Modern LLMs can rehydrate whole essays from tiny prompts. Makes me think that the prompt should be the work itself. Why write a whole essay (or book!) when you can craft a bulleted list that reinflates to the whole thing? #ai #asknostr

We are in the “horseless carriage” era of #AI. Everyone is building horses faster and faster horses (web apps). The real paradigm shift happens when we drop all that vibe coded bullshit and empower the AI to just do the tasks itself.

First-order effect: Everyone’s learning to vibe code, creating a glut of apps. Second-order effect: No one wants to learn and use your app. They just want AI to do it. Third-order effect: Companies are incentivized to deprioritize web apps and mobile in favor of MCP servers. Conclusion: get good at MCP. #ai

If there’s one thing I’ve learned from #AI, it’s that all my ideas are absolutely correct.

“Read the Fucking Manual” (RTFM) was a completely viable response to user error before AI. No longer. Reading docs is for the AI to do. Your app should be able to read its own docs and explain to the user whatever they need to know. This is the new bar. #ai

The problem with middle curve software engineers is that they think their job is to write code. It’s not. Your job is to produce VALUE. The code is just a vehicle. The prompts are the software now, and it’s your job to engineer it. #ai #vibes #meme

In the age of AI-assisted programming, a “documentation first” strategy makes sense. That is, before you write the code, start with a code comment describing the code. // The following function takes // an array of strings and returns // a new string that concatenates // the list with an optional separator. Then let the AI autocomplete the code for you. #programming #ai

#programming #ai #philosophy Programming is the art of turning fuzzy requirements into hard computer instructions. Translating a poorly defined “want” into code. As higher-level interfaces emerge, increasingly mediated by AI, the job of programmer is still the same, but increasingly philosophical. What is it that you (your manager, your customer) really want? What should happen when things go wrong? What does value mean in this context? These are programming questions in the age of AI. It’s less about syntax, and more about satisfying contextual human desires.