Usability of ChatGPT for searching for information sources

Sources of information on the internet are highly variable. The internet is actively used by a large percentage of the world population. Thanks to this fact there exists a vast amount of traceable information sources. The skill of searching for information correctly was never more relevant. In this aspect, the usage of AI, ChatGPT in particular, is becoming widespread. A large number of common users are using the outputs from this chatbot for satisfying their information needs. Outputs of this tool are often used directly for getting the needed information. Based on my own experiences with how time consuming the process of searching for sources can be, I wonder if ChatGPT can be used to speed up this process. The main objective of this work is specifically to evaluate the quality of sources provided by this chatbot when used for the search process. In addition to this main objective, I also evaluate three stated hypotheses in this thesis, which reflect my assumptions about the relationship between the currency of sources concerning their topics, their specialty, and their format. For needed context, I carried out an expert search of existing literature. In theoretical section I summarize the various views on defining, main areas of the field of artificial intelligence and the key stages of its evolution and furthermore in this section I focus on the area of natural language processing (NLP). Again, I provided an overview of the history of NLP, areas of application and I briefly characterize components and techniques that I consider essential. In the practical part of the thesis, I carry out the evaluation of sources provided by the ChatGPT tool. I obtained these through querying for sources for several token topics I selected. I performed this collection twice to assess possible effects of the time gap. I evaluated all sources according to CRAAP test criteria and finally assessed their overall quality.