Simulacrum of Knowledge Work | Hacker News

Hacker News commenters debated whether AI-generated knowledge work is mostly convincing-looking but shallow output, or just the latest version of a longstanding quality problem.

Some argued that humans also produce bad work, so relying on visible mistakes has always been a weak proxy for judging substance.

Others said reviewers now need deeper checks and better verification methods, especially as polished LLM output becomes easier to produce.

The thread suggested that improved models, stronger evaluation, and domain-specific validation could change how knowledge work is assessed.