Study: Large Language Models Generate "Kitsch"
A new preprint study from arXiv claims that large language models (LLMs) systematically generate works that can be classified as "kitsch", a result of their training methodologies.
Vad har hänt
A study published as a preprint on arXiv entitled "LLMs Generate Kitsch" presents the argument that large language models (LLMs) tend to produce material that is kitschy. The researchers suggest that this is a direct consequence of how these models are trained. The study aims to explain the tension between generated content often being rated highly in controlled studies while simultaneously being perceived as generic.
Key facts
| Källpublikation | arXiv cs.CL |
|---|---|
| Studiens titel | LLMs Generate Kitsch |
| Ämne | Generativ AI, Kitsch, Språkmodeller |
| Publikationstyp | Preprint |
”Large Language Models (LLMs) are increasingly used to generate pictures, texts, music, videos, and other works that have traditionally required human creativity.”
”We propose to resolve this tension by arguing that LLMs systematically generate kitsch, and that this is a consequence of the way in which they are trained.”
”We also show empirically that readers perceive LLM-generated stories as kitschier, if we control for their definition of "kitsch".”
Varför det spelar roll
The phenomenon of "kitsch" is described as work that is appealing but lacks depth or originality. If LLMs systematically generate kitsch, it impacts their utility in creative domains such as art, literature, and music. It raises questions regarding originality, copyright, and the future role of human creativity versus machine generation.
Vem påverkas
Researchers and developers within AI and NLP are affected, as the study questions current LLM training methods. Users of generative AI, particularly in creative professions, need to understand the limitations of the generated output. Consumers of AI-generated content are also affected, as they may come to expect a certain generic quality.
EU-status
Ej relevant för EU-status.
Mer att veta
The study is a preprint on arXiv, which means it has not yet undergone a full peer-review process. The results should therefore be regarded as preliminary.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilka bolag berörs?
Länken öppnar i nytt fönster och leder till utgivarens egen sida.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.