Bas Dijkstra, a test automation professional with 18 years of experience, published a blog post titled “I Am Tired of AI” on September 17, 2024, that has now surged to the top of Hacker News with over 1200 points and 1100+ comments. Dijkstra’s complaint is threefold: as a professional, he sees AI-powered testing tools producing faster but not better results; as a conference program committee member, he automatically rejects proposals clearly written by ChatGPT because they sound dull and identical; and as a human being, he finds AI-generated art, music, and social media posts boring compared to human creations. He acknowledges using AI sparingly but insists it cannot replace skilled human judgment.

The Hacker News thread reflects a deep community split. Animats argues that LLMs have shown their limits—useful but unreliable—and without a confidence metric the industry heads toward another AI winter. Datahack counters that dismissing LLMs echoes early dismissals of the Internet and mobile, and that the exponential improvement curve makes optimism warranted. YeGoblynQueenne points out that LLMs are not new but a direct descendent of Shannon’s 1948 statistical language models, suggesting we are past the early era. The comment section oscillates between these poles, with many agreeing that LLMs are good for narrow, low-stakes tasks but dangerous for anything important.

The post and its reception capture a growing sentiment among technical professionals: the AI hype cycle has produced genuine utility, but also exhaustion and skepticism about whether the industry is building what people actually need.