Artificial Intelligence (A.I.) - Introduction - Use of A.I. tools in teaching and research - LibGuides at University of Victoria Libraries

Upcoming Workshops

Questions? AskUs!

Questions?
Contact Your Subject Librarian

Defining Artificial Intelligence now and then

What is Artificial Intelligence?

Initial appearance of the term & the concept it describes

The term "Artificial Intelligence" was coined by computer scientist John McCarthy and his team at the University of Stanford in 1955. Not only is the term mentioned for the first time in A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence, but an initial outline of the concept it is intended to characterize is also provided:

...the artificial intelligence problem is taken to be that of making a machine behave in ways that would be called intelligent if a human were so behaving.

McCarthy et al, 1955

Recent definitions - surprisingly similar

This early concept by McCarthy et al has not lost its relevance. More recent definitions have continued to appear very similar to this original notion:

Al is best understood as a set of techniques aimed at approximating some aspect of human or animal cognition using machines.

Calo, 2017

Defining A.I. - Concepts and technologies

Artificial Intelligence – An umbrella term

Artificial Intelligence is often used as an umbrella term under which various types, concepts or specific technologies may be categorized. A few examples of those types and technologies, that are relevant to understanding the current generation of A.I. applications are:

Different A.I. types, for example
- Weak/Narrow vs. Strong/General A.I.
- Multitasking/Multimodal/Generalist A.I.

Various technological concepts, for example
- Machine learning
- Neural networks
- Deep learning

Please note that the terminology around A.I. is not being used consistently across resources, especially when it comes to A.I. typology. While consistency is relatively high and definitions are mostly clear in regards to A.I.-related technologies, the interchangeable use of some terms around certain types of A.I. can lead to confusion. UVic Libraries will vet and include resources here, that might help clarify A.I.-related terminology and that make it more comprehensive.

Types of Artificial Intelligence

Different types of Artificial Intelligence that can be found in the literature include...

Weak/Narrow vs. Strong/General A.I.
- Weak/Narrow A.I. = A.I. with limited capabilities, even if highly specialized in what it is supposed to do. ChatGPT is still considered a Weak A.I., for example, despite its impressive and diverse capabilities.
- Strong/General A.I. (AGI) = An Artificial Intelligence, that meets or exceeds human-like cognitive capabilities and is not limited to certain tasks. The development of an AGI is the declared long-term goal of A.I. research and development, but at this point in time, this goal has not yet been achieved.
Multitasking/Multimodal/Generalist A.I. = A.I. that is designed and trained to (or in some cases: taught itself to) do more than one task. The boundaries between uni-tasking A.I. tools and generalist ones can be fluid, making precise definitions and clear language around this topic challenging.

A.I. Technologies

Technological concepts that are summarized under the A.I. umbrella term include

Machine learning = Computer systems performing tasks they are not programmed to do, instead they are programmed to learn how to do them
Neural networks = Computer systems resembling the human brain
Deep learning = Machine learning by example, where neural networks are using large sets of training data to train themselves, for example how to recognize patterns
Large language models (LLMs) = Deep learning applications, utilizing massive text corpora to train themselves how to generate, recognize, summarize, translate, predict almost any kind of text