Abstract

This article analyses ChatGPT from the perspective of virtue epistemology, arguing that, despite certain benefits, the tool is more prone to fostering epistemic vices than cultivating virtues. After outlining the central notions of virtue epistemology, particularly the distinction between intellectual virtues and vices, the paper assesses the epistemic impact of large language models (LLMs) in science and education. While ChatGPT may offer practical advantages such as speed, accessibility, and pedagogical support, it also encourages intellectual passivity, weakens memory, authorship, and critical thinking, and reproduces social biases. These dynamics generate what recent studies describe as “cognitive debt,” a progressive erosion of intellectual autonomy and responsibility. The article highlights that delegating epistemic tasks to ChatGPT often distorts purposes, privileging results over processes essential to intellectual flourishing. It concludes that the widespread and uncritical use of LLMs risks consolidating laziness, epistemic insouciance, and dishonesty, thereby undermining rather than enhancing our capacities as epistemic agents. Keywords: Virtue-based epistemology. ChatGPT. Intellectual vices.