Skip to main content

What is Generative AI?

Artificial Intelligence (AI) is: “The capacity of computers or other machines to exhibit or simulate intelligent behaviour; the field of study concerned with this.” (OED, 3rd ed., Dec 2021). This includes a range of techniques used in research to find patterns in complex data, including machine learning and generative AI.

Generative AI (Gen AI) is a type of artificial intelligence system that automatically generates content in response to written prompts. The generated content can include text, software code, images, videos and music. Generative AI models learn the patterns and structure of the data they are trained with. They then generate new data that has similar characteristics.

Generative AI has the potential to transform education and learning and to develop new forms of creativity. At the same time, it is full of biases, inaccuracies, and other issues. We need to hedge our moments to explore its full potential while recognising its flaws and issues.

Prof. Jeff Grabill, Deputy Vice-Chancellor: Student Education, University of Leeds

Gen AI is usually trained using data from webpages, social media conversations, survey data, private datasets and other online content. It generates its outputs by identifying and repeating common patterns, for example, which words typically follow other words, or which pixels of images should come after other pixels. This is usually done by ingesting large datasets and statistically analysing the distribution of words or pixels or other elements in the data.

Remember:

  • While Gen AI is often accurate, even in complex conversations, it is not reliable and can make simple mistakes.
  • Gen AI can simulate intelligent conversation, but it doesn't possess true consciousness or human-like thinking.
  • Though Gen AI responses might seem insightful, it doesn't genuinely understand or comprehend topics as humans do.

While many of us will use Gen AI in creative ways in our work, we always need to verify that its outputs are accurate and appropriate for our needs.

Text AI

In response to a human-written prompt, text Gen AI generates text that usually appears as if a human has written it.

Yet, just like human-written texts, text Gen AI outputs can be superficial, inaccurate, untrustworthy and full of errors.

Despite appearances, text Gen AI does not understand either the prompt written by the human or the text that it generates. You may have encountered this if you have ever tried to have a conversation with a text Gen AI tool and seen it contradicting itself. You may also have seen a text Gen AI tool choosing words that are not quite right for the context or intended meaning of the output. It is just providing the words that would statistically be most likely to appear in a response to the prompt you provided.

Every time that we use a text Gen AI tool, we need to consider its output from a critical perspective.

See a list of common text Gen AI tools

Multimedia AI

Multimedia Gen AI can generate image, video or music outputs based on human-written prompts. Some can also respond to visual or audio prompts.

The appearance of image, video and music Gen AI outputs might appear novel. However, usually they are only complex combinations of the millions of images, videos or music that they have ingested during their training.

Gen AI uses earlier works in its outputs and this creates substantial challenges in copyright, intellectual property and the ownership of content. Two examples of this are the 2023 Writers Guild of America (WGA) strike in Hollywood on this topic, or the issues with a song created by AI that mimics the work of Drake and The Weeknd. These highlight deep issues with ethics and ownership when it comes to AI generated works.

See a list of common multimedia Gen AI tools

Learn more

The National Centre for AI have produced a short introduction to Generative AI technology and its implications for education, with more detail about the different tools and their capabilities and limitations: A Generative AI Primer