Skip to Main Content Saint Mary's Libraries

What is ChatGPT?

ChatGPT (Chat Generative Pre-trained Transformer) is a LLM (large language model) chatbot developed by OpenAI and launched in 2022. It enables users to refine and steer a conversation with the chatbot toward a desired length, format, style, and level of detail. 

Large language models like ChatGPT are designed to mimic human language and use mathematical models to predict the next word based on what you are asking for. They don't tell you where the got the information they're pulling from or who is working on them behind the scenes. Most are unregulated and influenced by how we interact with it.

There is a free version of ChatGPT, but a more advanced version is available for paid subscribers. According to their Privacy Policy, they "may share your Personal Information with third parties without further notice to you."

Some competing products include Bing Chat (Microsoft), Bard (Google), and LLaMA (Meta).

How does ChatGPT work?

Large language models like ChatGPT work by taking an input text and repeatedly predicting the next word. It was fine-tuned to target conversational usage. It was trained on large collections of text, such as books, articles, and public web pages. Time magazine reported that OpenAI used Kenyan workers to flag hateful content; these workers were paid less than $2 per hour.

Tips & Tricks

Benefits of using ChatGPT or similar tools:

  • Can provide simple explanations to well known topics, similar to a tool like Wikipedia
  • Can provide sample text for you to edit
  • Can create a list of keywords to help you in research

Drawbacks of using ChatGPT or similar tools:

  • There is a mix of correct and incorrect information
  • Has limited knowledge of the world after 2020
  • Biased based on bias in its training material
  • Known to produce citations of articles that do not exist
  • Only trained on freely available information; anything behind a paywall (for example, academic research like what you find in library databases) wasn't included in the training
  • Often down or slow to generate responses during peak usage windows