Unveiling Universal Knowledge: Fieldshifting Across Scientific Disciplines

A developmental biology lab in the US noticed that they could read papers in neuroscience, if they just substituted some words (neuron -> cell, neural -> cellular, behaviour -> morphogenesis). In this way, the number of papers that they could learn from increased dramatically. They pursued this practice in their group meetings, until 2023 when they realized the practice could be automated with LLMs. This not only sped up the process, making it possible to translate an entire paper automatically, but also opened up the technique to more disciplines. They found they could translate papers into philosophy and other disciplines, and still get useful new knowledge out of them.

In this talk I will discuss the implications of this practice for freeing scientific inquiry, and what it means about the deep commonalities in knowledge. I will also demonstrate an online tool that can be used for interactively fieldshifting between diverse disciplines with iterative feedback. This maximizes the intellectual distance the user can cross, as the fully automatic tools are somewhat brittle.