Abstract：Through the looking glassenvisioning new library technologiesAI-text generators as explained by ChatGPT This is the first in a series of columns focused on the intersection of text-based generative artificial intelligenceAIand librariesFor yearsexperts have been predicting that AI was poised to change multiple industriesyet much of that potential was largely inaccessible to most consumersNow companies are starting to release products that allow users to experience and interact with AI that seems remarkably responsiveThe purpose of this column is to provide a foundational understanding of the technology behind ChatGPTthe chatbot released by OpenAI that has suddenly generated headlines on the topicText is foundational to modern communicationThe ability to rapidly generate text is poised to transform everything from educationto programming and not least of all librariesChatGPT is just the most prominent current example of a set of technologies that are likely to continue to develop rapidlyChatGPT has been made public at this time as part of its beta to help improve itwhich will in turn foster future developmentBy the time you read thisit may not even be the most well-known example of textbased generative AIThis column will focus on understanding the building blocks of the technology will better enable information professionals to understand the possibilities for its futureIn reading itit is important to understand the context that this technology is moving quicklyand that it has applications beyond this particular toolThese foundational concepts also provide a foundation for future columns that will delve further into the areas most applicable to information professionalsThis was written almost entirely through asking ChatGPT questions and verifying the answers along with some light editingWriting in this manner raises the specter of many of the quandaries facing its futurein terms of attributionbut also when broaching topics beyond its area of expertiseits accuracyChatGPT grants the ability to use the text it generatesbut a host of other legal issuesincluding copyrightremain much murkierKafka2023The column summarizes the deep learning model that makes ChatGPT functionas well as some of its key features such as its use of attention mechanisms and transformer architecture that allow its responses to focus on the right conceptsIt also explores some of the potential controversies around this technologywith an emphasis on bias and false informationFor information professionals having a foundation for understanding these concepts will continue to be relevant into the future Consulting ServiceThe article mentioned that ChatGPT can be used as a tool to answer questions and provide informationwhich is similar to the librarys consulting service functionThis article is concentrated in the AItechnology generated by textespecially the ChatGPT published by OpenaiThis technology is based on deep learning modelsespecially using attention mechanisms and converter architectureswhich enables it to generate corresponding textsThe article also discussed the potential controversy of this technologyfocusing on prejudice and error information