AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Quoting Armin Ronacher

Collected Oct 22, 2026

Simon Willison published a short post quoting Armin Ronacher, who remarked that as enjoyable as AI is, the industry spent roughly thirty years producing a great deal of bad software and is now, in his framing, being punished for it by the universe. The line is filed alongside tags covering security, generative AI, Ronacher himself, and AI security research plus LLMs.

The remark lands in a security context. LLM-driven tools consume enormous volumes of existing code, documentation and system interfaces, and older systems often carry exactly the accumulated shortcuts and inconsistencies Ronacher alludes to. This suggests the joke doubles as a diagnosis: AI agents inherit the technical debt of the software they are pointed at, and in practice much of the difficulty in deploying them involves coping with that inheritance rather than with the models themselves.

Why it matters: developers and product teams building on LLMs should expect to spend real effort on the quality and safety of the systems those models touch. A reasonable inference is that teams with cleaner interfaces and stronger security practices will get more reliable results from AI tooling than those sitting on long-untended code.

Read at Simon Willison

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

As fun as AI is, we built so much bad software over the last 30 years, and now the universe is punishing us for our sins. — Armin Ronacher Tags: security , generative-ai , armin-ronacher , ai , ai-security-research , llms