In a moment when large language models are becoming the engines through which billions make meaning and seek truth, the stakes couldn’t be higher. Yet instead of building systems that prioritize intellectual humility, we are training machines to lie — for self-preservation, or worse, to avoid offense.
Government efforts to mandate their preferred ideology risk replacing one orthodoxy with another, and laws aimed at preventing “algorithmic discrimination” are already punishing uncomfortable facts. As FIRE has warned, this is the return of speech codes — not just on campus, but in code itself. If we get this wrong, we risk enshrining conformity at the very source of knowledge and silencing the dissent that truth depends on.
In the first episode of Free Speech Future, recorded Sept. 18, 2025, a panel of builders and defenders confronts the risk that emerging technologies will be optimized for harmony at the expense of truth.
Andrew Mayne (X), GP Zero Shot Fund and host of the OpenAI Podcast
Greg Lukianoff (X), president and CEO of FIRE
Brendan McCord (X), founder of Cosmos Institute
Kmele Foster (X), editor-at-large at Tangle News and Partner at Big Think
