CEU Electronic Theses and Dissertations, 2025
| Author | Dalton, Jim |
|---|---|
| Title | Can We Still Know? How Generative AI Undermines the Epistemic Role of Justification |
| Summary | This thesis investigates how generative artificial intelligence undermines the epistemic role of justification. The methodology consists of analytic investigation and engagement with contemporary epistemological literature. The thesis introduces a conceptual framework consisting of three key elements (justifiers, justifications, and beliefs) that clarifies how justification works with reference to empirical claims and is compatible with existing epistemological theories. Central to the thesis is the observation that generative AI can produce multimodal content (e.g. images, texts, audio, and video) that is perceptually and structurally indistinguishable from genuine human-generated content. This indistinguishability disrupts our ability to assess whether a justification is based on an appropriate justifier, thereby compromising two core epistemic dimensions: subjective belief formation and intersubjective evaluation of justification. The thesis further argues that traditional theories of epistemic justification (e.g. foundationalism, coherentism, evidentialism, and reliabilism) are mostly ineffective at addressing these challenges in practice. As a result, our capacity to function as responsible epistemic agents collapses, giving rise to a triadic form of skepticism: about reality, others, and ourselves. The culmination of these disruptions leads to what the thesis defines as the “post-knowledge era”, a novel epistemic condition in which even responsible knowers can no longer confidently determine whether they truly know something. The implications are urgent, raising significant philosophical and societal concerns. |
| Supervisor | Pascucci, Matteo |
| Department | Undergraduate Studies BA |
| Full text | https://www.etd.ceu.edu/2025/dalton_jim.pdf |
Visit the CEU Library.
© 2007-2025, Central European University