Copy, Paste, Create! Unearthing the Magic in Magic Media

Generated by Canva's Magic Media, Prompt: "Moon rising over the mountains"

The issue of generative AI has led artists all over social media on a rampage. Google searches are flooded with AI photos, social platforms with AI-generated media, and even spaces once thought to be exclusively for human creativity—like Pinterest and DeviantArt—have been infiltrated. To add to this mix, Canva, a graphic design tool celebrated for its creativity and user-friendly features, launched Magic Media on October 4, 2023.

The Wonderful World of Magic Media

Magic Media is Canva’s AI-powered tool designed to generate images in sets of four, based on prompts of at least five words. Users can choose from various artistic styles, with different capabilities depending on their Canva subscription:

• Free users have up to 50 image generations.
• Pro users enjoy up to 500 image generations monthly, can create AI-generated vectors or modify existing Canva library assets, as well as have up to 25 image generations on other platforms like DALL-E by OpenAI as part of their subscription.

At first glance, this AI appears to be a harmless creative assistant. However, as with any generative model, achieving high accuracy requires massive amounts of training data. Just as human creativity is shaped by exposure to various stimuli—learning from experiences, adapting techniques, and refining skills over time—AI functions similarly. It learns by processing vast datasets of images and their labels, identifying patterns, and piecing together elements to match a given prompt. The crucial ethical question, however, lies in where this training data comes from.

A key concern is whether these AI models are trained on copyrighted images. Humans can draw inspiration from copyrighted work, but there are ethical and legal boundaries—such as giving proper credit and ensuring that original work is not outright copied. AI, however, lacks this understanding. It doesn’t decide what is ethical; it follows its programming (similar to the Robot in Robot & Frank). This raises alarms when AI-generated images display distorted watermarks, suggesting they may have been trained on copyrighted works without consent.

Canva states in their FAQ section that its machine learning model “scans millions of images across the internet along with the text associated with them” to create unique images. They even add how

“You may use images you create with our AI Text-to-Image generators for any legal purpose, including personal or commercial projects.”

At first, this may sound like there are no issues in terms of usage and infringing copyright, however, the same answer section includes a conflicting statement:

“Please note, you may not have exclusive rights to your images.”

Especially since other people may generate an identical image (or features of an image), for example the moons shown here (both mountainous images with moons in the background were generated using the same prompt):

But, most crucially:

“Canva doesn’t provide any guarantee that the images you create are cleared for use (particularly if the image you create looks like someone else’s work). If you are going to use images for any commercial purpose, it may be a good idea to seek local legal advice.”

This contradiction is concerning. Canva encourages users to create AI-generated images for any purpose, yet distances itself from legal accountability—raising questions about how ethical and safe AI-generated art truly is. Additionally, if AI is pulling images from across the internet, this likely includes copyrighted artworks and social media posts, making it inevitable that AI outputs could resemble existing works.

My Experience and Thoughts on Using AI for Art

Despite these concerns, I’ve personally experimented with AI-generated art. Sometimes, I’d use it to generate backgrounds, which I later blurred and painted over, or to create color palettes as a starting point. The goal was never to let AI dominate the artistic process, but rather to use it as a tool to enhance creativity. One of the most interesting aspects of using AI was its unpredictability. While some images turned out surprisingly coherent, others were complete messes—featuring double moons, floating clouds on grasslands, or distorted human features (which you may have already seen floating around when images generated by DALL-E were first introduced).

This highlights a major limitation: AI struggles with fine details and artistic meaning. It doesn’t “understand” composition or storytelling in the way a human artist does. Instead, it stitches together pieces from its dataset, sometimes resulting in odd imagery. While this can be humorous and even inspiring, it also reinforces why AI cannot replace human artistry—it lacks the depth, intentionality, emotion, and other nuances of human art.

AI is here to stay, but it should never replace real artists. Every AI-generated image is, in some way, a reflection of human-created work—because AI cannot create from nothing; it only remixes existing content. That being said, AI can still be an invaluable tool when used ethically. It can help conceptualize ideas, generate references, or provide inspiration, but it should not be treated as a replacement for artistic expression. We must ensure that AI remains a creative assistant rather than a creative replacement.


Source:

https://www.canva.com/ai-image-generator/