AI: Democratizing Design
Midjourney prompt: Kid making artwork
A new kid on the block
Machine learning algorithms have been around for years, and designers and engineers have been putting design goals in generative programs to calculate all possible permutations for performance, spatial requirements, manufacturing and cost related challenges. Artificial computations are helping CAD design in virtually all manufacturing industries, and they are particularly gifted in generating structural optimizations.
Recently, there is a new emerging AI entering the field of design that offers imaging and rendering solutions rather than solving technical conundrums. This new artificial intelligence automatically generates and edits images following text instructions you feed it. Whatever illustration you can imagine, just describe it in words and the AI will spit out various interpretations in a matter of seconds.
AI has learned the relationship between images and text used to describe them and will create pieces of art based on what you type. Imagine it to be like a Google image search function that translates your thoughts into renderings and illustrations. All you need to do is describe your musings as detailed as possible and the machine learning system will generate multiple renderings with great accuracy, considerable expression and high image-fidelity.
The artificial learning software trains itself and creates solutions by accessing millions of images that form the raw material for an extensive iteration exercise and, learning from each iteration, the AI applies improvements in successive versions until it obtains an outcome that falls within the parameters and constraints of the original text input. Some algorithms, such as the much talked about DALL·E 2, can even create stylistically similar variations based on a pre-existing image rather than text.
Midjourney prompt: An engineer is working on a complex machine
AI programs are creators not creatives
AI can create art in a fraction of time so this must spell the end for visual arts and other creative disciplines. Not quite. Artificial intelligence mimics creativity based on previous work, it has no will, it has no goals or any storytelling ability. Image generation systems lack any real understanding of the world, or the consciousness and emotion to set forth unbridled creativity.
Understanding these subtleties and nuances are a true challenge for machine learning. The systems are highly generative and cannot convey morality, social awareness and ‘good’ design or art is created by copying and iteration rather than substance. Choosing the ‘best’ of anything means factoring in a plethora of subjective requirements that often conflict with each other and AI lacks the empathy to do so. Algorithms can make abundant variations but lacks emotional context.
Without this autonomous creative thinking image generating systems are merely tools that can help the ideation process, they have the potential to fuel creativity as a source of inspiration, but they will not replace human-like decisions; desire, feel, beauty, story and ownability are just a few of the aspects that are currently beyond the AI’s reach.
Midjourney prompt: A friendly helpful robot, hyper detailed, 3D painter, trending in artstation
A collaborator, not an enemy
Image-generating systems are less of a replacement and more a new partner that helps designers to explore numerous variations in an expedited manner. They will alter the conceptual phase with an immense increase in productivity and the amount of content produced. Artificial intelligence is going to be an aid in efficiency and inspiration rather than a competitor completing works.
The more input we feed it, the better the output we get. Designers that have early access to Image generating platforms found it to improve human creativity rather than replacing it. Algorithms generate images in different ways than us and they produce outcomes that we are not able to conjure up. That makes image algorithms excellent tools for inspiration and guidance.
Midjourney prompt: A friendly humanoid robot with liquid metal armor helping humans, hyper realistic, hyper detailed, unreal engine
AI, your data crunching partner
Being good at replicating repetitive or predictive tasks, AI can also analyze vast amounts of data to help with design prototyping. Feed your initial sketches, add a set of parameters and the algorithm will generate multiple variations in an instant, thus automating part of the iteration process with great efficiency.
Recognizing predictive patterns also allows AI to restore and de-pixelate images that suffer from poor lighting, bad contrast or color blemishes. Generative programs can transform low-quality images into sharp, printable assets. Furthermore, they can speed up mundane tasks such as cutting and separating main from secondary objects while placing them on different layers so that creatives can concentrate on design rather than preparation.
Midjourney prompt: A community of robots looking up at the sky, azure, cinematic lighting, in the style of tron
Give a person a tool and he or she will find innovative ways to use it. Image generating algorithms will allow virtually anyone to explore their creative ideas without the immediate need to involve designers and illustrators. But while AI can assist in visualizing your thoughts or add value to the creative process, it will not replace the design industry.
Image generating technologies present a series of exciting opportunities as sources of inspiration for creatives and as tools for everyone else to better translate their thoughts into visual concepts. It will be interesting to watch how the collaboration between humans and AI will evolve. How it will change the creative process; how it will make our lives easier.
As a closing thought, one aspect of artificial image computations worries me, algorithms are creating digital artwork by morphing imagery from the same, albeit very large, databases. Over time this repetitive process of mashing-up existing materials into new iterations may result in outcomes that look satiated and dated. And then there is the possible issue with plagiarism as all outcomes are iterations of existing work. But apart from this; we live in interesting times and artificial imaging intelligence will positively influence the creative process.
Midjourney prompt: Futuristic new world, cinematic lighting, hyper detailed, unreal engine, octane render