If you pay attention to technology news you may be aware of advancements in Artificial Intelligence (A.I.) that are coming quicker and quicker each week. The current form of A.I. that is making news is called a chatbot and a user interacts with it by typing a prompt, similar to a text message, into a text box. The very complex and specifically designed software of the A.I. produces a response based on the prompt and sends it to the user. This process can be done repeatedly so using an A.I. chatbot can feel like exchanging text messages with a friend, colleague, or customer service representative. If the user of an A.I. chatbot installs resources to allow them to talk to the chatbot and for the chatbot to talk back, the experience is very similar to using the personal assistants Siri, Alexa, or Cortana to streamline a routine task like doing a Google search or putting a meeting into a calendar app.
There are several A.I. chatbots in development currently; some of the big names are Open AI’s ChatGPT, Google’s Bard AI, Microsoft’s Bing AI, Poe by Quora, ChatSonic, YouChat, among many others. Why has this form of A.I. become so popular? Previously, users were put on a waiting list to access these tools and only limited numbers could use the tools to prevent slowdowns of the complex software. However, the barrier to entry for using ChatGPT is relatively low—all that is required is an internet connection and a Google account, and people are seeing the benefits.
Farhad Manjoo in an article reprinted in the New York Times notes that Nicholas Carlson, the editor in chief of an international publishing and media company, has come to think of ChatGPT as “a two-player word processor,” helping them overcome writing obstacles and enhancing productivity. These can be simple tasks like asking ChatGPT to replace a word in a sentence to help you recall the right one, or analyze a writing sample for tone, style, reading level, and length and produce content on requested topics in the same style. It is an expert in aiding individuals in language-based professions by reducing time spent on certain tasks. Journalists, marketers, writers, and editors are finding ChatGPT valuable.
So how does ChatGPT work? ChatGPT functions as a language predictor, generating responses based on patterns and examples in its vast repository of written text. ChatGPT’s software understands the patterns in the huge volume of text samples it has processed and it recreates those patterns based on what the user asks it to do. When accessing ChatGPT, users are greeted with notices highlighting its limitations: occasional generation of incorrect information, potential for biased content, and limited knowledge of events beyond 2021. A chatbot that relies on predicting the next word and does not have a mechanism to verify the information it is giving can easily lie to the user. Users who have honed their information literacy skills before they chat are able to more easily sort out what is helpful and truthful from what is false. Instead of an encyclopedia or a librarian, ChatGPT functions more for sparking creativity.
ChatGPT is a powerful tool for lifelong learning. It can be utilized for brainstorming, summarizing, prose generation and improvement, generating writing or craft project prompts, and language learning, to name just a few. It works well as a personal sounding-board for exploring topics and ideas. It can have positive impacts on personal growth including casual research and career explorations, and a way to get advice (as long as it is verified by another source) without judgement (or with a judgmental tone if you prefer to “prime” it that way).
Super users of ChatGPT will be aware of the main concepts of priming, personas, and prompt shaping. To prime ChatGPT you identify which persona or tone it should use to respond to your prompts. ChatGPT can be primed with specific personas, such as a career counselor, travel agent, or biologist, influencing its responses accordingly. While priming does not improve accuracy, persistent users can leverage this technique to their advantage, producing better results. Prompt shaping is the concept that the user can change and update their prompts to get a better response from ChatGPT about any particular topic.
So why do librarians care about ChatGPT? The limitations of this A.I. chatbot align with core principles of information literacy, the accuracy, bias, and currency of information, that are cornerstones of a librarian’s professional activities. Additionally, ChatGPT’s inability to cite sources raises questions about the authority and reliability of the information it generates. It is clear that ChatGPT is not a substitute for other information seeking methods with transparent citations and information about the authority and accuracy of a source. ChatGPT has made a substantial impact in the world of academic librarians who have observed an influx of research source lists that are generated by ChatGPT, some of which contain completely fictional citations. These instances highlight ChatGPT’s limitations in generating factual information and reliable sources for research purposes. Libraries can serve as hubs for individuals to explore the potential benefits of ChatGPT in their work and offer guidance on its usage, including outlining the responsibilities users have when using ChatGPT for scholarship.
Moreover, ChatGPT can serve as a gateway to the numerous AI tools already available for generating text, images, videos, research presentations, spreadsheets, and more. The barrier to entry for using ChatGPT is relatively low, as previously described. Patrons who choose to explore ChatGPT are likely to appreciate an information-savvy guide who can illuminate the possibilities and drawbacks of a tool that is already shaping the world. This is exactly where librarians can make an impact.
