What is GenAI and how does it work?
In recent years the development and availability of Generative Artificial Intelligence (Generative AI or GenAI) tools has been a major theme for discussions around education and society. GenAI has been viewed by some as a useful tool to unlock productivity and increase efficiency, while others see it as a threat to the environment and a “plagiarism machine”.
The concept of Artificial Intelligence is well known in a general sense, and in the metaphors in science fiction, but what do we mean by GenAI?
The following definitions of GenAI are useful to consider, and to examine the bias and different perspectives we can encounter:
- IBM: “Generative AI refers to deep-learning models that can generate high-quality text, images, and other content based on the data they were trained on.” (Martineau 2023)
- Jisc: “AI text generators such as ChatGPT are trained on a large amount of data scraped from the internet, and work by predicting the next word in a sequence.” (Webb 2024)
- AI generated definition: “Generative AI refers to a category of artificial intelligence systems designed to create new content - such as text, images, audio, video, or code - based on patterns learned from existing data. These models don’t just analyze [sic] or classify data; they generate new data that resembles the training material.” (Microsoft 365 Copilot, 2025)
In summary, the key points for defining GenAI are:
- A GenAI tool is able to create content in a variety of multimedia formats.
- Content created by GenAI tools is based on large amounts of data they have been trained on.
- The content created is informed by replicating patterns rather than a genuine understanding of the topic.
- The user will write prompts that inform how the GenAI tool will use any data it interacts with while making content.
How do GenAI tools work?
GenAI tools are incredibly complex tools and there are significant differences in how they all function. However, the following is a simplified description of the general process.
- Trained on vast amounts of text (or media)
- Large knowledge base that has been harvested. Sometimes informed by live internet access.
- Predicts the next word
- Text responses are based on what the tool believes the next word is likely to be.
- No actual understanding
- The tool does not understand the input from users or the content created.
- No quality control. No verification. Mistakes, hallucinations and misinformation are common.
- Responds based on input
- The context and content of the prompt will influence how the tool behaves, and the nature of what it creates.
A harsher description of GenAI tools has been provided by researchers from the University of Glasgow (Hicks, Humphries and Slater 2024).
References
- Hicks, M.T., Humphries, J. and Slater, J. (2024) ‘ChatGPT is bullshit’, Ethics and Information Technology, 26(38). Available at: https://doi.org/10.1007/s10676-024-09775-5
- Martineau, K. (2023) ‘What is generative AI?’, IBM, 20 April. Available at: https://research.ibm.com/blog/what-is-generative-AI (Accessed: 15 September 2026).
- Microsoft Copilot 365 (2025) Microsoft Copilot 365 response to John MacMillan, 28 August.
- Webb, M. (2024) Generative AI - a primer. Available at: https://www.jisc.ac.uk/reports/generative-ai-a-primer (Accessed: 15 September 2026).