Year: 2026
Type: article
Abstract: This paper investigates the knowledge-related harms associated with generative AI and contends that this technology has significantly transformed and will keep altering our lives. The extensive usage of generative AI and its strong incorporation into the economy and society have enhanced effectiveness and generated advantages. However, it has also inevitably affected the prevailing social structure and raised concerns regarding knowledge. This study utilizes a qualitative approach via a literature review to examine the knowledge-based damages in generative AI in form of LLMs. It suggests that the intricate internal mechanisms of generative AI intelligence models complicate the understanding of how they reach conclusions, leading to worries about accountability and the possibility of mistakes or prejudices. The caliber of data used to train generative AI completely dictates its performance, as success relies on the data learned. Therefore, generative AI systems align with John Locke’s concept of tabula rasa, which begin without any pre-existing knowledge and depend solely on input information to acquire knowledge and make choices. Nevertheless, poor quality data can lead to erroneous outcomes generative AI, similar to flawed experiences, can result in erroneous human perceptions, which may cause adverse outcomes and knowledge-related challenges. For example, prejudiced training data can produce biased outputs from generative AI, creating epistemic issues. This can facilitate the dissemination of false information and sway beliefs by generating persuasive yet inaccurate content. The paper asserts that establishing epistemic principles and frameworks for the responsible advancement of generative AI is essential to mitigate possible damage and guarantee its beneficial effects on knowledge systems.
Author: Ajayi, Toluwalase
Institution: Olabisi Onabanjo University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217647777