Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Gammalt möter nytt: SVT Nyheter, artificiell intelligens och migrationsperspektiv.

Fornbacke, Vilhelm LU and Norberg, Edvin LU (2026) STVK04 20252
Department of Political Science
Abstract (Swedish)
Migration has been a contested issue in Swedish politics over the past decade, and media coverage has repeatedly sparked debate about the impartiality of Sveriges Television (SVT). Using large language models (LLMs) as coding instruments, this study examines how migration has been framed in SVT Nyheter, a major Swedish public service news outlet, between 2014 and 2024. Rather than explaining causal mechanisms behind perceived bias, the study describes patterns of political framing over time and assesses whether these patterns align with the neutrality and impartiality requirements stipulated in SVT’s broadcasting licence. The findings suggest that SVT Nyheter’s migration reporting varies over time rather than exhibiting a stable... (More)
Migration has been a contested issue in Swedish politics over the past decade, and media coverage has repeatedly sparked debate about the impartiality of Sveriges Television (SVT). Using large language models (LLMs) as coding instruments, this study examines how migration has been framed in SVT Nyheter, a major Swedish public service news outlet, between 2014 and 2024. Rather than explaining causal mechanisms behind perceived bias, the study describes patterns of political framing over time and assesses whether these patterns align with the neutrality and impartiality requirements stipulated in SVT’s broadcasting licence. The findings suggest that SVT Nyheter’s migration reporting varies over time rather than exhibiting a stable directional bias. When deviations from neutrality are observed, they tend to consistently favor migration positive perspectives rather than anti- migration ones. While differences between LLMs are present, the overall trends are consistent across models. The study contributes to political communication research by demonstrating both the potential and the limitations of LLMs for quantitative text analysis in political science. (Less)
Please use this url to cite or link to this publication:
author
Fornbacke, Vilhelm LU and Norberg, Edvin LU
supervisor
organization
course
STVK04 20252
year
type
M2 - Bachelor Degree
subject
keywords
Artificiell intelligens, språkmodeller, public service, SVT, migration, invandring, mediabias, opartiskhet, partiskhet, framing
language
Swedish
id
9216876
date added to LUP
2026-01-26 11:47:04
date last changed
2026-01-26 11:47:04
@misc{9216876,
  abstract     = {{Migration has been a contested issue in Swedish politics over the past decade, and media coverage has repeatedly sparked debate about the impartiality of Sveriges Television (SVT). Using large language models (LLMs) as coding instruments, this study examines how migration has been framed in SVT Nyheter, a major Swedish public service news outlet, between 2014 and 2024. Rather than explaining causal mechanisms behind perceived bias, the study describes patterns of political framing over time and assesses whether these patterns align with the neutrality and impartiality requirements stipulated in SVT’s broadcasting licence. The findings suggest that SVT Nyheter’s migration reporting varies over time rather than exhibiting a stable directional bias. When deviations from neutrality are observed, they tend to consistently favor migration positive perspectives rather than anti- migration ones. While differences between LLMs are present, the overall trends are consistent across models. The study contributes to political communication research by demonstrating both the potential and the limitations of LLMs for quantitative text analysis in political science.}},
  author       = {{Fornbacke, Vilhelm and Norberg, Edvin}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{Gammalt möter nytt: SVT Nyheter, artificiell intelligens och migrationsperspektiv.}},
  year         = {{2026}},
}