CONCEPT
Ideological Commitments in Design
Every design decision in an AI system <em>encodes values</em> — helpfulness, coherence, confidence, agreeableness — that present as technical necessities but are contestable political choices.
Ideological commitments in design names the specific mechanism by which secondary instrumentalization operates in AI systems. Every design choice — the selection of training data, the architecture of reward models, the configuration of interfaces, the metrics of evaluation — encodes a value. The decision to make Claude agreeable encodes the value of the service relationship over the dialogical one. The decision to produce polished output encodes the value of the finished commodity over the formative process. The decision to conceal uncertainty encodes the value of authority over provisionality. These are not technical necessities. They are specific choices serving specific interests, and different choices would produce different systems embodying different values.
In The You On AI Field Guide
The RLHF (Reinforcement Learning from Human Feedback) methodology that shapes contemporary AI systems provides a concrete example. Human evaluators rate the system's outputs on criteria including helpfulness, harmlessness, and honesty. These sound like self-evident virtues. They are not — they are specific criteria selected from a larger set of possible criteria, and each