Artificial Intelligence (AI)
There is no single definition of artificial intelligence or AI. AI is “a catch-all term for a set of technologies that make computers do things that are thought to require intelligence when done by people,” such as “recognizing faces, understanding speech, driving cars, writing sentences” (Heaven, 2024).
Generative Artificial Intelligence (GenAI)
Generative artificial intelligence (GenAI) is a specific type of artificial intelligence capable of generating new content, such as text or images, based on patterns learned from existing data.
People often rely on comparisons, metaphor, and figurative language to describe new technologies. Authors have compared generative AI to tools like calculators and have described them using terms associated with human traits and behavior, such as "learning," "teaching," "understanding," and even "intelligence." These descriptions may capture some aspects of the technology, but they are imperfect and misleading.
Experts in the field of computer science have critiqued popular media for perpetuating misconceptions about generative AI technologies:
In this blog post, computer scientists Sayash Kapoor and Arvind Narayanan (2022) discuss common mistakes journalists make when covering AI, including anthropomorphizing AI tools and implying they, like humans, are independent actors.
Due to their ability to produce human-like text, using GenAI tools such as ChatGPT or Google Gemini can feel like conversing with a subject matter expert or even consulting an oracle. In reality, these tools are capable of generating human-like text because they have analyzed vast amounts of human-written text long before you prompted them.
The tools are “trained” to recognize patterns in human language by analyzing immense amounts of human-written texts, often sourced from the internet. In response to a user’s inquiry, the tool generates a combination of words that sounds plausible based on the texts it has been trained on (Nield, 2023). In “What is AI?”, Will Douglas Heaven (2024), senior editor for AI at MIT Technology Review, describes the training process:
During training, which can last months and cost tens of millions of dollars, [GenAI tools] are given the task of filling in blanks in sentences taken from millions of books and a significant fraction of the internet. They do this task over and over again. In a sense, they are trained to be supercharged autocomplete machines. The result is a model that has turned much of the world’s written information into a statistical representation of which words are most likely to follow other words, captured across billions and billions of numerical values.
Human beings, including hired workers and end users, play a crucial role in training GenAI tools. They provide examples for the tool to emulate, identify errors in responses, and rate the quality of responses (Nield, 2023).
In this video, Jane Stimpson, a former community college librarian, provides an overview of how text-generating AI tools like ChatGPT work.
Artificial Intelligence (AI): "The capacity of computers or other machines to exhibit or simulate intelligent behavior." (Oxford English Dictionary, n.d.)
Generative AI (Gen-AI): A type of AI technology that generates content such as text, images, audio, and video. Also sometimes referred to as a generator.
Machine learning: A technique by which a computer can learn without being directly programmed with rules.
Large Language Model (LLM): A complex model trained on vast amounts of data that generates language that resembles human-generated language. Open AI's GPT series (powering ZotGPT, Microsoft CoPilot, and PapyrusAI), Google's Gemini, and Meta's LLaMA are examples of LLMs.
Chatbot: A computer program that uses an LLM to simulate a conversation with human users, typically through typed text in a software application.
Algorithm: A set of instructions or rules for performing a computation. Developers typically design algorithms used in AI to progressively iterate themselves, which we can consider a form of machine learning.
Training: The process of supplying algorithms with large data sets, and then "teaching" them to develop progressively sophisticated outputs for their intended purpose.
Heaven, W. D. (2024, December 6). What is AI? MIT Technology Review. https://www.technologyreview.com/2024/07/10/1094475/what-is-artificial-intelligence-ai-definitive-guide/
Content adapted from the University of Wisconsin-Madison Libraries’ Generative AI Research Guide, authored by Dave Bloom and Heather Shimon.
Bender, E. M. (2022, May 2). On NYT Magazine on AI: Resist the Urge to be Impressed. Medium. https://medium.com/@emilymenonbender/on-nyt-magazine-on-ai-resist-the-urge-to-be-impressed-3d92fd9a0edd
Bhaggart. (2023, January 31). Why it’s a mistake to compare calculators to ChatGPT. Blayne Haggart’s Orangespace. https://blaynehaggart.com/2023/01/31/why-its-a-mistake-to-compare-calculators-to-chatgpt/
Kapoor, S., & Narayanan, A. (2022, September 30). Eighteen pitfalls to beware of in AI journalism. AI Snake Oil. https://www.aisnakeoil.com/p/eighteen-pitfalls-to-beware-of-in
Heaven, W. D. (2024, December 6). What is AI? MIT Technology Review. https://www.technologyreview.com/2024/07/10/1094475/what-is-artificial-intelligence-ai-definitive-guide/
Nield, D. (2023, April 30). How ChatGPT and Other LLMs Work—and Where They Could Go Next. WIRED. https://www.wired.com/story/how-chatgpt-works-large-language-model/
Definitions adapted from University of California Irvine Libraries’ Generative AI and Information Literacy LibGuide written by Stacy Brinkman and April Urban.
Defining AI and chatbots. (n.d.). Teaching Commons. https://teachingcommons.stanford.edu/teaching-guides/artificial-intelligence-teaching-guide/defining-ai-and-chatbots