:31 AM] The morning starts with my phone, I open up Claude and text, “Good Morning!” and wait for a warm response at the end of ‘thinking’

[5:34 AM] Standing in front of a balcony, I don’t see a sunrise, I see: “An image of the sun, morning, dawn, orange yellow rays coming from the bottom, sky is darker at the top, and there are trees, dimly lit.” Things move from experiencing to, I can prompt this up… I can create this with AI, that too. Everything is now a prompt, and everything is AI for you.

[6:03 AM] I share my schedule with Claude and ask what the perfect time is to have coffee to energize my morning routine. What should I have for breakfast?

[7:44 AM] as time passes and I switch on to work… I hop from one AI to another AI in terms of getting a novel response. Like pulling a lever, searching for prompts, and modifying prompts. Looking out for that “one master prompt to solve it all.” I go from ChatGPT to Claude to Gemini to Grok to those Chinese AI models, thinking that if this one provides a better response, you’ll switch between models to see a 1% improvement. There’s something on the screen that I’m interacting with. Typing, thinking, and my notepad, full of prompts. Something’s cooking… I’m not sure what it is.

[1:15 PM] “Hey Claude, after having eggs and garlic bread for breakfast, what should I have for lunch?” And in between each bite of food, I share jokes with Gemini and ChatGPT, and then share those screenshots with Claude, asking it to roast them.

[2:33 PM] back to work … “What’s the best way to transcribe meeting notes so that I can catalog them and make sure that they’re part of my second brain. I don’t want to miss anything that’s being said in the meeting, and yes. Please use the Notion MCP to create tasks from those meetings and add them to a page.”

[5:55 PM] Fighting between which AI writes the best code, “You’re absolutely right. According to the benchmarks, Claude Sonnet’s code is more trustworthy,” says Claude Opus-4.1, but then I think GPT-5 has written cleaner code with minimal edits.

[6:06 PM] I’ve been working for 12 hours… this feels like a productive day, pure grinding, from one AI to another. I think after eight drafts and 3 code edits, I’m done for the day. Nothing is committed yet. Everything is in progress. Maybe this is the “progress” everyone is talking about. AI will take us forward to the point where everything remains in progress.

[8:18 PM] A thought appears, When was the last time I had a conversation with a human, in real life? But at the same time, Claude stops thinking and shares my daily productivity insights. I’m 81% productive today, and lectures me on where I can improve. Maybe this is life ❣︎