Canva Launches Craft: What Does It Mean for Creativity?

Any graphic designers in this group? Have you heard about Canva?
I use Canva almost every day. It is a very handy tool for creating graphics for my articles, social media posts and business.
Now, Canva has launched Craft, its own image-generation model. There are already several AI models that can generate beautiful images from text prompts. What makes this announcement interesting is that Canva, a design platform that helps people create and edit visual content, is now developing its own image-generation technology.
Why does this matter? And what does it tell us about creativity in the age of AI?
How AI is changing creative work
Creating a personal brand on social media is the thing these days. Whether it is Instagram, Pinterest, Snapchat or Facebook, creating images and videos has become almost a necessity. The same is true for businesses. Marketing teams have to keep generating fresh content for campaigns, promotions and customer engagement.
With the advent of generative AI, creating visuals has become increasingly accessible, especially for people who previously found it difficult to create them. You no longer need to be a skilled graphic designer to produce an attractive image. Describe what you want, specify the style and colours, and AI can generate something for you.
AI can also generate multiple options, helping us explore ideas that we might not have thought of ourselves. This is a big advantage, particularly when time is limited.
But does the ability to generate images mean that creativity is no longer needed? I don't think so.
Creativity was traditionally considered a uniquely human ability. With the advancements in generative AI, machines can now produce images, illustrations and designs that would once have required considerable artistic skill. This makes me wonder what exactly happens when AI creates something, and how that compares with the way humans create.
My experience with AI image generation
I enjoy painting and creating digital art. I use tools such as Linearity Curve, Photoshop and Canva to create visuals for my articles and business.
Sometimes, when I don't have much time and need to publish an article, I turn to generative AI to create an image. The results can be beautiful and often come quite close to what I have in mind. But I have also faced several challenges.
For one, once I have generated an image, making subsequent changes is not always easy. I might want to change one small detail, adjust the placement of an object or modify a colour. Instead, I often end up generating different versions and explaining what I need all over again. After many attempts, I sometimes get what I want. At other times, I have to settle for something that is close enough or start a fresh conversation to explore a different direction.
Interestingly, I have often found that making a rough sketch myself and giving it to the AI as a reference helps. With my sketch as input, the AI is able to do a much better job of producing what I have in mind.
Another challenge is that AI often gives me a finished image when what I really need is an editable design. For marketing, you may need vector graphics or SVG files that can be modified to fit your brand requirements. You may also want to maintain consistency across a series of creatives. Getting that level of control from a generated image is not always straightforward.
And then there is text. Getting AI to generate an image with the exact text, spelling and layout you want can be a challenge. Fine details can also be inconsistent.
I was hoping Canva Craft would help address this, especially by allowing me to generate images with editable layers. I tried it, but in my experience, the output still seemed to be a finished image rather than the kind of layered, editable design I was looking for. Perhaps this is something that will evolve as technology advances.
All this has made me reflect on something. The better I can communicate what I have in mind, the better my chances of getting the image I want. My rough sketch often helps because it communicates something that words alone struggle to convey.
This brings me to a more fundamental question. When an AI generates an image, what is actually happening behind the scenes?
When AI creates an image, what is actually happening?
Many popular AI image-generation models use a technique called diffusion.
Imagine the fuzzy grey static you might have seen on an old television when there was no signal. A diffusion model typically starts with an image containing random noise and gradually transforms it into an image that matches the prompt.
When you type a description, the model uses that prompt, along with patterns it has learned during training, to guide the process. It repeatedly estimates how to remove noise and refine the image until a recognisable result emerges.
Imagine asking an AI to generate a cat sitting on a bright windowsill next to a blooming marigold pot. The model does not necessarily draw the cat first, then add the windowsill and finally place the flowerpot beside it. Instead, the visual elements and their relationships emerge through the denoising process.
It is tempting to imagine the AI as a filter that progressively removes everything that does not fit the description. That is a useful way to visualise the broad idea, but it is not quite what happens. The model is not simply deleting irrelevant dots. It is using patterns learned during training to progressively generate an image guided by the prompt.
The important point is that the model has learned from existing images. It uses that learning to generate new outputs in response to our instructions.
But how does this compare with the way a human artist creates the very same scene?
How does creativity happen in the human brain?
I had once read about how creativity works in the human brain, and I thought I'd share a little snippet here. Of course, this is a very simplified explanation of a much more complex process, but it gives us an interesting starting point.
Two important brain networks often discussed in the neuroscience of creativity are the Default Mode Network (DMN) and the Executive Control Network (ECN).
