X Design Head Benji Taylor: No Longer Thinks Taste is a Permanent Boundary Large Language Models Cannot Cross
X and SpaceXSI design head Benji Taylor stated that he no longer believes taste is a permanent boundary that large language models (LLMs) cannot cross. He believes this crossing has already occurred.
On October 13, Taylor posted on X: "I no longer see any compelling reason to believe that taste is some permanent boundary LLMs won’t cross. It’s already happening." The post received thousands of likes and hundreds of replies. Taylor is currently responsible for design at X and SpaceXSI. He previously led the design at Coinbase Base and founded Los Feliz Engineering, which was acquired by Aave.
Taylor had previously stated that craftsmanship is a differentiating factor in an era of mediocrity. This statement shifts the notion of taste from a permanent human boundary to one that can be crossed by models. The discussion involves the aesthetic quality of model-generated content and data issues. Some responses suggest that taste is a locatable potential space embedding.
Funding is shifting from viewing taste as a human-exclusive differentiation to judging that models are crossing that boundary. Design work is transitioning from relying on personal taste to coexisting with model outputs. The narrative of "taste being irreplicable" is under pressure. The performance of models in visual and interface generation is presented as evidence.
Source: Public Information
ABAB AI Insight
Benji Taylor has transitioned from design entrepreneurship to Aave CPO, then to the design head at Coinbase Base, and now to X and SpaceXSI, focusing on design. He has long been concerned with interface craftsmanship and user experience. Previously, he viewed craftsmanship as a means of differentiation.
He directs resources and judgments towards the idea that models are crossing the boundary of taste. His post directly denies taste as a permanent human boundary. The discussion has shifted to the aesthetic level of model-generated content and data-driven improvements. His role in design at X and SpaceXSI connects this judgment to actual product interfaces.
Similar design heads have shifted from viewing taste as a human boundary to believing that models are crossing it, a path that has appeared in other discussions of AI-generated content. We are currently in a phase where models are continuously improving in visual and interface generation, and the narrative of taste is being reassessed.
Essentially, this belongs to technological substitution. The capabilities of models replace the assumption that "taste is uncrossable." The mechanism lies in the training data and improvements in generation quality, where aesthetic judgments and style outputs are no longer pre-set as permanently exclusive to humans, allowing design work to shift from relying on personal taste to collaborating or competing with model outputs, while positioning craftsmanship and context as remaining differentiation directions.
ABAB News · Cognitive Laws
- Taste shifts from a permanent human boundary to the judgment that models are crossing it.
- Craftsmanship as a means of differentiation needs to be repositioned after the enhancement of model capabilities.
- Design heads publicly deny the uncrossable nature of taste, bringing the aesthetics of generated content into a verifiable range.