Generative AI refers to various machine-learned models that generate content such as text, images, code, audio, and more. This type of artificial intelligence, such as Large Language Models (e.g., ChatGPT, Gemini) or Diffusion models (e.g., Dall-E, Stable Diffusion), is trained by feeding it large amounts of data. Through statistical associations, it can produce new content based on its training data.

However, Generative AI often makes assumptions when missing information in prompts. For example, if you ask for an image of Enschede, the model might associate Enschede with typical elements of the Netherlands, like cyclists, water, mills, and canal houses—sometimes leading to even unrealistic images (e.g., cyclists on water, multiple fingers, etc).

This hub focuses on Generative AI because of its accessibility and broad applicability for regular users, as well as its distinct impact compared to other AI types. Unlike traditional AI with hard-coded algorithms that are predictable and explainable, Generative AI involves a 'black box' of associations that can be unpredictable.

Watch the video lecture from UT's prof. dr. Giancarlo Guizzardi (Week of Education 2024) about what artificial intelligence is and how it connects to the humanities: