Guest commentary: AI? On the gradual formation of thoughts through writing for oneself

Artificial intelligence may make everyday writing easier, but the convenience comes with a cost, argues Benedikt Haufs. When people surrender the free play of their own minds, they also risk losing something of their vitality and personality.

Haufs points to several signs that a text may have been generated or structurally altered by AI. Its language may lean toward conspicuous pathos, stock phrases and repetition. Ideas can recur in both content and structure, sometimes almost endlessly. Such writing may also lack detail and provide weak sources.

The deeper flaws appear when the text is read carefully. AI-written passages can contain fine cracks in structure and chronology, with pieces of information colliding because the links between them are missing. On specialized subjects, the system may use isolated details to construct a plausible-looking picture without recognizing that the information is outdated, incorrect or limited to one perspective.

Even an originally human-written text can change fundamentally after AI editing, Haufs says. He compares this to an organic farmer whose neighbor secretly adds carefully measured pesticides to the field. The text may still carry the author’s name, but its structure has become an AI product.

The commentary frames the risk as a gradual loss of self-reliance. Earlier generations worked routinely with their hands, while that everyday ability has already declined. Now intellectual work is increasingly being outsourced as well. Without repeatedly searching for ideas and forming them independently, people may lose their feel for things, including the meanings hidden between the lines. Haufs compares developing a text with one’s own mind to a stonemason’s work, extracting substance from what one has personally thought and gathered.