Latent image generation on paper. Photo: H.E. Max Haarich.
Generative Artificial Intelligence is the absolute megatrend in the world of work. Creative and design professions, in particular, are suddenly faced with unprecedented possibilities: Generative AI can now create texts, images, and sounds in seconds that would have taken humans days, weeks, or months to produce. At the same time, Generative AI can also produce nonsense that humans wouldn’t have thought of in a hundred years.
The goal of the workshop is to develop a deep understanding of the theoretical and technological foundations as well as practical knowledge for the successful use of Generative AI, with a focus on image generation (Text-to-Image):
- What is artificial intelligence, and how does it differ from human intelligence?
- What is Generative AI, and how does a text prompt turn into an image?
- What are common tools and workflows for image generation?
First, we will discuss the basic requirements for an AI that is meant to understand human texts and images—and then we will become the AI ourselves: In playful group exercises, we will explore individual AI workflows such as data preparation, model training, image recognition, image generation, or model fine-tuning, and discuss how to handle distortions and bias. Based on this understanding, we will explore the open-source image generation software Stable Diffusion to bring our own ideas to life.
No prior knowledge is required, but a laptop is needed to use free online tools (many require registration via a Google account). Those who have a high-performance laptop with a graphics card are welcome to bring it on the second and third days.