Understanding some basic aspects of how artificial intelligence generates text helps us evaluate the tools and use them responsibly. Since artificial intelligence is a rather broad and vague term, in this context we will focus on discussing tools based on generative language models.

Simply put, these language models work similarly to as the autocomplete function in a word processing program.The task of the language models is to guess what the next word in a sentence will be, based on self-learning algorithms trained on large datasets.

In recent years, this type of artificial intelligence has undergone rapid development and has become so advanced that it can convincingly mimic natural language. We are also seeing similar progress in generative models for images and sound, as well as the ability to combine these different models so that, allowing us to generate video and audio through written instructions.

This development presents us with many new opportunities, but also introduces a number of new challenges—legal, ethical, and educational. After we’ve taken a closer look at the technology itself, we will return to these challenges.

Would you like to know more?

AI Art: How to see like a machine

The video below is from the Museum of Modern Art in New York. It shows how artists think about and use AI, and also provides insight into how one can train artificial intelligence using data.

Videoklipp: Museum of Modern Art (MoMA). (2023). AI Art: How artists are using and confronting machine learning | How to see like a machine. youtube.com

Sosiale medier Nyhetsmedier Faglitteratur Kunstig intelligens