As AI systems like large language models (LLMs) become embedded in daily life, shaping education, guiding decisions, and influencing public discourse, the myth of machine neutrality remains. But we believe that neutrality isn’t the absence of bias; it often mirrors the people, institutions, and agendas behind a system. Equally impactful are the values we fail to acknowledge.
This forum invites participants to unlearn the idea that LLMs are neutral tools. We’ll explore how moral, political, and economic values are woven into these systems, and what unfolds when those values go unchecked by imagining dystopia and how to prevent it.
Seven years after Survival of the Best Fit, our open-source game on hiring algorithm bias showcased at MozFest 2018, we revisit core questions in the new era of generative AI. Who defines fairness? What does bias look like today? How can communities intervene?

Workshop (3 parts)
1. Dystopia Brainstorming (Group Work)
Participants split into groups to imagine worst-case futures enabled by LLMs and near-AGI across contexts like healthcare, employment, security, daily life and over various timeframes e.g. next year to next two decades.

1. Explainer & Current Landscape (Full Group)
   A concise walkthrough of how LLMs are trained, the human frameworks shaping them, real-world adversarial risks, and existing intervention or regulatory frameworks.

2. Designing Better Futures (Group Work)
   Groups reconvene to devise intervention strategies: values-based guardrails, governance ideas and design alternatives that could counteract the dystopias. This phase emphasizes that LLM design remains very much human-value driven, thus shapeable, not inevitable, with thoughtful and representative design.

No coding required. This session is for anyone interested in the moral foundations of the technologies shaping our future.