In his recent paper, Metcalf (2026) argues that, at least in the humanities, widespread AI-generated articles are likely to result in a kind of erosion of the knowledge base of an institution or system. Building on this framework of technical incrementalism, this paper complements it with the commodification of the image of knowledge, the integration of spectacles, and Guy Debord’s algorithms to map two poles of LLM use in scientific authorship, from minimal technical assistance to the complete delegation of the writing process to machines. The massive use of LLMs encourages the production of technical, homogenous, and innovation-poor scholarly work, an undeniable common condition today. Behind this, academic economic pressures and publication incentive systems exacerbate this, particularly in philosophy and the humanities. Therefore, solutions to this problem cannot stop at the individual level. Possible long-term solutions include structural changes to the scientific publishing ecosystem, assessments based on contributions to knowledge, and fair incentive systems.
