Závěrečná práce: Martina Stankovičová: Analysis of Bias and Reliability in LLM-Powered Search Engines
Bakalářská práce
Analysis of Bias and Reliability in LLM-Powered Search Engines
Anotace
Vzostup veľkých jazykových modelov (LLM) nemožno podceniť. Postupne sa stávajú súčasťou takmer všetkého, najmä na internete. Prirodzene, to zahŕňa aj internetové vyhľadávače. Mnohé štúdie sa zaoberali politickým skreslením LLM a ich vplyvom na správanie používateľov pri vyhľadávaní informácií. Táto práca sa zameriava na podrobné preskúmanie troch populárnych vyhľadávačov založených na LLM: Perplexity …více
Abstract
The rise of Large Language Models cannot be underestimated. They are slowly becoming a part of almost everything, especially on the internet. Naturally, this includes search engines. Many studies have been conducted about the bias of LLMs and their effects on searching behaviors. This thesis aims to closely examine three of the most popular LLM-based search engines: Perplexity, You.com, and Brave Search …více
Zadání práce
This thesis aims to evaluate the performance of LLM-powered search engines (such as Perplexity or You.com) in providing links to primary news sources. The study will focus on the bias, partisanship, and reliability of sources attributed by these engines while considering user prompts, ethical implications, and the broader societal impact.
Specific tasks:
-
Provide an overview of LLM-powered search engines and the available databases on bias, partisanship, and the reliability of news sources.
-
Develop a web application to automate information retrieval. Based on user input, the application should query multiple LLM-powered search engines via API, extract the recommended sources, and retrieve information regarding their bias and reliability.
-
Select a representative sample of events and analyze the responses of search engines, focusing on the bias and reliability of the sources they provide.
-
Discuss the ethical implications of bias in LLM-based search engines.
26. 5. 2025 14:04, Mgr. Tomáš Foltýnek, Ph.D., učo 4374
Přílohy
Práce na příbuzné téma
Seznam prací, které mají shodná klíčová slova.
-
Exploring The Potential of Generative AI as a Topic in The EFL Classroom
Mgr. Viktor Šťastný -
Adoption of LLMs in the PV260 Software Quality Course: Exploration and Evaluation
Ing. Markéta Opichalová -
A Quest for Information: Enhancing Game-Based Learning with LLM-Driven NPCs
Mgr. Tereza Tódová -
Automating Event Stream Definition in Kafka Ecosystem Using Large Language Models
Ing. Patrik Szabó -
Use of Large Language Models in Programming Courses
Bc. Jakub Čížek -
Virtual AI partner for training situations in teaching
Ing. Mikuláš Heinz -
Synthetic Dataset Generation for Customer Inquiry Classification
Mgr. Andrej Srnka -
Evaluating the Ability of Large Language Models to Provide Code Quality Feedback
Ing. Michal Adler




