Year: 2024
Type: book-chapter
Abstract: Increasingly, human-generated digital text is becoming more prevalent. Various forms of digital content, such as digital journals, scientific publications, and social media posts, are progressively being accompanied by formerly analog materials, as exemplified by the digitization of archival resources. This trend paves the way for the growing integration of information technology tools, including artificial intelligence models, across diverse fields in the future. At their best, AI models will enhance natural language processing, enabling the generation of extensive textual content without the necessity for manual scrutiny of each passage. This streamlines tasks for researchers, freeing them from mundane activities by delegating automation-friendly duties to machines. This evolution prompts a consideration of the specific tasks AI can address, the ways AI can facilitate them, and the resulting implications for contemporary and future research methodologies. However, alongside AI’s potential lies a gamut of threats that surface when the inherent limitations of AI remain inadequately grasped. In this manuscript, we engage in a contemplation of these dimensions and endeavor to elucidate the mechanics of AI models, with a particular emphasis on transformer models, through a research-oriented lens.
Authors: Toivanen, Ida, Poso, Venla, Lindroos, Jari
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7034652913