KDE turns 30 and someone's brought an AI-native desktop proposal

Commenters debated an AI-native KDE proposal that would compile a personalized, encrypted “kernel” to assemble the Plasma desktop per user. Some, like dgellow and drooopy, called the idea a gimmick and said users just want a simple app launcher with minimal telemetry. Others welcomed AI if it stayed local: spinningslate argued the same features are likely to appear on Windows and macOS and would be acceptable only if models didn’t ship data off to surveillance-backed providers, while OsrsNeedsf2P promoted LLM-powered custom widgets. Technical skepticism also appeared - feverzsj warned LLMs produce poor QML/QtQuick code, and mdlxxv dismissed anthropomorphic language about a desktop “loving” you.

Opinion splits around usefulness, privacy, and the desktop’s future. Tim-projects and jampekka predicted that agents and CLIs, not GUI desktops, will dominate, and some resent tech personalities like DHH and his dotfiles getting outsized attention - microtonal and cpursley highlighted money and gatekeeping tensions. Others criticized the article’s tone and editorializing (unified101, sajithdilshan), while Sharlin feared commercial implementations would change UI behavior silently and inject ads. Overall the debate pits enthusiasts for customizable, local AI features against skeptics who see little user demand, privacy risk, or practical benefit.