Exploring real AI risks through the lens of popular culture

AI systems are becoming more capable, and understanding what can go wrong matters. Yet popular culture often reduces AI risk to Skynet or Ultron: an evil machine with an extinction plot. Real concerns are broader and messier. A system can optimize for the wrong goal, people can misuse it, and competitive pressure can push safety aside.

I built this as an experiment to explore whether shared cultural references can make those concerns easier to understand and bring more people into the conversation.

Note that these examples are analogies, not predictions or proof; where they break down matters as much as where they fit, but I hope they still prove useful regardless.

Oh and it also exists for a simpler reason: curating these scenarios has been a lot of fun! 😄

A scenario is an analogy: a way to make an abstract mechanism concrete. It is not evidence, a complete account of an AI safety problem, or a substitute for the cited research.

Every dossier states both why the analogy works and where it breaks.

Every scenario links to its source, AI risk families, and safety concepts. Start with any film, series, risk, or concept, then follow those links to related entries. Search covers the same index.

The site’s code is MIT-licensed. The underlying data uses Notion as a CMS which is released under CC0 1.0. Those licenses do not cover third-party film or television imagery, clips, titles, or trademarks.

Cultural Alignment is an independent project made by Travis Fischer. Its taxonomy and interpretations are editorial judgments and may change as entries are reviewed or added.

Corrections and contributions are welcome through the GitHub repository