Generative Artificial Intelligence (GenAI) in Higher Education

A set of ordered instructions that are used to solve a problem based on some understanding of available options. Algorithms are used heavily in computer science and mathematics but also have relevance to other fields like philosophy.

"The capability of a computer system to show human-like intelligent behavior characterized by certain core competencies, including perception, understanding, action, and learning...An AI application refers to the integration of AI technology into a computer application field with human–computer interaction and data interaction" (Wirtz, Weyerer, & Geyer, 2018, p. 599).

"A component of an information system that implements AI technology and uses computational, statistical, or machine-learning techniques to produce outputs from a given set of inputs."

AI that can only function within the parameters for which it has been programmed (Wirtz, Weyerer, & Geyer, 2018).

"A type of machine learning based on artificial neural networks in which multiple layers of processing are used to extract progressively higher level features from data" (Oxford Languages definition retrieved using Google).

Large language model that is trained on a broad set of data and then can be re-used in a range of tasks. There are both possibilities and risks inherent in this type of model (Bommasani et al., 2021; Murphy, 2022)

"A GPT is a language model relying on deep learning that can generate human-like texts based on a given text-based input. A user “feeds” the model with a sentence, and the transformer creates coherent paragraph-based information extracted from publicly available datasets. The technology can process any type of text, including guitar tabs or computer code" (Atelier Services, 2023)

A statistical model used in neural networks that assigns probabilities to word sequences and then is able to make predictions about which word might occur in a particular place in a sequence. Examples include GPT-3, GPT-4, and PaLM2. (Carlini et al., 2021; Petroni et al., 2019)

ML is a subset of AI that uses data to identify patterns, make predictions, and in other ways apply its “understanding” to situations beyond what it was originally programmed for.

The capacity of computer systems and the code employed to analyze, interpret, and generate language as it is naturally written or spoken (Aydin & Erdem, 2022)

A process of revealing flaws or potential vulnerabilities in AI models or systems. These efforts can be aimed at improving safety (reducing harm to users causes by system malfunctions) and/or security (defending systems from malicious actors).

AI that uses unsupervised ML and is able to transfer the “understanding” it has gained to new situations and toward the accomplishment of new goals which have not been pre-determined (Goertzl, 2014; Pham & Sampson, 2022).

Involves some human processing of data (labeling/tagging) before it is fed into an AI system (Celik et al., 2022)

ML that can process data without it being labeled by humans (Celik et al., 2022).