Generative AI – what is it and what is it capable to do?

In the media, there have been more and more special coverages on the topic of artificial intelligence in recent times, for example in the Tagesanzeiger or the NZZ magazine. There was talk of generative AI, which independently generates content, such as texts or artistic works like images. What possibilities already exist and how will this affect working life and society? We discussed this and more with expert Dr. Souhir Ben Souissi from BFH School of Engineering & Computer Science in the following interview.

What exactly is meant by generative AI?

Dr Souhir Ben Souissi researches and teaches at the Department of Technology & Computer Science at BFH.

Souhir Ben Souissi: Generative AI is a type of Artificial Intelligence that focuses on generating new content or information autonomously. It uses techniques such as deep learning to generate and manipulate data. Its purpose is to create new, unseen and unique content based on the data it is given and trained on. This type of AI understands patterns in existing data and creates original content such as text, images, audio, or video … Generative AI can be used to assist creators with generating new ideas, products, and services.

Can you give examples of applications using generative AI?

Several applications are using generative AI, such as:

  1. Text Generation: Generative AI  is used to generate text, helping users in the writing and editing processes of their work. One of the more recent architectures was presented by OpenAI, called ChatGPT3[1]. This API can provide users with context-specific, human-like conversations that can be used for various applications, such as customer service chatbots, and automated dialogue systems.
  2. Virtual Personal Assistants:  Generative AI generates conversations and natural-language responses to help virtual personal assistants better understand and respond to user requests.
  3. Autonomous Cars: Generative AI is used to generate driving scenarios, such as road conditions, obstacles, and other vehicles, to help train autonomous cars.
  4. Healthcare: Generative AI can generate possible chemical compositions to help the research and development process for drug discovery.

Can it be used with different technologies?

There are virtually no limitations on how these systems can be composed with other technologies, especially with other more classical forms of software. The output they produce, either in the form of text, code, or images, can be readily understood by other sub-systems to form a more general solution.

Is it possible for generative AI to generate artwork? How does this work?

Yes, it is possible for generative AI to generate artwork, for example, in the form of images. In the domain of image generation, e.g., the neural architecture works by learning how to de-noise images from its dataset. Then it can start from an input of pure noise and, with the guidance of a textual prompt, iteratively de-noise its input until it converges into an output that we can visibly recognize as a genuine new image. This is achieved by training the AI with data that is either provided by domain experts or collected from the internet. Once the AI is trained, it is able to generate unique pieces of art.

  1. For Music: Generative AI can be employed to generate original music from audio samples or from scratch, as well as to compose and remix existing music. As an example, Holly Herndon is a singer who has employed cutting-edge Artificial Intelligence to create a generative clone of her own voice, enabling her to sing in any language and in any tone, even those she does not speak. Through her music, Holly demonstrates how AI can expand the capacity and artistry of the human voice[2].
  2. For Visual Art: Generative AI can be used to generate images, such as photographs, paintings, and other forms of art. DALL·E [3] is an example of such an AI system, providing generated images based on user input (usually referred to as “prompt”). Here, there are some examples:

Prompt 1: An impressionist oil painting of dance

Prompt 2: A Matisse oil painting for dance

What are the consequences for society of such technology, especially for artists? 

Generative AI has the potential to revolutionize how humans interact with machines, generate new insights, and create new opportunities for economic growth and job creation. Artistically, it can provide a valuable tool for both professional and amateur creators to better express themselves creatively through lyrics, music, live visuals, and augmented performance. Moreover, it has the potential to democratize both the artistic creation process and the use of advanced AI technologies for content generation. However, it is essential to consider the potential risks and ethical implications of Generative AI before implementing it on a large scale, such as data privacy and security, job displacement, and algorithmic bias.

Are there also new chances that arise from such technologies for artists?

Generative AI technologies offer a range of opportunities for artists to explore new creative possibilities. By providing the ability to experiment with different combinations of colors, shapes, and textures, generative AI can open up new pathways for creating and refining artwork. Moreover, it can help artists become more efficient in their creative process, enabling them to dedicate more time to the creative elements of their artwork.

About the person

Dr. Souhir Ben Souissi is a tenure-track professor for Data Engineering at the Institute for Data Applications and Security (IDAS) at BFH School for Engineering & Computer Science. Her research interests include medical decision systems, semantic web technologies and multi-criteria decision systems.

Save the Date: TRANSFORM Conference on AI

What can AI do in public service? This is the topic of this year’s TRANSFORM conference at BFH Wirtschaft. It will take place on May 3rd in the Bern City Hall. Speakers from science, administration and other organisations will discuss these questions: Where are we today with the application of AI? What experiences does the administration have with AI? Where is there potential for the use of AI in administrative work processes? What are the possible risks and opportunities?

Paulina Grnarova from DeepJudge and Bertrand Loison from the Federal Statistical Office have already confirmed their keynotes. More information on the programme and the registration form will follow soon on this page.





Creative Commons Licence

AUTHOR: Mascha Kurpicz-Briki

Dr Mascha Kurpicz-Briki is Professor of Data Engineering at the Institute for Data Applications and Security IDAS at Bern University of Applied Sciences, and Deputy Head of the Applied Machine Intelligence research group. Her research focuses, among other things, on the topic of fairness and the digitalisation of social and community challenges.

Create PDF

Related Posts

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *