Choose Your Own Dystopia: How Our Decisions on AI Can Lead Us to Different Futures

The talk argues that utopias rarely reflect how things go in practice, so dystopian fiction is a more useful lens for thinking about artificial intelligence. The speaker opens with a short exercise in which the audience writes down as many "what if AI goes wrong" scenarios as they can in one minute, then sets up the rest of the session as a guided tour of those scenarios.

The middle section walks through a series of fictional futures, drawn from films, TV series, books and games. Each one maps to a plausible risk: strained natural resources and e-waste, mass job loss and inequality, surveillance and predictive policing, social scoring, AI in military systems, manipulation of memory and reality, emotional dependence on machines, questions of AI rights and accountability, loss of control over powerful systems, and over-reliance on technology that could leave people unable to work when it fails. The speaker stresses that these categories overlap and that the list is not exhaustive. The session is deliberately non-technical and aimed at a broad audience.

The closing part turns the stories into practical questions. Who makes the decisions, who bears the harms, whether the benefits outweigh the risks, and what safeguards exist are the questions to keep asking. For builders, the talk recommends bias testing, transparency, feedback loops and internal policies, and for everyone it recommends scrutinising the products they use and supporting sensible regulation. The Q&A touched on blaming AI for tax errors and on how reliance on technology can be limited.