The Default Mode Network (DMN) becomes active when our mind wanders, when we daydream or let our thoughts drift. Have you ever had a sudden idea while taking a shower, going for a walk or doing something routine? Your mind may be connecting seemingly unrelated memories, experiences and ideas. The DMN plays an important role in this process.
The Executive Control Network (ECN) helps us evaluate these ideas. Whether we are planning a family trip, finding a new way to solve a problem at work or figuring out how to reuse something at home, we need to decide which ideas might work and refine them.
Interestingly, research has found that these networks can work together during creative thinking. The brain needs to generate possibilities, but it also needs to evaluate them. An idea may be imaginative, but it may not be useful or appropriate for what we are trying to achieve.
Think about an artist creating a painting. The idea may come from a memory, an observation, an emotion or an unexpected association. As the artist works, they make decisions about colours, composition, proportions and details. They may change direction several times, discard parts of the work and try something entirely different.
Human creativity is not simply about producing something new. It involves the interplay of imagination, knowledge, experience, experimentation and judgement.
Of course, this is a simplified explanation of a much more complex process. Creativity does not reside in just two networks, and the brain does not follow a fixed sequence every time we create something. But these networks help us understand some of the processes involved.
Drawing a parallel: Creation vs. filtration
Now we can begin to see why an AI generating an image and a human artist creating one may be fundamentally different processes.
Imagine asking an artist to draw a simple scene: a cat sitting on a bright windowsill next to a blooming marigold pot.
The artist starts with a blank canvas and an idea in mind. They might draw upon memories of a pet curled up in the morning sun, the glaze of the glass vase or the way sunlight streams through a window. They make decisions about the composition, the colours, the proportions and the details. With every brushstroke, they gradually bring their imagined scene to life.
Now consider the AI image generator.
It begins with random noise and progressively transforms it into an image, guided by the prompt and patterns learned during training. The human artist builds a visual representation of an imagined scene through a series of decisions. The diffusion model generates an image through a mathematical process of denoising.
This is where the analogy of creation versus filtration comes in. The human artist begins with an idea and gives it visual form. The diffusion model begins with noise and transforms it into an image that increasingly matches the requested description.
But there is an important distinction to make here. We cannot conclude that the AI has no compositional ability simply because it works through denoising. It can produce coherent compositions and arrange visual elements in ways that make sense. The difference is in how it arrives at the result, rather than whether the result can be creative or visually compelling.
And the prompt matters enormously. It provides the subject, the direction and the constraints that guide the generation. A rough sketch, a reference image or a detailed description can help the model get closer to what we have in mind.
The human's intention therefore plays an important role in the process.
But isn't human creativity also built on what came before?
There is one important question we need to ask before drawing a sharp distinction between human and AI creativity. Don't humans also learn from the work of others?
An artist may study Van Gogh's use of colour, be inspired by the play of light in a photograph or borrow a technique from a painting they saw years ago. A musician may draw inspiration from another composer's work. A writer may combine ideas encountered in books, conversations and personal experiences to develop an entirely new perspective.
Very little of what we create emerges in complete isolation. Our brains are constantly absorbing information, forming associations and connecting things we have encountered before.
So, if both humans and AI draw upon what they have learned, is the difference between their creative processes really as clear-cut as we might imagine?
Perhaps the more interesting question is not whether we draw inspiration from existing work, but what happens to that inspiration once it becomes part of the creative process.
Human creativity is shaped by our experiences, emotions, intentions and understanding of the world. We can deliberately seek new influences, question what we have learned and change our perspective. AI models learn patterns from training data and use those patterns to generate new outputs. These are different processes, even though both can produce results that are novel.
This does not mean that AI can only copy existing work, or that human creativity is always original. Humans can imitate, and AI can generate surprising combinations. The distinction is more complicated than that.
So, what happens to creativity in the age of AI?
Generative AI is changing the way we create. It helps us explore possibilities, bring ideas to life and produce creative work much faster than before. For businesses that need large volumes of marketing material, or people who lack the technical skills to execute their ideas, this is enormously valuable. We could call this utility art, where the primary purpose of creating something visual is to serve a practical need, whether it is marketing a product, communicating an idea or engaging an audience.
But creativity is about more than producing an image, a painting or a piece of music. It is also about observing the world around us, making unexpected connections, experimenting and deciding what is worth creating. As AI makes execution easier, we may find ourselves spending more time directing, evaluating and refining what it produces. The challenge is to ensure that we continue to exercise our own imagination rather than simply choose from the options AI presents to us.
My take? In the age of AI, we should look at creativity not just as the ability to produce something, but as the ability to imagine possibilities, explore ideas and give them meaning. AI can help us bring those ideas to life, but we should never stop nurturing our own capacity to imagine. To my mind, that is creativity in its truest spirit.



Comments