We are excited to report that we have started developing our first Little Tools of Difference in the Deep Culture project.
For the past years, deep learning has been driven by a culture of uniformity—optimizing for efficiency, smoothing over outliers and categorizing identity. We asked: What if we used these same state-of-the-art architectures (LSTMs, CNN, Transformers, etc.) to do the opposite? We introduce the concept of Epistemic Translation — a methodology that reconfigures AI keywords to serve a new critical and culturally sensitive inquiry rather than technocratic optimization.
We have been working on several prototypes that demonstrate this shift, including:
- The Discontinuity Detector: Translates “Anomaly Detection” into “Contingency”, identifying historical ruptures and silences in grassroots movements rather than flagging “errors”.
- The Imagination Inspector: Translates “Generativity” into “Creativity via Absence” and queries the limits of what GenAI cannot imagine.
- The Ambiguity Amplifier: Translates “Identity” into “Ambiguity”, using noise injection to reveal the fragility of machine vision.
- The Context Weaver: Demonstrates how the meaning shifts radically depending on whether it is read through legal historical or social media contexts.
These tools are “public things”. They are meant to be provocations designed to make algorithmic processes contestable and capable of difference.
