When everything looks polished, what can we trust?
Kyle Kingsbury, known for testing databases with Jepsen, wrote that LLMs erode trust in documentation, in software and in the companies that make them. The signals we used to judge a mature project are now easy to fake, he says. Simon Willison shared the quote.
Other stories echo him. Eurogamer reports that AI-written fan wikis full of errors are muddying games culture (83 points on Hacker News). A blog post titled 'So I guess everyone uses LLMs to think now' drew 112 comments. The top AI story on Hacker News today, Nolan Lawson's 'The AI cost of JavaScript frameworks', got 376 points and 292 comments; our source gives only the title.
Even the tools react. 'unslop', a Claude skill that turns 'AI-looking' web design into a real designed interface, got 5,320 GitHub stars.
When anything can look professional, looking professional stops meaning quality. For normal people, it means double-checking wikis, docs and reviews. For builders, it means proving their work, not just presenting it.
