Usability Evaluation of the Local Large Language Models
Kivimäki, Teemu (2025-06-12)
Usability Evaluation of the Local Large Language Models
(12.06.2025)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
avoin
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2025061670380
https://urn.fi/URN:NBN:fi-fe2025061670380
Tiivistelmä
Local artificial intelligence (AI) features have recently gained remarkable attention, which indicates the growing need to recognize the existing challenges for wider adaptation of these technologies. Holistic approach is needed since the usability evaluation of locally installed LLMs requires a comprehensive view due to the challenges faced by different target groups and varying levels of configuration and implementation methods. Factors such as limited hardware resources and different user expectations need to be considered accordingly.
This thesis aims to recognize current advantages and challenges of using local large language models (LLMs) in average use environments such as homes and small to medium-size businesses and draw a broader view of the trend with methods such as heuristic evaluation of the user interfaces and by conducting a user study utilizing a general-use locally installed LLM (Meta Llama3-8B).
Regulations such as European Union’s General Data Privacy Regulation (Regulation (EU) 2016/679) create responsibilities for businesses to handle data in a secure and responsible manner. Likewise, the home users can also benefit from more confidential AI environments depending on their use cases. In this context, some of the key focuses of local LLMs strongly relate to privacy and data security. While using AI features, it is important that the end-user stays in control of their data and proper distinctions are drawn between local and cloud processing, as this approach allows users to make more informed decisions about their own data, thus enhancing a more user-centric approach to LLMs and AI more broadly.
This thesis aims to recognize current advantages and challenges of using local large language models (LLMs) in average use environments such as homes and small to medium-size businesses and draw a broader view of the trend with methods such as heuristic evaluation of the user interfaces and by conducting a user study utilizing a general-use locally installed LLM (Meta Llama3-8B).
Regulations such as European Union’s General Data Privacy Regulation (Regulation (EU) 2016/679) create responsibilities for businesses to handle data in a secure and responsible manner. Likewise, the home users can also benefit from more confidential AI environments depending on their use cases. In this context, some of the key focuses of local LLMs strongly relate to privacy and data security. While using AI features, it is important that the end-user stays in control of their data and proper distinctions are drawn between local and cloud processing, as this approach allows users to make more informed decisions about their own data, thus enhancing a more user-centric approach to LLMs and AI more broadly.