Since the public release of ChatGPT in November of 2022, teachers and students have been grappling with how generative artificial intelligence tools can be used productively in the context of education.
When considering the use of tools like ChatGPT, it helps to start with an understanding of what they are and how they work. The term “generative artificial intelligence” refers to computational tools that generate content–such as text and images–in response to user prompts based on predictions drawn from very large datasets. GPT-3, for example, was trained on roughly 500 billion words scraped from text on the Internet and digitized books.
If something is being called generative AI, it probably involves a large language model. Large language models (LLMs) take user input and attempt to predict the words most likely to follow from that input. Some LLMs have been designed to draw from a specific body of texts, and some have been designed to incorporate web or database searches into their responses, but as a default an LLM does not search any source external to itself. It is guessing what words are likely to appear together based on the context that user input provides.
Generative AI tools are being integrated into many research databases. These tools may appear in the form of automatically generated summaries, suggested resources, research topic suggestions, and chat bots. It is important for researchers to be aware when the text they are seeing is likely to have been generated by a large language model, because we cannot assume the accuracy or intention of words generated by machines. As well, Grinnell College’s academic honesty policy requires citation of generative AI material if used in academic work.