(Harnessing skills on giving prompts to AI to create images)
“Artificial intelligence is no longer just the future, it is the present, it is already here,” said Vujković.
The skills of posing queries to artificial intelligence are the skills of the future. Therefore, as part of the European Year of Skills and in collaboration with the Montenegrin Association for Artificial Intelligence MAIA, Europe House organised two workshops on how to use Chat GPT and other tools for prompt engineering, i.e., harnessing skills on how to pose queries to generative artificial intelligence.
The workshops were led by Miodrag Vujković on behalf of the MAIA Montenegrin Association for Artificial Intelligence.
During the first workshop, we gained skills on generating text using AI, while the second workshop was dedicated to learning how to generate images.
“Creating images using generative artificial intelligence is a very interesting thing, although image generating tools have not reached the technical advancement that text generation has reached. The reason is that image-generating tools are technically and infrastructurally more demanding,” emphasised Vujković.
It takes much more computing power to generate one image than to generate one article.
During the workshop, we had the opportunity to get acquainted with various AI tools that can be used for generating images.
There are several standard tools for generating images. The first tool that appeared with serious generation capabilities was Midjourney, followed later by Stable Diffusion and DALL-E.
“There are very small differences between these tools,” Vujković emphasises. “The only difference is perhaps in the aesthetics of the obtained results, but through prompt engineering we can adjust the aesthetics,” Vujković added.
Generating one image consumes as much power as charging a phone from 0% to 100%, and although generating images is demanding, this technology is improving from week to week.
During the workshop held at Europe House, we learned what are the elements of a good prompt for generating an image. We learned how to formulate a prompt using: Subject; Setting; Atmosphere; Mood, and more.
We learned that we will get the best results if we input the prompt in English, that a conversation with an artificial intelligence machine is no different from the conversation with a human, and that with a little imagination and skills it is easy to transform words into images.