From the course: Introduction to Prompt Engineering for Generative AI

The AI-generated image landscape

From the course: Introduction to Prompt Engineering for Generative AI

The AI-generated image landscape

“

- [Instructor] Large language models have brought about a pretty interesting phenomenon: text to image. This means you can turn words into images that look like drawings and photographs. Now, this brought about some interesting controversies, some of which are, are artists getting enough credit for the works that were used in order to train these models? How do we define art? And this is a similar debate to the debate that was around when photographs came about. And finally, what is the potential for misuse, in particular with fake news? Now we're going to look at two pretty popular language to image models. One is DALL-E by OpenAI. We know that it's based on the Transformer model architecture, and it's trained on millions of images that have descriptions. The other is Midjourney, which is also extremely powerful. It uses Discord as its interface, which may take a bit to get used to, and it does have a little bit less free capabilities, so if you're into generating AI art with Midjourney I recommend getting a monthly usage plan.

Contents