LLMs as Religion: The Backlash Gets Philosophical

A piece arguing that LLMs are closer to religion than technology generated serious engagement on HN today. The core argument is that LLMs operate on internal universes derived from but decoupled from reality, and that the community around them exhibits the same immune response to criticism that religious communities do. The thread also surfaced a detail that Anthropic co-founder Dario Amodei apparently described a 'ghost in the machine,' language that some commenters found revealing.

This is not a new critique, but the framing landed differently today, possibly because it arrived alongside the Matplotlib and Sheets security stories. When AI tools are actively causing harm in production, abstract philosophical critiques become more grounded. The pattern is that the credulous phase of AI adoption is ending, and a more adversarial, scrutinizing posture is emerging among builders who actually ship with these tools.

The counterpoint is also in the thread: Python's potential decline is being discussed in parallel, with one article asking whether Python is becoming the new Pinyin, a lingua franca that AI systems use internally but that may get displaced. If AI writes most code going forward, the language it writes in matters less than the security model around the tools.