Development of a Browser Plugin to Simplify Working with Language Models

The aim of this bachelor's thesis is the design and implementation of a plugin for the Google Chrome web browser that assists users in verifying information on the internet. The application is implemented as a browser extension. It is written using the JavaScript programming language. The application operates independently, without need for server communication, with the exception of requests directed to the large language model provider. All evaluation logic is included directly in the client-side code. This implementation enables a high level of scalability. For full functionality of the plugin, users must have the appropriate access keys, specifically an API key for the large language model, as well as a Google Search API Key and a Programmable Search Engine ID. Following the implementation phase, user testing was conducted, which demonstrated the effectiveness of the proposed solution in practical scenarios and helped identify areas for future improvement. The partial goals of the thesis include the analysis of approaches suitable for adapting large language models to improve the reliability of information evaluation. This analysis was conducted through a review of academic literature. The thesis also includes an analysis of the functionality of existing extensions focused on information verification, carried out using a data collection method.